The European Union Versus the Internet

Earlier this summer the Internet breathed a sigh of relief: the European Parliament voted down a new Copyright Directive that would have required Internet sites to proactively filter uploaded content for copyright violations (the so-called “meme ban”), as well as obtain a license to include any text from linked sites (the “link tax”).

Alas, the victory was short-lived. From EUbusiness:

Internet tech giants including Google and Facebook could be made to monitor, filter and block internet uploads under amendments to the draft Copyright Directive approved by the EU Parliament Wednesday. At their plenary session, MEPs adopted amendments to the Commission’s draft EU Copyright Directive following from their previous rejection, adding safeguards to protect small firms and freedom of expression…

Parliament’s position toughens the Commission’s proposed plans to make online platforms and aggregators liable for copyright infringements. This would also apply to snippets, where only a small part of a news publisher’s text is displayed. In practice, this liability requires these parties to pay right holders for copyrighted material that they make available.

At the same time, in an attempt to encourage start-ups and innovation, the text now exempts small and micro platforms from the directive.

I chose this rather obscure source to quote from for a reason: should Stratechery ever have more than either 50 employees or €10 million in revenue, under this legislation I would likely need to compensate EUbusiness for that excerpt. Fortunately (well, unfortunately!), this won’t be the case anytime soon; I appreciate the European Parliament giving me a chance to start-up and innovate.

This exception, along with the removal of an explicit call for filtering (that will still be necessary in practice), was enough to get the Copyright Directive passed. This doesn’t mean it is law: the final form of the Directive needs to be negotiated by the EU Parliament, European Commission, and the Council of the Europe Union (which represents national governments), and then implemented via national laws in each EU country (that’s why it is a Directive).

Still, this is hardly the only piece of evidence that EU policy makers have yet to come to grips with the nature of the Internet: there is also the General Data Protection Regulation (GDPR), which came into effect early this year. Much like the Copyright Directive, the GDPR is targeted at Google and Facebook, but as is always the case when you fundamentally misunderstand what you are fighting, the net effect is to in fact strengthen their moats. After all, who is better equipped to navigate complex regulation than the biggest companies of all, and who needs less outside data than those that collect the most?

In fact, examining where it is that the EU’s new Copyright Directive goes wrong — not just in terms of policy, but also for the industries it seeks to protect — hints at a new way to regulate, one that works with the fundamental forces unleashed by the Internet, instead of against them.

Article 13 and Copyright

Forgive the (literal) legalese, but here is the relevant part of the Copyright Directive (the original directive is here and the amendements passed last week are here) pertaining to copyright liability for Internet platforms:

Online content sharing service providers perform an act of communication to the public and therefore are responsible for their content and should therefore conclude fair and appropriate licensing agreements with rightholders. Where licensing agreements are concluded, they should also cover, to the same extent and scope, the liability of users when they are acting in a non-commercial capacity…

Member States should provide that where right holders do not wish to conclude licensing agreements, online content sharing service providers and right holders should cooperate in good faith in order to ensure that unauthorised protected works or other subject matter, are not available on their services. Cooperation between online content service providers and right holders should not lead to preventing the availability of non-infringing works or other protected subject matter, including those covered by an exception or limitation to copyright…

This is legislative fantasizing at its finest: Internet platforms should get a license from all copyright holders, but if they don’t want to (or, more realistically, are unable to), then they should keep all copyrighted material off of their platforms, even as they allow all non-infringing work and exceptions. This last bit is a direct response to the “meme ban” framing: memes are OK, but the exception “should only be applied in certain special cases which do not conflict with normal exploitation of the work or other subject-matter concerned and do not unreasonably prejudice the legitimate interests of the rightholder.”1 That’s nearly impossible for a human to parse; expecting a scalable solution — which yes, inevitably means content filtering — is absurd. There simply is no way, especially at scale, to preemptively eliminate copyright violations without a huge number of mistakes.

The question, then, is in what direction those mistakes should run. Through what, in retrospect, are fortunate accidents of history,2 Internet companies are mostly shielded from liability, and need only respond to takedown notices in a reasonable amount of time. In other words, the system is biased towards false negatives: if mistakes are made, it is that content that should not be uploaded is. The Copyright Directive, though, would shift the bias towards false positive: it mistakes are made, it is that allowable content will be blocked for fear of liability.

This is a mistake. For one, the very concept of copyright is a government-granted monopoly on a particular arrangement of words. I certainly am not opposed to that in principle — I am obviously a benefactor — but in a free society the benefit of the doubt should run in the opposite direction of those with the legal right to deny freedom. The Copyright Directive, on the other hand, requires Internet Platforms to act as de facto enforcement mechanisms of that government monopoly, and the only logical response is to go too far.

Moreover, the cost of copyright infringement to copyright holders has in fact decreased dramatically. Here I am referring to cost in a literal sense: to “steal” a copyrighted work in the analog age required the production of a physical product with its associated marginal costs; anyone that paid that cost was spending real money that was not going to the copyright holder. Digital goods, on the other hand, cost nothing to copy; pirated songs or movies or yes, Stratechery Daily Updates, are very weak indicators at best of foregone revenue for the copyright holder. To put it another way, the harm is real but the extent of the harm is unknowable, somewhere in between the astronomical amounts claimed by copyright holders and the zero marginal cost of the work itself.

The larger challenge is that the entire copyright system was predicated on those physical mediums: physical goods are easier to track, easier to ban, and critically, easier to price. By extension, any regulation — or business model, for that matter — that starts with the same assumptions that guided copyright in the pre-Internet era is simply not going to make sense today. It makes far more sense to build new business models predicated on the Internet.

The music industry is a perfect example: the RIAA is still complaining about billions of dollars in losses due to piracy, but many don’t realize the industry has returned to growth, including a 16.5% revenue jump last year. The driver is streaming, which — just look at the name! — depends on the Internet: subscribers get access to basically all of the songs they could ever want, while the recording industry earns somewhere around $65 per individual subscriber per year with no marginal costs.3 It’s a fantastic value for customers and an equally fantastic revenue model for recording companies; that alignment stems from swimming with the Internet, not against it.

This, you’ll note, is not a statement that copyright is inherently bad, but rather an argument that copyright regulation and business models predicated on scarcity are unworkable and ultimately unprofitable; what makes far more sense for everyone from customers to creators is an approach that presumes abundance. Regulation should adopt a similar perspective: placing the burden on copyright holders not only to police their works, but also to innovate towards business models that actually align with the world as it is, not as it was.

Article 11 and Aggregators

This shift from scarcity to abundance has also had far-reaching effects on the value chains of publications, something I have described in Aggregation Theory (“Value has shifted away from companies that control the distribution of scarce resources to those that control demand for abundant ones“). Unfortunately the authors of the Copyright Directive are quite explicit in their lack of understanding of this dynamic; from Article 11 of the Directive:

The increasing imbalance between powerful platforms and press publishers, which can also be news agencies, has already led to a remarkable regression of the media landscape on a regional level. In the transition from print to digital, publishers and news agencies of press publications are facing problems in licensing the online use of their publications and recouping their investments. In the absence of recognition of publishers of press publications as rightholders, licensing and enforcement in the digital environment is often complex and inefficient.

In this reading the problem facing publishers is a bureaucratic one: capturing what is rightfully theirs is “complex and inefficient”, so the Directive provides for “the exclusive right to authorise or prohibit direct or indirect, temporary or permanent reproduction by any means and in any form, in whole or in part” of their publications “so that they may obtain fair and proportionate remuneration for the digital use of their press publications by information society service providers.”4

The problem, though, is that the issue facing publishers is not a problem of bureaucracy but of their relative position in a world characterized by abundance. I wrote in Economic Power in the Age of Abundance:

For your typical newspaper the competitive environment is diametrically opposed to what they are used to: instead of there being a scarce amount of published material, there is an overwhelming abundance. More importantly, this shift in the competitive environment has fundamentally changed just who has economic power.

In a world defined by scarcity, those who control the scarce resources have the power to set the price for access to those resources. In the case of newspapers, the scarce resource was reader’s attention, and the purchasers were advertisers…The Internet, though, is a world of abundance, and there is a new power that matters: the ability to make sense of that abundance, to index it, to find needles in the proverbial haystack. And that power is held by Google. Thus, while the audiences advertisers crave are now hopelessly fractured amongst an effectively infinite number of publishers, the readers they seek to reach by necessity start at the same place — Google — and thus, that is where the advertising money has gone.

This is the illustration I use to show the shift in publishing specifically (this time using Facebook):

A drawing of Aggregation Theory - Facebook and Newspapers

This is why the so-called “link tax” is doomed to failure — indeed, it has already failed every time it has been attempted. Google, which makes no direct revenue from Google News,5 will simply stop serving Google News to the EU, or dramatically curtail what it displays, and the only entities that will be harmed — other than EU consumers — are the publications that get traffic from Google News. Again, that is exactly what happened previously.

There is another way to understand the extent to which this proposal is a naked attempt to work against natural market forces: Google’s search engine respects a site’s robot.txt file, wherein a publisher can exclude their site from the company’s index. Were it truly the case that Google was profiting unfairly from the hard word of publishers, then publishers have a readily-accessible tool to make them stop. And yet they don’t, because the reality is that while publishers need Google (and Facebook), that need is not reciprocated. To that end, the only way to characterize money that might flow from Google and Facebook (or a €10-million-in-revenue-generating Stratechery) to publishers is as a redistribution tax, enforced by those that hold the guns.

Here again the solution ought to flow in the opposite direction, in a way that leverages the Internet, instead of fighting it. An increasing number of publishers, from large newspapers to sites like Stratechery, are taking advantage of the massive addressable market unlocked by the Internet, leveraging the marketing possibilities of free social media and search engine results, and connecting directly with readers that care — and charging them for it.

I do recognize this is a process that takes time: it is particularly difficult for publishers built with monopoly-assumptions to change not just their business model but their entire editorial strategy for a world where quality matters more than quantity. To that end, if the EU wants to, as they say in the Copyright Directive, “guarantee the availability of reliable information”, then make the tax and subsidy plan they effectively propose explicit. At least then it would be clear to everyone what is going on.

The GDPR and the Regulatory Corollary of Aggregation

This brings me to a piece of legislation I have been very critical of for quite some time: GDPR. The intent of the legislation is certainly admirable — protect consumer privacy —although (and this may be the American in me speaking) I am perhaps a bit skeptical about just how much most consumers care relative to elites in the media. Regardless, the intent matters less than the effect, the latter of which is to entrench Google and Facebook. I wrote in Open, Closed, and Privacy:

While GDPR advocates have pointed to the lobbying Google and Facebook have done against the law as evidence that it will be effective, that is to completely miss the point: of course neither company wants to incur the costs entailed in such significant regulation, which will absolutely restrict the amount of information they can collect. What is missed is that the increase in digital advertising is a secular trend driven first-and-foremost by eyeballs: more-and-more time is spent on phones, and the ad dollars will inevitably follow. The calculation that matters, then, is not how much Google or Facebook are hurt in isolation, but how much they are hurt relatively to their competitors, and the obvious answer is “a lot less”, which, in the context of that secular increase, means growth.

This is the conundrum that faces all major Internet regulation, including the Copyright Directive; after all, Google and Facebook can afford — or have already built — content filtering systems, and they already have users’ attention such that they can afford to cut off content suppliers. To that end, the question is less about what regulation is necessary and more about what regulation is even possible (presuming, of course, that entrenching Google and Facebook is not the goal).

This is where thinking about the problems with the Copyright Directive is useful:

  • First, just as business models ought to be constructed that leverage the Internet instead of fight it, so should regulation.
  • Second, regulation should start with the understanding that power on the Internet flows from controlling demand, not supply.

To understand what this sort of regulation might look like, it may be helpful to work backwards. Specifically, over the last six months Facebook has made massive strides when it comes to protecting user privacy. The company has shut down third-party access to sensitive data, conducted multiple audits of app developers that accessed that data, added new privacy controls, and more. Moreover, the company has done this for all of its users, not just those in the EU, suggesting its actions were not driven by GDPR.

Indeed, the cause is obvious: the Cambridge Analytica scandal, and all of the negative attention associated with it. To put it another way, bad PR drove more Facebook action in terms of user privacy than GDPR or a FTC consent decree. This shouldn’t be a surprise; I wrote in Facebook’s Motivations:

Perhaps there is a third motivation though: call it “enlightened self-interest.” Keep in mind from whence Facebook’s power flows: controlling demand. Facebook is a super-aggregator, which means it leverages its direct relationship with users, zero marginal costs to serve those users, and network effects, to steadily decrease acquisition costs and scale infinitely in a virtuous cycle that gives the company power over both supply (publishers) and advertisers.

It follows that Facebook’s ultimate threat can never come from publishers or advertisers, but rather demand — that is, users. The real danger, though, is not from users also using competing social networks (although Facebook has always been paranoid about exactly that); that is not enough to break the virtuous cycle. Rather, the only thing that could undo Facebook’s power is users actively rejecting the app. And, I suspect, the only way users would do that en masse would be if it became accepted fact that Facebook is actively bad for you — the online equivalent of smoking.

For Facebook, the Cambridge Analytica scandal was akin to the Surgeon General’s report on smoking: the threat was not that regulators would act, but that users would, and nothing could be more fatal. That is because:

The regulatory corollary of Aggregation Theory is that the ultimate form of regulation is user generated.

If regulators, EU or otherwise, truly want to constrain Facebook and Google — or, for that matter, all of the other ad networks and companies that in reality are far more of a threat to user privacy — then the ultimate force is user demand, and the lever is demanding transparency on exactly what these companies are doing.

To that end, were I a regulator concerned about user privacy, my starting point would not be an enforcement mechanism but a transparency mechanism. I would establish clear metrics to measure user privacy — types of data retained, types of data inferred, mechanisms to delete user-generated data, mechanisms to delete inferred data, what data is shared, and with whom — and then measure the companies under my purview — with subpoena power if necessary — and publish the results for the users to see.

This is the way to truly bring the market to bear on these giants: not regulatory fiat, but user sentiment. That is because it is an approach that understands the world as it is, not as it was, and which appreciates that bad PR — because it affects demand — is a far more effective instigator of change than a fine paid from monopoly profits.

I wrote a follow-up to this article in this Daily Update.

  1. Full text of the “meme exception”:

    Despite some overlap with existing exceptions or limitations, such as the ones for quotation and parody, not all content that is uploaded or made available by a user that reasonably includes extracts of protected works or other subject-matter is covered by Article 5 of Directive 2001/29/EC. A situation of this type creates legal uncertainty for both users and rightholders. It is therefore necessary to provide a new specific exception to permit the legitimate uses of extracts of pre-existing protected works or other subject-matter in content that is uploaded or made available by users. Where content generated or made available by a user involves the short and proportionate use of a quotation or of an extract of a protected work or other subject-matter for a legitimate purpose, such use should be protected by the exception provided for in this Directive. This exception should only be applied in certain special cases which do not conflict with normal exploitation of the work or other subject-matter concerned and do not unreasonably prejudice the legitimate interests of the rightholder. For the purpose of assessing such prejudice, it is essential that the degree of originality of the content concerned, the length/extent of the quotation or extract used, the professional nature of the content concerned or the degree of economic harm be examined, where relevant, while not precluding the legitimate enjoyment of the exception. This exception should be without prejudice to the moral rights of the authors of the work or other subject-matter. [↩︎]

  2. That link is about Section 230, which is a U.S. law shielding Internet platforms from liability for what their users upload, but the same principle broadly speaking applies in the E.U. presently [↩︎]
  3. That $65 figure is an estimate of the amount paid out by streaming services like Spotify; the total number per listener is lower, thanks to family plans and shared accounts [↩︎]
  4. The first quotation is from EU Directive 2001/29/EC which is explicitly evoked in the new Copyright Directive, from whence comes the second quotation. [↩︎]
  5. The company does, of course, collect data to be used in advertising elsewhere [↩︎]

The iPhone Franchise

Apple released a new flagship iPhone yesterday, the iPhone XS. This isn’t exactly ground-breaking news: it is exactly what the company has done for eleven years now (matching the 11-year run of non-iOS iPods, by the way1). To that end, what has always interested me more are new-to-the-world non-flagship models: the iPhone 5C in 2013, the iPhone 8 last year (or was it the iPhone X?), and the iPhone XR yesterday. Each, I think, highlights critical junctions not only in how Apple thinks about the iPhone strategically, but also about how Apple thinks about itself.

The iPhone 5C

It’s hard to remember now, but the dominant Apple narrative in 2013, after a five-year iPhone run that saw the company’s stock price increase around 700%, was that the company was at risk of low-end disruption from Android and high-end saturation now that smartphone technology was “good enough”.

Apple's stock price during the iPhone era

This was, for me, rather fortuitous: Stratechery launched in the middle of the Apple-needs-a-cheap-iPhone era, providing plenty of fodder not only for articles defending Apple’s competitive position,2 but also multiple articles speculating on what the iPhone 5C would cost and how it would be positioned.

For the record, I guessed wrong, and I knew I was wrong Two Minutes, Fifty-six Seconds into the keynote.

It was at two minutes, fifty-six seconds that Tim Cook said there would be a video – a video! – about the iTunes Festival.

And it was awesome.

In case you didn’t watch the whole video (and you really should – it’s only a couple of minutes; due to a copyright claim I had to embed Apple’s full-length keynote), this clip of the ending captures why it matters:

Message: Apple is cool.
Message: Apple is cool.

This was Apple, standing up and saying to all the pundits, to all the analysts, to everyone demanding a low price iPhone:

NO!

No, we will NOT compete on price, we will offer something our competitors can’t match.

No, we are NOT selling a phone, we are selling an experience.

No, we will NOT be cheap, but we will be cool.

No, you in the tech press and on Wall Street do NOT understand Apple, but we believe that normal people love us, love our products, and will continue to buy, start to buy, or aspire to buy.

Oh, and Samsung? Damn straight people line up for us. 20 million for a concert. “It’s like a product launch.”

Apple's iTunes Festival video on the left, Samsung's Galaxy SIII commercial mocking those standing in line on the right
Apple’s iTunes Festival video on the left, Samsung’s Galaxy SIII commercial mocking those standing in line on the right

This attitude and emphasis on higher-order differentiation — the experience of using an iPhone — dominated the entire keynote and the presentation of features, with particularly emphasis throughout on the interplay between software and hardware.

In fact, that understated Apple’s position in the market: as I discussed last year the iPhone 5C — which in retrospect, was really just an iPhone 5 replacement in Apple’s trickle-down approach to serving more price-sensitive customers — was a bit of a failure: Apple customers only wanted the best iPhone, and those that couldn’t afford the current flagship preferred a former flagship, not one that was “unapologetically plastic”.

Thus the first lesson: Apple wouldn’t go down-market, nor did its customers want it to.

The iPhone X

Last year, meanwhile, was in many respects the opposite of the iPhone 5S and 5C launch, at least from a framing perspective. The iPhone 8 was the next in line after the iPhone 7 and all of the iPhones before it; it was the iPhone X that was presented as being out-of-band — “one more thing”, to use the company’s famous phrase. The iPhone X was the “future of the smartphone”, with a $999 price tag to match.

A year on, it is quite clear that the future is very much here. CEO Tim Cook bragged during yesterday’s keynote that the iPhone X was the best-selling phone in the world, something that was readily apparent in Apple’s financial results. iPhone revenue was again up-and-to-the-right, not because Apple was selling more iPhones — unit growth was flat — but because the iPhone X grew ASP so dramatically:

iPhone Revenue, Units, and ASP on a TTM basis

This was the second lesson: for Apple’s best customers, price was no object.

The iPhone XR

To be clear, the overall strategy and pricing of the iPhones XS and XR were planned out two to three years ago; that’s how long product cycles take when it comes to high-end smartphones. Perhaps that is why the lessons of the iPhone 5C seem so readily apparent in the iPhone XR in particular.

First off, while the XR does not have stainless steel edges like the iPhones X or XS, it is a far cry from plastic: the back is glass, like the high end phones, and the aluminum sides not only look premium but will be hidden when the phone is in a case, as most will be. What really matters is that the front looks the same, with that notch: this looks like a high-end iPhone, with all of the status that implies.

Second, the iPhone XR is big — bigger than the XS (and smaller than the XS Max, and yes, that is its real name). This matters less for 2018 and more for 2020 and beyond: presuming Apple follows its trickle-down strategy for serving more price-sensitive markets, that means in two years its lowest-end offering will not be a small phone that the vast majority of the market rejected years ago, particularly customers for whom their phone is their only computing device, but one that is far more attractive and useful for far more people.

Third, that 2020 iPhone XR is going to be remarkably well-specced. Indeed, probably the biggest surprise from these announcements (well, other than the name “XS Max”) is just how good of a smartphone the XR is.

  • The XR has Apple’s industry-leading A12 chip, which is so far ahead of the industry that it will still be competitive with the best Android smartphones in two years, and massively more powerful than lower-end phones.
  • The XR has the same wide-angle camera as the XS, and the same iteration of Face ID. Both, again, are industry-leading and will be more than competitive two years from now.
  • The biggest differences from the XS are the aforementioned case materials, an LCD screen, and the lack of 3D Touch. Again, though, aluminum is still a premium material, Apple’s LCD screens are — and yes there is a theme here — the best in the industry, and 3D Touch is a feature that is so fiddly and undiscoverable that one could make the case XR owners are actually better off.

There really is no other way to put it: the XR is a fantastic phone, one that would be more than sufficient to maintain Apple’s position atop the industry were it the flagship. And yet, in the context of Apple’s strategy, it is best thought of as being quite literally ahead of its time.

The iPhone XS

There is, of course, the question of cannibalism: if the XR is so great, why spend $250 more on an XS, or $350 more for the giant XS Max?

This is where the iPhone X lesson matters. Last year’s iPhone 8 was a great phone too, with the same A11 processor as the iPhone X, a high quality LCD screen like the iPhone XR, and a premium aluminum-and-glass case (and 3D Touch!). It also had Touch ID and a more familiar interface, both arguably advantages in their own right, and the Plus size that so many people preferred.

It didn’t matter: Apple’s best customers, not just those who buy an iPhone every year, but also those whose only two alternatives are “my current once-flagship iPhone” or “the new flagship iPhone” are motivated first-and-foremost by having the best; price is a secondary concern. That is why the iPhone X was the best-selling smartphone, and the iPhone 8 — which launched two months before the iPhone X — a footnote.

To be sure, the iPhone X had the advantage of being something truly new, not just the hardware but also the accompanying software. It was the sort of phone an Apple fan might buy a year sooner than they had planned, or that someone more price sensitive might choose over a cheaper option. The XS will face headwinds in both regards: it is faster than the iPhone X, has a better camera, comes in gold — it’s an S-model, in other words — but it’s hard to see it pulling forward upgrades; it’s more likely natural XS buyers were pulled forward by the X. And, as noted above, the XR is a much more attractive alternative to the X than the 8 was to the X; most Apple fans may want the best, but some just want a deal, and the XR is a great one.

Apple should be fine though: overall unit sales may fall slightly, but the $1,099 XS Max will push the average selling price even higher. Note, too, that the XR is only available starting at $749; the longstanding $650 iPhone price point was bumped up to $699 last year, and is now a distant memory.3

Apple's Fall 2018 iPhone Lineup

To put it another way, to the extent the XR cannibalizes the XS, it cannibalizes them with an average selling price equal to Apple’s top-of-the-line iPhone from two years ago; the iPhone 8 is $50 higher than the former $550 price point as well.

Mission Impossible iPhone

This is what I meant when I said Apple’s second iPhone models capture how the company has changed not only its strategy but how the company seems to view itself:

  • 2013 was a time of uncertainty, with a sliding stock price and a steadily building clamor heralding Apple doom via low-end disruption; the company, though, found its voice with the 5C and declared its intentions to be unapologetically high-end; the 5C’s failure, such that it was, only cemented the rightness of that decision.
  • In 2017 the company, for the first time in ten years, started to truly test the price elasticity of demand for the iPhone: given its commitment to being the best, just how much could Apple charge for an iPhone X?
  • This year, then, comes the fully-formed iPhone juggernaut: an even more expensive phone, with arguably one of the weaker feature-driven reasons-to-buy to date, but for the fact it is Apple’s newest, and best, iPhone. And below that, a cheaper iPhone XR that is nearly as good, but neatly segmented primarily by virtue of not being the best, yet close enough to be a force in the market for years to come.

The strategy is, dare I say, bordering on over-confidence. Apple is raising prices on its best product even as that product’s relative differentiation from the company’s next best model is the smallest it has ever been.

Here, though, I thought the keynote’s “Mission: Impossible”-themed opening really hit the mark: the reason why franchises rule Hollywood is their dependability. Sure, they cost a fortune to make and to market, but they are known quantities that sell all over the world — $735 million-to-date for the latest Tom Cruise thriller, to take a pertinent example.

That is the iPhone: it is a franchise, the closest thing to a hardware annuity stream tech has ever seen. Some people buy an iPhone every year; some are on a two-year cycle; others wait for screens to crack, batteries to die, or apps to slow. Nearly all, though, buy another iPhone, making the purpose of yesterday’s keynote less an exercise in selling a device and more a matter of informing self-selected segments which device they will ultimately buy, and for what price.

I wrote a follow-up to this article in this Daily Update.

  1. Specifically, the original iPod was released in October, 2001, and the 7th generation iPod Nano in September, 2012; the last iPod Touch was released in July, 2015 [↩︎]
  2. I.e. Two Bears and What Clayton Christensen Got Wrong [↩︎]
  3. Hilariously, Senior Vice President of Worldwide Marketing Phil Schiller said in reference to the $749 price point, “That’s less than the iPhone 8 Plus. I’m really proud of the work the team has done on that”; Apple shareholders are surely proud that the price is $50 higher than the iPhone 8! [↩︎]

Uber’s Bundles

With Uber, nothing is easy.

Start with profitability, or the lack thereof: two weeks ago the company reported its quarterly “earnings”,1 and once again the losses were massive: $891 million on $2.8 billion in revenue. Clearly the business is failing, no?

Well, like I said, it’s not that easy: unlike a company like MoviePass, Uber has positive unit economics — that is, the company makes money on each ride. This is clear intuitively: Uber keeps somewhere between 20%–30% of each fare,2 from which it pays insurance costs, credit card fees, etc., and keeps the rest.3 According to last quarter’s numbers “the rest” totaled $1.5 billion, for a gross profit margin of 55% (13% of Uber’s total bookings). Moreover, margin is improving — it was 47% a year ago — mostly because Uber is managing to both take a higher percentage of fares even as it has reduced its spending on promotions and driver incentives (Cost of Revenue, meanwhile, appears to correspond very closely to gross bookings).

The problem for Uber is trifold: first, the company continues to spend massive amounts of money on “below-the-line” costs: $2.2 billion for Operations and Support, Sales and Marketing,4 Research and Development, General and Administrative, and Depreciation and Amortization. Second, it seems likely that a good portion of the company’s improving margin stems from exiting more difficult markets like Russia and Southeast Asia, as opposed to improvements in its core markets in the United States, Europe, and Oceania. And most concerning of all, Lyft seems to be outgrowing Uber.

Uber’s Lyft Problem

Lyft is a problem for Uber with riders, investors, and drivers.

From a rider perspective, Lyft has, unsurprisingly, benefited from the self-inflicted disaster that was 2017 (although to be fair, 2017 was the year of revelations of problems that had been in place for years). The consumer benefit of services like Uber and Lyft has always been clear, and Uber’s aggressive expansion paid off when the service became the default choice for a large portion of the market, something that is critical for a commodity offering with two-sided network effects. The problem is that Uber gave riders plenty of reasons to question their default choice with not just a sexual harassment scandal, and not just a lawsuit alleging the theft of intellectual property from Google, and not just allegations of brazenly circumventing local regulators, but all three (and honestly, this understates things).

This was particularly problematic because that two-sided network effect wasn’t that strong: sure, Uber was more likely to monopolize driver time given its larger user base, but as long as drivers are independent contractors Uber can’t do anything to prevent them from multihoming, that is, being available on both Uber and Lyft’s networks at the same time. Lyft was ready-and-able to absorb unhappy Uber riders, because they were effectively using Uber’s drivers to accommodate them.

The timing could not have been worse: it was only a few months prior that Lyft appeared to be for sale and unable to find a buyer; it seemed that former-CEO Travis Kalanick was going to win one of his biggest gambles, turning down an offer to acquire Lyft in 2014 in exchange for 18% of Uber.

It proved to be Kalanick’s biggest mistake, at least from a business perspective: within weeks of the Uber scandal explosion Lyft raised $600 million, and a month later formed a partnership with Waymo, Google’s self-driving car company. Suddenly the best way to invest in the most promising self-driving technology was Lyft; unsurprisingly Lyft has since raised an additional $2.3 billion, including an investment from Google Capital.

Uber’s Competitive Context

The reason this context matters is that a proper analysis of Uber’s business is fundamentally different today than it was two years ago — or four years ago, when I wrote Why Uber Fights. That is when I made the argument that even though Uber’s two-sided network effects were relatively weak thanks to the lack of driver lock-in, the fact that ride-sharing was a commodity market meant its head start and brand would lead to slow-but-steady growth in marketshare, eventually starving Lyft due to an inability to raise funds based on increasingly inferior financial results.

I stand by that analysis: it is exactly what happened, and Uber came very close to knocking Lyft out. At the same time, it is also no longer applicable, because Lyft no longer has any problem raising money, while Uber appears to be having a hard time holding onto its market share (as an indirect indicator of Uber’s waning power with consumers, note Uber’s recent inability to defeat a cap on ride-sharing in New York, after doing just that three years ago). To that end, the prospect of Lyft being present in the market for the foreseeable future means Uber’s needs a new strategy than simply squeezing Lyft dry.

Welcome to bundles, Uber-style.

Uber’s Consumer Bundle: Transportation-as-a-Service

Uber CEO Dara Khosrowshahi laid out a new vision for the Uber app in an interview with Kara Swisher earlier this year at the Code Conference:

DK: We are thinking about alternative forms of transport. If you look at Jump, the dockless bicycle startup Uber acquired earlier this year, the average length of a trip at Jump is 2.6 miles. That is, 30 to 40 percent of our trips in San Francisco are 2.6 miles or less. Jump is much, much cheaper than taking an UberX. To some extent it’s like, “Hey, let’s cannibalize ourselves.” Let’s create a cheaper form of transportation from A to B, and for you to come to Uber, and Uber not just being about cars, and Uber not being about what the best solution for us is, but really being about the best solution for here.

KS: So bikes, scooters?

DK: Bikes, perhaps scooters. I wanna get the bus network on. I wanna get the BART, or the Metro, etc., onto Uber. So, any way for you to get from point A to B.

KS: Wait, you wanna start your own BART? No.

DK: No, no, no. We’re not gonna go vertical. Just like Amazon sells third-party goods, we are going to also offer third-party transportation services. So, we wanna kinda be the Amazon for transportation, and we want to offer the BART as an alternative. There’s a company called Masabi that is connecting Metro, etc., into a payment system. So we want you to be able to say, “Should I take the BART? Should I take a bike? Should I take an Uber?” All of it to be real-time information, all of it to be optimized for you, and all of it to be done with the push of a button.

KS: So, any transportation?

DK: Any transportation, totally frictionless, real time.

In case you had any question about how serious Khosrowshahi is about the concept, he told the Financial Times in an interview yesterday:

During rush hour, it is very inefficient for a one-tonne hulk of metal to take one person 10 blocks…We’re able to shape behaviour in a way that’s a win for the user. It’s a win for the city. Short-term financially, maybe it’s not a win for us, but strategically long term we think that is exactly where we want to head…

We are willing to trade off short-term per-unit economics for long-term higher engagement…I’ve found in my career that engagement over the long term wins wars and sometimes it’s worth it to lose battles in order to win wars.

This is very much a bundle, and like any bundle, what makes the economics work in the long run is earning a larger total spend from consumers even if they spend less on any particular item. To that end, as Khosrowshahi notes, the real enemy is the car in the garage; to the extent Uber can replace that the greater its opportunity is.

Uber's consumer bundle

Moreover, the more that Uber can handle all of an end user’s transportation needs through the sort of complexity inherent in building such a service, the stickier Uber becomes for consumers. Granted, Lyft is promising to build the same thing, but Uber is a bit ahead and still has the bigger war chest, which may prove more helpful in a land grab as opposed to the current war of attrition. Moreover, Uber still has a significant geographic advantage over Lyft, which only just started expanding internationally, making it a better option for travelers.

Uber’s Driver Bundle: Uber Eats

Uber Eats, meanwhile, has the potential to be a very attractive business in its own right: Khosrowshahi said at Code Conference that the business has “a $6 billion bookings run rate, growing over 200 percent.” Uber takes 30% of that ($1.6 billion), as well as a $5 delivery fee from customers, out of which it pays drivers a pickup fee, drop-off fee, and per-mile rate (of which it keeps 25%); according to The Information, the service isn’t making money yet, but it is much more profitable than Uber’s ride-sharing business was at a similar scale.

Leaving aside the drivers for a moment, this is a classic aggregation play, where owning consumer demand gives Uber the ability to attract suppliers, increasing consumer demand in a virtuous cycle. Jason Droege, the Uber Vice President and Head of UberEverything, told Eater in an interview this past summer:

I think that we’re all here to service the consumer, right? And the eater. And I think eaters today want convenience, they want value, they want flexibility, and they want choice. And delivery offers all of those things. And restaurants choose to participate in delivery. And so if they don’t believe it’s valuable as a channel to connect to their consumers, or maybe new consumers, or reach new people with their brand, then that’s okay. We’re here to provide a conduit between the two. Not to tell them how to run their business.

It certainly is an open question as to whether services like Uber Eats help or hurt established restaurants; this New Yorker article recounts a number of anecdotes about restauranteurs who are a bit fuzzy on exactly how much Uber Eats is costing them, much like Uber drivers that forget to account for the wear-and-tear on their cars. At the same time, Uber is creating entirely new opportunities for restaurants focused on delivery, just as it did for drivers that only wanted to work sometimes, or couldn’t find any other job at all, as well as companies to service them like HyreCar.

Moreover, Uber Eats has a leg-up in the space because of Uber itself: the latter can acquire customers from the former (both because of owned-and-operated advertising as well as reducing drop-off because Uber already has payment details), and all of those huge marketing and G&A expenses from building out teams in every city Uber operates is easily leveraged for Uber Eats. This of course applies to driver acquisition costs as well.

The biggest payoff, though, comes from effectively bundling opportunities for drivers. The problem for any standalone restaurant delivery app is that the vast majority of orders come at lunch and dinner, but the driver may wish to work at other times of the day as well. With Uber that is easy: just pick up riders (Uber drivers can drive for just Uber, just Uber Eats, or both). In other words, Uber has more and more ways to monopolize a driver’s time, to the driver’s benefit personally and Uber’s benefit competitively.

Uber's driver bundle

To be sure a GrubHub driver, to take one Uber Eats competitor at random, could also drive for Lyft (or Uber, for that matter), but that is where rewarding drivers for a certain number of rides in a given time period is particularly effective: because drivers can complete their “Quests” with Uber ride-sharing trips or Uber Eats trips, it often makes more sense to simply stick with Uber.

More broadly, the challenge Uber faces with drivers derives from the same fungibility that makes the service possible in the first place. To that end, the best way to approach the driver market is not to compete against this reality but to embrace it, and having multiple services that utilize the same driver pool accomplishes exactly that.

Self-Driving Cars: A Bundle as the Way Forward

Self-driving cars, meanwhile, remain Uber’s white whale. The company received a $500 million investment from Toyota yesterday, and will work to incorporate its technology into Toyota Sienna minivans.

This is definitely not the divesture of the unit that The Information says has been mooted; the unit has apparently cost Uber $2 billion over the last two years. Of course, as I noted above, that cost pales in comparison to the strategic impact of losing Google as a potential partner to Lyft.

Still, it is never too late to consider doing the right thing: I continue to believe that Uber’s investment in self-driving cars was a strategic mistake. Yes, its biggest cost is drivers, and a theoretical Google ride-sharing service could, were it at scale, completely undercut Uber, but that is the shallowest possible way to analyze how this market might have played out.

Keep in mind the point I just made about drivers: sure it sounds attractive to convert your most expensive supply input, which must be paid on a marginal basis and that you don’t control, into a fixed cost that you have exclusive rights to. That, though, means massively more capital expenditures for a business that is currently losing around a billion dollars a quarter. Worse, it means competing with Google in an area — machine learning — where the search giant has a massive advantage.

Moreover, in the long run it seems unlikely that Google would want to build up a vertical Uber competitor: it remains far more logical, both financially and in terms of Google’s historical margin profile, to license out their technology. To be sure, if Waymo’s technology were superior, they would have wholesale transfer pricing power, which Tren Griffin describes as:

The bargaining power of company A that supplies a unique product XYZ to Company B which may enable company A to take the profits of company B by increasing the wholesale price of XYZ

Here’s the thing though: Uber is better equipped than anyone else to deal with Waymo’s potential ability to extract margin for superior self-driving technology. After all, the company is already paying for driving technology — the technology just happens to be a human!

Of course that doesn’t mean Uber should settle for paying Waymo instead of drivers: the ride-sharing service remains the best possible way to go-to-market for everyone working on self-driving technology. To that end Uber should be willing to partner with anyone and everyone — and to share its technology with whoever wants it. In the long run Uber has market power thanks to its network, and it will best exploit that power to the extent it can engender competition amongst suppliers in the self-driving car space.

Uber's self-driving bundle

Moreover, it seems certain that whenever self-driving cars come along (and even Waymo is having trouble), they will not be suitable for all of the environments Uber operates in. That makes Uber uniquely suited to bundle self-driving car service with traditional Uber car service, as well as all of the other transportation services it plans to offer to consumers. This “bundle” will allow self-driving technology to come-to-market gradually when and where it makes sense, while still giving riders the confidence they can get from anywhere to anywhere.

To be fair, Khosrowshahi has signaled the desire to partner with multiple self-driving partners, including Google. I suspect, though, that will be hard to accomplish as long as Uber is pursuing its own exclusive technology. To that end Khosrowshahi should cut the cord with Uber’s self-driving program sooner rather than later, or perhaps even open-source it; the money savings are in fact the second most important potential benefit.


There was a certain satisfying simplicity to the brutality of Uber’s original strategy under Kalanick: be as aggressive as possible to establish an early lead, and then leverage Uber’s seemingly limitless ability to raise money to spend its competitors into submission. In the end, though, that same brutality did Kalanick in, and his strategy along with it.

That left Khosrowshahi with a much more complicated situation: not only did he need to fix Uber internally, he needed to create an entirely new strategy to win in a market that was fundamentally altered because of Uber’s crisis. The bundling of services for users, of opportunities for drivers, and ideally of technologies for self-driving makes sense as an alternative.

This strategy is, though, befitting the nature of the situation, considerably more complicated, and the commensurate chances of success — and ultimately, of profitability — a fair bit lower. In other words, Uber’s boardroom drama may be over, but the company remains perhaps the most compelling in tech.

I wrote a follow-up to this article in this Daily Update.

  1. Uber voluntarily shares high level numbers with media outlets (the Wall Street Journal has collected them here), but the numbers are selective, unaudited, and come with no financial documents [↩︎]
  2. The company now deducts the amount spent on driver incentives and promotions, in addition to driver earnings on a percentage basis, from its overall bookings; this is a very welcome improvement to the company’s reporting. Previously it was unclear whether or not this spending was accounted for correctly [↩︎]
  3. You can see an old breakdown from a 2015 leaked document here [↩︎]
  4. As noted in an earlier footnote, Uber does appear to be deducting promotional expenses that apply to specific rides from booking, so these are non-unit marketing costs [↩︎]

Facebook’s Story Problem — and Opportunity

This is hardly the first Stratechery article about Facebook to start with Snapchat. The “camera company” née social network reported earnings that were, as they say, mixed. Revenue beat expectations by 5%, but user growth, after slowing considerably for two years, went in the opposite direction: Snap’s Daily Active Users were still up 9% year-over-year, but for the first time declined sequentially (by 2%):

Snapchat's Daily Active Users

So what happened?

Snap’s Growth Excuses

Snap’s stated rationale for its slowing growth has shifted over those two years:

  • In its S-1, filed in February 2017, the company blamed “technical issues”:

    In mid-2016, we launched several products and released multiple updates, which resulted in a number of technical issues that diminished the performance of our application. We believe these performance issues resulted in a reduction in growth of Daily Active Users in the latter part of the quarter ended September 30, 2016.

  • During the pre-IPO roadshow, CEO Evan Spiegel blamed Snapchat’s performance on Android:

    I think broadly speaking if you look at rest of world growth as a proxy for Android, you can start getting an understanding for the performance issues we face on Android in the last two quarters.

  • On the company’s first earnings call, Spiegel pointed to the company’s restraint:

    I’d love to speak a little bit to the DAU question, because it’s a question that we get all the time. And I think one of the reasons why it’s such a popular question is because there’s a lot of this thing in our industry called growth hacking, where you send a lot of push notifications to users or you try to get them to do things that might be unnatural or something like that. And I think while that’s the easy way to grow daily actives quickly, we don’t think that those sorts of techniques are very sustainable over the long term. And I think that can ultimately impact our relationship with the customer.

  • A year ago, on the company’s third conference call, Spiegel blamed a shift in measurement:

    This quarter, we grew our Daily active users at a lower rate than we would have liked, adding 4.5 million new users. This can be partially attributed to our decision to report our daily active users as an average over the entire quarter, where a strong September was offset by the more modest months of July and August. Ultimately though, we want to drive more user growth in 2018.

  • In March the blame was put on Snapchat’s re-design, along with that old bugaboo Android:

    As we have mentioned on our past two earnings calls, a change this big to existing behavior comes with some disruption, especially given the high frequency of daily engagement of our community. While we had an average of 191 million daily active users in Q1, our March average was lower, but still above our Q4 average. We are already starting to see early signs of stabilization among our iOS users as people get used to the changes, but still have a lot of work to do to optimize the new design, especially for our Android users.

  • Last quarter the re-design again took center stage:

    While our monthly active users continue to grow this quarter, we saw 2% decline in our daily active users. This was primarily driven by a slightly lower frequency of use among our user base due to the disruption caused by our redesign. It has been approximately six months since we broadly rolled out the redesign of our application and we have been working hard to iterate and improve Snapchat based on the feedback from our community.

Note that this last explanation is a bit different than the others: Snap was admitting that its core users were using the product less. The problem is that while social networks making blunders in the user experience is hardly a new phenomenon, the difference between Snapchat and, well, Facebook, is the lack of growth — going on for two years now — to make up for it.

And, of course, there is the reason for slowing growth that Snap’s executives can’t bring themselves to acknowledge: Instagram Stories.

Copying Audacity Redux

In August 2016, when Instagram launched what its own executives admitted was a rip-off of Snapchat’s Stories feature, I wrote in The Audacity of Copying Well that the effect would not be to steal Snapchat’s users, at least not yet; rather, Instagram Stories looked poised to kill Snapchat’s growth:

Facebook is leveraging one of their most valuable assets: Instagram’s 500 million users. The results, at least anecdotally, speak for themselves: I’ve seen more Instagram stories in the last 24 hours than I have Snapchat ones. Of course a big part of this is the novelty aspect, which will fade, and I follow a lot more people on Instagram than I do on Snapchat. That last point, though, is, well, the point: I and my friends are not exactly Snapchat’s target demographic today, but for the service to reach its potential we will be eventually. Unless, of course, Instagram Stories ends up being good enough…

Instagram and Facebook are smart enough to know that Instagram Stories are not going to displace Snapchat’s place in its users lives. What Instagram Stories can do, though, is remove the motivation for the hundreds of millions of users on Instagram to even give Snapchat a shot.

Unfortunately for Snap, Instagram Stories were more than good enough: they were soon significantly better, particularly once you take performance into account (not just on Android, but on iOS too). In fact, they were so good that Instagram’s user growth actually accelerated after their introduction, despite starting from a base of over 500 million users:1

Instagram's Monthly Active Users

In short, Instagram Stories not only prevented users from defecting to Snapchat, it also took all of Snapchat’s growth, and now, after Snapchat’s re-design snafu, is likely taking users away.

Long-term Versus Short-term

Two weeks ago, after Facebook experienced the largest single day market-cap decline in U.S. corporate history, I argued that, if you looked at the company through any lens but a financial one, the company was stronger than ever.

For all of the company’s travails and controversies over the past few years, its moats are deeper than ever, its money-making potential not only huge but growing both internally and secularly; to that end, what is perhaps most distressing of all to would-be competitors is in fact this quarter’s results: at the end of the day Facebook took a massive hit by choice; the company is not maximizing the short-term, it is spending the money and suppressing its revenue potential in favor of becoming more impenetrable than ever.

That reference to “taking a hit by choice” was primarily about the reduced margins Facebook is projecting thanks to its dramatically increased spending on security; it certainly hurts in the short-term, but keeping people on the platform and regulators away has massive long-term value.

It should be noted, though, that Instagram Stories in particular (along with Stories on Facebook’s app) are in a similar vein: their strategic impact, particularly in terms of Facebook’s competitive position relative to Snapchat, will be felt for years. There is a financial impact in the short-to-medium term, though, and it may be significant.

The News Feed Ad Unit

The foundation of Facebook’s advertising business is the News Feed ad, which I described five years ago as the best display advertising unit ever:

It’s better for an advertising business to not be a platform. There are certain roles and responsibilities a platform must bear with regards to the user experience, and many of these work against effective advertising. That’s why, for example, you don’t see any advertising in Android, despite the fact it’s built by the top advertising company in the world.

So a Facebook app owns the entire screen, and can use all of that screen for what benefits Facebook, and Facebook alone.

Something like this:

The News Feed Ad

You can’t help but see the advertising, which makes it particularly attractive to advertisers. Brand advertising, especially, is all about visuals and video (launching soon!), but no one has been able to make brand advertising work as well on the web as it does on TV or print. There is simply too much to see on the screen at any given time.

This is the exact opposite experience of a mobile app. Brand advertising on Facebook’s app shares the screen with no one. Thanks to the constraints of mobile, Facebook may be cracking the display and brand advertising nut that has frustrated online advertisers for years.

To be sure, the fact that News Feed ads take over the screen isn’t the only reason Facebook’s stock is, even with the recent drop, up 148% since that article: in the intervening years the company has doubled its userbase, increased ad load in the News Feed, and managed to increase its price-per-ad thanks to Facebook’s superior targeting capabilities.

Still, the ad unit itself matters:

Facebook Advertising Growth Metrics

That huge anomaly between 2014 and 2016 marked the dramatic reduction in Facebook’s side-bar ads, which had been the company’s chief money-maker for years; that reduced the number of impressions, but the resultant shift in ad inventory mix to News Feed ads led to an even more dramatic increase in Facebook’s reported price-per-ad.2

The years after the mix shift have only reinforced just how good the News Feed ad unit is: by 2016 the shift was complete and Facebook just grew and grew, quarter after quarter (remember, that chart is growth rates; a chart of absolute numbers would be up-and-to-the-right3). 2017 was even more interesting: the company said it would stop increasing ad load in the News Feed, which is why impressions fell, but the price-per-ad increased in response. This demonstration of pricing power is as clear an indication as you can get that Facebook’s News Feed ad was highly differentiated.

The Stories Challenge

That’s the thing about Stories, though: while more people may use Instagram because of Stories, some significant number of people view Stories instead of the Instagram News Feed, or both in place of the Facebook News Feed. In the long run that is fine by Facebook — better to have users on your properties than not — but the very same user not viewing the News Feed, particularly the Facebook News Feed, may simply not be as valuable, at least for now.

Facebook CFO David Wehner said as much multiple times on Facebook’s recent earnings call (all emphasis mine):

  • In his prepared remarks:

    There are several factors contributing to [revenue growth] deceleration. For example, we expect currency to be a slight headwind in the second half versus the tailwinds we have experienced over the last several quarters. We plan to grow and promote certain engaging experiences like Stories that currently have lower levels of monetization, and we are also giving people who use our services more choices around data privacy, which may have an impact on our revenue growth.

  • In response to a question about monetizing Instagram:

    Instagram has more heavy usage of Stories, so that’s an area of continued growth opportunity because the effective levels of monetization in Stories are lower. On the demand side, we see a good traction across both platforms, and we’re rolling out more ability for advertisers to leverage ads in Stories with more formats and the like. So that’s, again, an important opportunity for growth is just continuing to build out more products on the demand side for Stories.

  • In response to a question about declining revenue growth:

    We’re going to be focusing on growing engaging new experiences like Stories and promoting those. And that’s going to have a negative impact on revenue growth.

COO Sheryl Sandberg summarized the fundamental question with regard to Stories:

We’ve seen great progress with Stories as a format for people to share on our platforms. We have 400 million people sharing with Instagram Stories, 450 million with WhatsApp Status. Facebook is newer, but we’re seeing good progress there. The question is will this monetize at the same rate as News Feed? And we honestly don’t know.

Nor do investors.

The Downside of Stories

In fact, there are two good reasons to be pessimistic: one from a user perspective and one from an advertiser perspective.

From a user perspective Story ads are far easier to skip: simply tap the screen, much as the user has probably already been doing for the last several minutes feverishly trying to catch up on their Story backlog. Part of what makes Stories such fantastic drivers of engagement is the combination of their disappearing nature — better check frequently to not miss anything! — combined with the simple mechanic of viewing them: tap tap tap. An absence of friction, though, is always a challenge when it comes to monetization.

Moreover, effective Story ads are more difficult to make: still photos can be used, but the most engaging ads will be video, which increases the production difficulty significantly — particularly when you remember the importance of grabbing the user’s attention immediately! This is a particular problem for Facebook because so much of its advertising comes from small- and medium-sized businesses. Sheryl Sandberg said on an earnings call a year-and-a-half ago:

We’re really excited to announce today that 65 million businesses are using our free Pages product and 5 million are using Instagram Business profiles. More and more of these businesses are becoming advertisers with over 4 million advertising on Facebook and over 500,000 on Instagram. As a result, our revenue base is becoming more diverse. In Q4, our top 100 advertisers represented less than a quarter of our ad revenue, which is a decline from Q4 last year.

This is tremendous diversification, particularly in comparison to traditional advertising, as this chart from MoffettNathanson shows:

https://stratechery.com/2013/mobile-makes-facebook-just-an-app-thats-great-news/

This is another reason why News Feed ads are so effective: advertising is measured on ROI — return on investment — and the “I” is just as important as the “R”; the investment necessary to achieve a favorable return is so much lower for News Feed ads than nearly any other medium, and experimentation is cheap-and-easy. Small wonder small- and medium-sized businesses flock to Facebook (and a reminder that Facebook’s advertising platform is a critical piece of building the economy of the future); to that end, though, it is fair to wonder just how much of Facebook’s advertiser base will flock to Instagram Stories, even as its users do exactly that.

Stories’ Potential Upside

That noted, Stories potentially have significant value for Facebook beyond strategic positioning. First and foremost, the fact that users find Stories even more engaging than the News Feed suggests there is at least the possibility to create even more engaging advertising as well.

That is particularly compelling because attentive readers may have realized my 2013 excerpt above about the effectiveness of News Feed ads got one thing wrong: I was convinced that taking over the screen would be valuable because of the potential for brand advertising; in fact, most of Facebook’s business is in direct response ads, where the goal is driving user action (app install, product purchase, newsletter signup, etc.) as opposed to simply building brand affinity. The latter remains very television-centric even as young people desert the medium, and every tech company is working feverishly to capture their share of that television money that will be going digital any day now!

Might Stories be an ad unit that works for brand advertising? Sandberg tried to make the case:

With 1 billion active people on the platform, I think Instagram is definitely both a direct response opportunity and an opportunity for discovery. Part of it’s the format. The format is so visually appealing and people are telling stories with pictures, so we see both anecdotally and in the data that this is a great place for people to become aware of a product in the first place.

This is in fact another way to look at that chart about the top 200 advertisers: the biggest spenders are brand advertisers, and while it is to Facebook’s credit that they serve the long tail, there remains a significant opportunity the company has mostly not tapped into, and oh-by-the-way, those are the advertisers best equipped to spend the money to make an effective Stories ad; after all, they’re already spending the money on TV.

That noted, it is possible the pot of brand advertising gold that tech companies are chasing may end up being at the end of the proverbial rainbow, always pursued and never obtained. Do big brands stick with TV because the medium is that much better for advertising, or because they were built for a mass market that on the Internet increasingly doesn’t exist? Indeed, one can make the case — as I have — that the fortunes of traditional television and its advertisers are completely intertwined; unsurprisingly, the disruptors of the latter use Facebook.


There was one small bit of good news for Snap in this analysis: a significant challenge for the company was the de facto requirement that advertisers devote outsized resources to the platform, killing any long-term ROI proposition. The issue is formatting: Snapchat ads are vertical, ideally video, and it was so much easier to simply buy a News Feed ad on Facebook and run it on Instagram. Now, though, an advertiser’s investment in vertical video ads can, at least in theory, be used on both Instagram and Snapchat (and all of Facebook’s other properties after its attempt to put Stories everywhere).

This is small solace to be sure: Instagram has a larger audience — including most of Snapchat’s — the entire Facebook advertising apparatus behind it, and all of the trends are in its favor. That is the sort of long-term gain for which Facebook is rightfully willing to bear the short-term pain; the question now is just how substantial that long-term gain may be, and relatedly, how long the short-term pain will persist.

  1. Facebook has only ever released Instagram MAUs at certain milestones; the growth rate was calculated based on the time in months between those milestones. Also note that Snap, to their credit, releases daily active users, which depresses their numbers relative to Instagram’s monthly active users [↩︎]
  2. The year-over-year increase in price-per-ad growth peaks at 335% in Q4 2014 [↩︎]
  3. Also, such a chart is impossible: Facebook doesn’t actually release the absolute number of impressions or price-per-ad [↩︎]

Free Daily Update: An Interview with Patreon CEO Jack Conte and Memberful CEO Drew Strojny

Stratechery occasionally conducts interviews; these are always in the subscriber-only Daily Update, in part as a benefit for subscribers, but also because Weekly Articles are more representative of what potential subscribers should expect if they choose to receive the Daily Update.

I am making an exception in this case: Patreon just announced that they are acquiring Memberful, the membership software that I use on Stratechery. It is an acquisition and space that is not only pertinent to the business of Stratechery, but also some of the major concepts that I talk about, particularly the evolution of media. To that end, and given the timeliness, this Daily Update is available for anyone to read here.

Facebook Lenses

While I was mostly unplugged on my vacation last week, with the news of Facebook’s disappointing earnings report and subsequent stock decline — the largest one-day loss by any company in U.S. stock market history — I couldn’t resist chiming in on Twitter:

I do regret the tweet a tad, and not only because “chiming in on Twitter” is always risky. Back when Stratechery started I wrote in the very first post that one of the topics I looked forward to exploring was “Why Wall Street is not completely insane”; I was thinking at the time about Apple, a company that, especially at that time, was regularly posting eye-popping revenue and profit numbers that did not necessarily lead to corresponding increases in the stock price, much to the consternation of Apple shareholders. The underlying point should be an obvious one: a stock price is about future earnings, not already realized ones; that the iPhone maker had just had a great quarter was an important signal about the future, but not a determinant factor, and that those pointing to the past to complain about a price predicated on the future were missing the point.

Of course that is exactly what I did in that tweet.

It’s worth noting, though, that while the explicit reasoning of those Apple stockholders may have been suspect, their sentiment has proven correct: in April 2013 Apple reported quarterly revenue of $43.6 billion and profit of $9.5 billion, and the day I started Stratechery the stock price was $63.25; five years later Apple reported quarterly revenue of $61.1 billion and profit of $13.8 billion, and on Friday the stock price was $190.98.

To be clear, I agreed with the Apple-investor sentiment all along: several of my early articles — Apple the Black Swan, Two Bears, and especially What Clayton Christensen Got Wrong — were about making the case that Apple’s business was far more sustainable with much deeper moats than most people realized, and it was that sustainability and defensibility that mattered more than any one quarter’s results.

The question is if a similar case can be made for Facebook: certainly my tweet taken literally was naive for the exact reasons those Apple investor complaints missed the point five years ago; what about the sentiment, though? Just how good of a business is Facebook?

As with many such things, it all depends on what lens you use to examine the question.

Lens 1: Facebook’s Finances

As is often the case with earnings, the move in Facebook’s stock was only a bit about the results and whole lot about future expectations. On Wednesday, Facebook’s stock closed at $217, but then its earnings showed revenue of $13.2 billion, slightly below Wall Street’s expectations; unsurprisingly, the stock slid about 8% in after-hours trading to around $200. The real drop was spurred by two comments on the earnings call from Facebook CFO Dave Wehner about Facebook’s expectations going forward.

First, with regards to revenue:

Turning now to the revenue outlook; our total revenue growth rate decelerated approximately 7 percentage points in Q2 compared to Q1. Our total revenue growth rates will continue to decelerate in the second half of 2018, and we expect our revenue growth rates to decline by high-single digit percentages from prior quarters sequentially in both Q3 and Q4.

Second, with regards to operating margin:

Turning now to expenses; we continue to expect that full-year 2018 total expenses will grow in the range of 50% to 60% compared to last year…Looking beyond 2018, we anticipate that total expense growth will exceed revenue growth in 2019. Over the next several years, we would anticipate that our operating margins will trend towards the mid-30s on a percentage basis.

From a purely financial perspective, both pieces of news are are less than ideal but at least understandable. In terms of revenue, Facebook’s growth is from a very large base, which means that this quarter’s 42% year-over-year revenue growth to $13.2 billion from $9.3 billion is, in absolute terms, 36% greater than the year ago’s 45% revenue growth (from $6.4 billion). To put it in simpler terms, massive growth rates inevitably decline even as massive absolute growth remains; as a point of comparison, Google in the same relative timeframe (14 years after incorporation) grew 35 percent to $12.21 billion (i.e. Facebook is better on both metrics).

As far as the operating margin decline, in a normal company — i.e. one with marginal costs — a revenue decrease would not necessarily lead to a meaningful decline in margin since selling fewer products would mean lower costs of goods sold. Facebook, of course, is not a normal company: the only marginal costs for the ads they sell are credit card fees; like most tech companies the vast majority of costs are “below the line” (mostly in Research & Development, but also Sales & Marketing and General & Administrative); it follows, then, that a decrease in revenue growth would, absent an explicit effort to decrease unrelated (to revenue) expense growth, lead to lower operating margins.

In fact, Facebook is not only not decreasing expenses, they are going in the opposite direction; expenses are growing faster than ever, even as revenue growth clearly fell off:

Facebook's revenue growth is decreasing even as its expense growth increases

I suspect it is this chart, more than anything else, that explains the drop in Facebook’s stock price: it’s not one thing or the other; it is both revenue growth slowing and expenses accelerating at the same time, with all indications from management are that the trends will continue.

Again, relatively speaking Facebook is in great shape financially — I already noted the company had better revenue and growth numbers than Google at a similar point, and their operating margins are substantially better as well — but there’s no question this is a pretty substantial shift in the company’s longterm outlook. The financial lens still provides a pretty positive view, but it is indeed less positive than before.

Lens 2: Facebook’s Products

It is always a bit confusing to write about Facebook, because there is both Facebook the company and Facebook the product, and there is no question the greatest amount of negativity has, for several years now, been centered around the latter. To that end, it is tempting to conflate the two; for example, the New York Times wrote in an article headlined Facebook Starts Paying a Price for Scandals:

For nearly two years, Facebook has appeared bulletproof despite a series of scandals about the misuse of its giant social network. But the Silicon Valley company’s streak ended on Wednesday when it said that the accumulation of issues was starting to hurt its multibillion-dollar business — and that the costs are set to continue playing out for months.

This is true as far as it goes, particular when it comes to expenses: Facebook is on pace to increase its security and content review teams to 20,000 people, a three-fold increase in 18 months; that is why CEO Mark Zuckerberg warned on an earnings call last year:

I’ve directed our teams to invest so much in security on top of the other investments we’re making that it will significantly impact our profitability going forward, and I wanted our investors to hear that directly from me. I believe this will make our society stronger, and in doing so will be good for all of us over the long term. But I want to be clear about what our priority is. Protecting our community is more important than maximizing our profits.

What is much less clear is what effect, if any, Facebook’s controversies have had on the top line. There were three factors that for many years made Facebook a monster when it came to revenue growth:

  • The number of users was increasing
  • Ad load (the number of ads shown in the News Feed) was increasing
  • The price-per-ad was increasing

A year ago, though, Facebook stopped increasing ad load; as I have documented, this did result in an even sharper increase in the price paid per-ad, but it was still a retardant on growth.

Then, over the last year, Facebook’s user growth started to slow, and in the most-profitable North American region, has effectively plateaued. That, though, isn’t because of Facebook’s controversies: it is because the app has run out of people! The company has 241 million monthly active users in the US & Canada, 65% of the total population of 372 million (including children who aren’t supposed to have accounts before the age of 13).

Given that degree of nearly total penetration, what is more important when it comes to evaluating Facebook’s health is that there is no indication the company is losing users. Sure, the numbers in North America decreased by a million in Q4 2017, but now that million is back; I expect something similar when it comes to the million users the company lost in Europe when it required affirmative consent from users to continue using the app because of GDPR.

The fact of the matter is that nothing has happened to diminish Facebook’s moat when it comes to attracting and retaining users: the number one feature of a social network is how many people are on it, and for all intents and purposes everyone is on Facebook — whether they like it or not.

Interestingly, Facebook is working to deepen that moat even further with its focus on Groups. Zuckerberg said on the earnings call:

There are more than 200 million people that are members of meaningful groups on Facebook, and these are communities that, upon joining, they become the most important part of your Facebook experience and a big part of your real world social infrastructure. These are groups for new parents, for people with rare diseases, for volunteering, for military families deployed to a new base and more.

We believe there is a community for every one on Facebook. And these meaningful communities often spend online and offline and bring people together in person. We found that every great community has an engaged leader. But running a group can take a lot of time. So we have a road map to make this easier. That will enable more meaningful groups to get formed, which will help us to find relevant ones to recommend to you, and eventually achieve our five-year goal of helping 1 billion people be a part of meaningful communities.

Zuckerberg is referring to his 2017 manifesto Building a Global Community; it is a particularly attractive goal from Facebook’s perspective because it makes the product stickier than ever.

All that noted, the most important reason to view Facebook through the lens of the company’s products is that the sheer scale of Facebook the app makes it easy to lose site of the still substantial growth potential of those other products. Instagram in particular recently passed 1 billion users, which is an incredible number that is still less than half of Facebook the app’s total users; by definition Instagram has reached less than half of its addressable market.

Moreover, Instagram has not only been untouched by Facebook’s controversies, it is such a compelling product that, anecdotally speaking, most “Facebook-nevers” or “Facebook-quitters” readily admit to using the service daily. The app also hasn’t come close to reaching its monetization potential: while the feed carries the same ad load as Facebook, the SnapChat-inspired Stories format that has exploded in usage has barely been monetized; in fact, Facebook’s executives attributed some of the company’s slowing revenue growth to increased Stories usage (instead of the feed). From a purely financial perspective this is certainly a cause for concern, but from a strategic perspective it means that Instagram is in an even stronger position that it was previously. Remember, revenue and profit are lagging indicators, and the explosion in Instagram Stories is an extreme example of why that is such an important fact to keep in mind.

WhatsApp is increasingly compelling as well: not only does the app remain the dominant communications medium in much of the world, but the addition of WhatsApp Status updates and Stories dramatically increases the monetization potential of the service — a potential that Facebook hasn’t even started to realize.1

There is some degree of long-term risk when it comes to products: Facebook acquired both Instagram and WhatsApp, but the company should not be allowed to acquire another social network of similar size and velocity to those two, and I doubt they would be. That concern, though, is very far in the future: for now the product lens suggests that Facebook is as strong as ever.

Lens 3: Facebook’s Advertising Infrastructure

This lens takes the exact opposite perspective of Lens 2; looking at the company from a product perspective shows four different apps, but looking at the company from an advertising perspective shows a single integrated machine.

This was a point Facebook executives touched on repeatedly in last week’s earnings call. Here is Wehner (emphasis mine):

In terms of Facebook versus Instagram, they’re obviously both contributing to revenue growth. Instagram is growing more quickly and making an increasing contribution to growth. And we’ve been pleased with how Instagram is growing. Facebook and Instagram are really one ads ecosystem.

Zuckerberg:

We’re also making progress developing Stories into a great format for ads. We’ve made the most progress here on Instagram, but this quarter, we started testing Stories ads on Facebook too…

COO Sheryl Sandberg added:

Since we have so many different places where you have Stories formats in Instagram and WhatsApp and Facebook, as volume increases of the opportunity, advertisers get more interested.

Zuckerberg and Sandberg were obviously talking about the potential for advertising in Stories, but that potential is simply a repeat of what has already happened with Feed ads: Facebook spent years building out News Feed advertising — not simply the display and targeting technology but also the entire back-end apparatus for advertisers, connections with non-Facebook data sources and points-of-sale, relationships with ad buyers, etc. — and then simply plugged Instagram into that infrastructure.

The payoff of this integrated approach cannot be overstated. Instagram got to scale in terms of monetization years faster than they would have on their own, even as the initial product team had the freedom to stay focused on the user experience. Facebook the app benefited as well, because Instagram both increased the surface area for Facebook ad campaigns even as it increased Facebook’s targeting capabilities.

The biggest impact, though is on potential competition. It is tempting to focus on the “R” in “ROI” — the return on investment — and as I just noted Instagram + Facebook makes that even more attractive. Just as important, though, is the “I”; there is tremendous benefit to being a one-stop shop for advertisers, who can save time and money by focusing their spend on Facebook. The tools are familiar, the buys are made across platforms, and as Zuckerberg and Sandberg alluded to with regard to Stories, the ads themselves only need to be made once to be used across multiple platforms. Why even go to the trouble to advertise anywhere else?

This is why the advertising lens is perhaps the most useful when it comes to understanding just how strong Facebook’s business remains, and why the Instagram acquisition in particular was such a big deal. For all the discussion of Facebook the app’s lock-in, it is very reasonable to wonder if engagement is decreasing over time, particularly amongst young people, or if controversies may drive down usage — or worse. Were Instagram a separate company, advertisers might find themselves with no choice but to spread out their advertising to multiple companies, and once their advertising was diversified, it would be a much smaller step to target users on other networks like SnapChat or Twitter. As it stands there is no reason to leave Facebook the advertising platform, no matter what happens with Facebook the app.

Lens 4: Facebook’s Multiplying Moats

Facebook’s advertising moat may be its most important, and its network moat its strongest, but the company has actually added moats, particularly in the last year.

The first is GDPR; this may seem counter-intuitive, given that Facebook said last week the regulation cost them a million users, and that one of the factors that would hurt revenue growth was the increased controls the company was giving users when it comes to controlling their personal information. Keep in mind, though, that GDPR applies to everyone, not just Facebook, and as Sandberg noted on the call (emphasis mine):

Advertisers are still adapting to the changes, so it’s early to know the longer-term impact. And things like GDPR and other privacy changes that may happen from us or may happen with regulation could make ads more relevant. One thing that we know that’s not going to change is that advertisers are always looking for the highest ROI opportunity. And what’s most important in winning budget is our relative performance in the industry, and we believe we’ll continue to do very well on that.

I made this exact point previously:

While GDPR advocates have pointed to the lobbying Google and Facebook have done against the law as evidence that it will be effective, that is to completely miss the point: of course neither company wants to incur the costs entailed in such significant regulation, which will absolutely restrict the amount of information they can collect. What is missed is that the increase in digital advertising is a secular trend driven first-and-foremost by eyeballs: more-and-more time is spent on phones, and the ad dollars will inevitably follow. The calculation that matters, then, is not how much Google or Facebook are hurt in isolation, but how much they are hurt relatively to their competitors, and the obvious answer is “a lot less”, which, in the context of that secular increase, means growth.

Secondly, all of those costs that Facebook are incurring for security and content review that are reducing operating margin? Perhaps the stock market would feel better if they were characterized as moat expansion, because that’s exactly what they are: any would-be Facebook competitor is going to have to make a similar investment, and do it from a dramatically lower revenue base.

Moreover, just as Facebook benefits from scaling its ad infrastructure to all of its products, it can do the same with its security efforts. Zuckerberg stated:

More broadly, our strategy is to use Facebook’s computing infrastructure, business platforms and security systems to serve people across all of our apps…We’re using AI systems in our global community operations team to fight spam, harassment, hate speech, and terrorism across all of our apps to keep people safe. And this is incredibly useful for apps like WhatsApp and Instagram as it helps us manage the challenges of hyper-growth there more effectively.

This is why the lens with which you view Facebook matters so much: the exact same set of facts viewed from a financial perspective are a clear negative; from a moat perspective they are a clear positive.

Lens 5: Facebook’s Raison D’être

Needless to say, once you view Facebook through anything but a financial lens the health of the business is hard to argue with (and frankly, the finances went from phenomenal to fantastic, but it’s all relative). That’s why I can’t help but wonder if there is something more fundamental about both the collapse in Facebook’s stock and the general celebration that followed.

To return to the early years of Stratechery, it was striking how widespread Facebook skepticism was; I first tried to argue otherwise five years ago today, and in 2015 felt compelled to write The Facebook Epoch that begins like this:

I’m fond of saying that few companies are as underrated as Facebook is, especially in Silicon Valley. Admittedly, it seems strange to say such a thing about a $245 billion company with a trailing 12-month P/E ratio of 88, but that is Wall Street sentiment; in the tech bubble many seem to simply assume the company is ever on the brink of teetering “just like MySpace”, never mind the fact that the social network pioneer barely broke 100 million registered users, less than 10% of the number of active users Facebook attracted in a single day late last month. Or, as more sober minds may argue, sure, Facebook looks unstoppable today, but then again, Google looked unstoppable ten years ago when social seemingly came out of nowhere: surely the Facebook killer is imminent!

That sentiment sure seems to be back in full force!

At the risk of veering into broad-based psychoanalysis, I think a lot of the Facebook skepticism is because so much of the content seems so shallow and petty, or in the case of the last few years, actively malicious. How can such a product survive?

In fact, it survives for the very reason it exists: Facebook began in Zuckerberg’s Harvard dorm room by quite literally digitizing offline relationships that already existed, both in real life and in actual physical “facebooks”. Facebook is so powerful because of this direct connection to the real world: it is shallow and petty and sometimes malicious — and yes, often good — because we humans are shallow and petty and sometimes malicious — and yes, often good.

By extension, to insist that Facebook will die any day now is in some respects to suggest that humanity will cease to exist any day now; granted, it is a company and companies fail, but even if Facebook failed it would only be a matter of time before another Facebook rose to replace it.

That seems unlikely: for all of the company’s travails and controversies over the past few years, its moats are deeper than ever, its money-making potential not only huge but growing both internally and secularly; to that end, what is perhaps most distressing of all to would-be competitors is in fact this quarter’s results: at the end of the day Facebook took a massive hit by choice; the company is not maximizing the short-term, it is spending the money and suppressing its revenue potential in favor of becoming more impenetrable than ever.

“Utter disaster” indeed.

  1. There is Messenger as well; I am more dubious of its long-term monetization potential because its natural advertising space — status updates and stories — are basically what Facebook is [↩︎]

The European Commission Versus Android

To understand how Google ended up with a €4.3 billion fine and a 90-day deadline to change its business practices around Android, it is critical to keep one date in mind: July 2005.1 That was when Google acquired a still in-development mobile operating system called Android, and to put the acquisition in context, Steve Jobs was, at least publicly, “not convinced people want to watch movies on a tiny little screen”. He was, of course, referring to the iPod; Apple would go on to release an iPod with video playback a few months later, but the iPhone was still a year-and-a-half away from being revealed.

In other words, Android, at least at the beginning, wasn’t a response to Apple;2 the real target was Microsoft (and to a lesser extent Blackberry), which seemed poised to dominate smartphones just as they had the desktop. That was an untenable situation for Google; then Vice-President of Product Management Sundar Pichai wrote on the Google Public Policy blog about the company’s challenges on PCs:

Google believes that the browser market is still largely uncompetitive, which holds back innovation for users. This is because Internet Explorer is tied to Microsoft’s dominant computer operating system, giving it an unfair advantage over other browsers. Compare this to the mobile market, where Microsoft cannot tie Internet Explorer to a dominant operating system, and its browser therefore has a much lower usage.

What mattered to Google was access to end users: that is what makes the Aggregation flywheel turn. On PCs the company had succeeded through a combination of flat-out being better, the fact that it was very simple to visit a new URL (and make it your homepage), and deals with OEMs to set Google as the homepage from the beginning. All would be more difficult to achieve on mobile, at least mobile as it was understood in 2005: applications were notoriously difficult to find and install, and Microsoft and Blackberry had locked down their operating systems to a much greater extent than Microsoft had on the PC.

Thus the Android gambit: Google decided to take on Microsoft directly in mobile operating systems, and its most powerful tool would not be the quality of the operating system, but the business model. To that end, while Google did, naturally, retool Android’s user interface once the iPhone was announced, the business model remained Microsoft kryptonite: whereas Microsoft charged a per-device licensing fee, just as it had with Windows, Android would not only be free and open-source, Google would actually share search revenue derived from Android with OEMs that installed the operating system.

Of course Android also ended up being a much better experience than Windows Mobile in the post-iPhone world, and the deal was irresistible to OEMs flailing for a response to the iPhone: get a (somewhat) comparable (sort-of) touch-based operating system for free, and even make money after the initial sale! Indeed, not only did Android effectively kill Microsoft’s mobile efforts, it went on to take over the world via a massive ecosystem of device makers and mobile carriers that competed to drive down costs and increase distribution.

Android’s Success

That Android increases competition was the focus of Pichai’s — now the CEO of Google — latest blog post in response to the ruling:

Today, the European Commission issued a competition decision against Android, and its business model. The decision ignores the fact that Android phones compete with iOS phones, something that 89 percent of respondents to the Commission’s own market survey confirmed. It also misses just how much choice Android provides to thousands of phone makers and mobile network operators who build and sell Android devices; to millions of app developers around the world who have built their businesses with Android; and billions of consumers who can now afford and use cutting-edge Android smartphones. Today, because of Android, there are more than 24,000 devices, at every price point, from more than 1,300 different brands…

What Pichai doesn’t say is that this competition is not so much a feature as it was the point: open-sourcing Android commoditized smartphone development meaning anyone could enter, even as few were in a position to profit over time. That included Google, at least at the beginning, which was by design: remember, the point of Android was not to make money like Windows, it was to stop Windows or any other operating system from getting between Google and users. Venture capitalist Bill Gurley explained in a 2011 post entitled The Freight Train That Is Android:

Android, as well as Chrome and Chrome OS for that matter, are not “products” in the classic business sense. They have no plan to become their own “economic castles.” Rather they are very expensive and very aggressive “moats,” funded by the height and magnitude of Google’s castle [(search advertising)]. Google’s aim is defensive not offensive. They are not trying to make a profit on Android or Chrome. They want to take any layer that lives between themselves and the consumer and make it free (or even less than free). Because these layers are basically software products with no variable costs, this is a very viable defensive strategy. In essence, they are not just building a moat; Google is also scorching the earth for 250 miles around the outside of the castle to ensure no one can approach it. And best I can tell, they are doing a damn good job of it.

Indeed they were, but the strategy had a built-in problem: Android was, well, open source, and just as that helped Android spread, it could just as easily be forked into an initially compatible operating system that didn’t connect to Google’s services — the castle that Google was trying to protect all along. Google needed a wall for its moat, and found one in the Google Play Store.

The Google Play Store and Google Play Services

The Google Play Store, not unlike Android’s user interface, was a response to the iPhone, specifically the highly successful launch of the App Store in 2008. And while the Play Store often lagged the App Store when it came to cutting-edge apps, particularly in the early days, it quickly became one of Google’s most valuable services, both in terms of making Android useful as well as making Google money.

Note, though, that the Play Store is not a part of Android: it has always been closed-source and exclusive to Google’s version of Android, just like other Google services like Gmail, Maps, and YouTube. The problem Google had with all of those apps, though, was that they were updated with the operating system, and OEMs and carriers — who only made money when a device was initially sold — were not particularly incentivized to update the operating system.

Google’s solution was Google Play Services; first released in 2010 as a part of Android 2.2 Froyo, Google Play Services was distributed via the Play Store and provided an easily updatable API layer that would, in the initial version, allow Google to update its own apps independent of operating system updates. It was an elegant solution to a real problem inherent in the free-wheeling model Google had taken for Android distribution: widespread fragmentation. Soon all of Google’s apps were built on top of Google Play Services, and then, in 2012, Google started opening it up to developers.

The initial version was quite modest; here is the announcement on Google+:

At Google I/O we announced a preview of Google Play services, a new development platform for developers who want to integrate Google services into their apps. Today, we’re kicking off the full launch of Google Play services v1.0, which includes Google+ APIs and new OAuth 2.0 functionality. The rollout will cover all users on Android 2.2+ devices running the latest version of the Google Play store.

Over the next several years, though, Google devoted more and more of its effort — and its most interesting APIs, like location and maps and gaming services — to Google Play Services; meanwhile, whatever equivalent service was in the open-source version of Android was effectively frozen in time. The net result is incredibly significant to teasing out this case: Google Play Services silently shifted ever more apps from Android apps to Google Play apps; today, no Google app will function on open-source Android without extensive reworking, and the same applies to ever more 3rd-party apps as well.

That noted, it is hard, in my estimation, to see this as being an antitrust violation. The fact of the matter is that Google was addressing a legitimate problem in the Android ecosystem, and the company didn’t make any developer use Google Play Services APIs instead of the more basic ones still available even today.

The European Commission Case

The European Commission found Google guilty of breaching EU antitrust rules in three ways:

  • Illegally tying Google’s search and browser apps to the Google Play Store; to get the Google Play Store and thus a full complement of apps, OEMs have to pre-install Google search and Chrome and make them available within one screen of the home page.
  • Illegally paying OEMs to exclusively pre-install Google Search on every Android device they made.
  • Illegally barring OEMs that installed Google’s apps from selling any device that ran an Android fork.

Taken in isolation, these seem to run from least problematic to most problematic.

  • The Google Play Store has always been an exclusive Google app; it seems that Google ought to be able to distribute it exclusively as part of a bundle if it so chooses.
  • Pinning all revenue from Google Search to exclusivity on all devices quite obviously makes it very difficult for alternative search services to build share (as they lack access to pre-installs, one of the most effective channels for customer acquisition); this seems to be more of a Google Search dominance issue than an Android dominance issue though.
  • Predicating the availability of any of Google’s apps, including the Google Play Store, on OEMs not taking advantage of the open source nature of Android on devices that will not include Google apps seems much more problematic than Google insisting its apps be distributed in a bundle. The latter is Google’s prerogative; the former is dictating OEM actions just because Google can.

This is where the history of Android matters; before Google Play Services, the primary challenge in building a competitive fork of Android would have been convincing developers to upload their apps to a new app store (since Google would obviously not want to put its apps, including the Play Store, on said fork). That fork, though, never materialized because of Google’s contractual terms barring OEMs from selling any devices built on such a fork.

Today the situation is very different: that contractual limitation could go away tomorrow (or, more accurately, in 90 days), and it wouldn’t really matter because, as I explained above, many apps are no longer Android apps but are rather Google Play apps. To run on an Android fork is by no means impossible, but most would require more rework than simply uploading to a new App Store.

In short, in my estimation the real antitrust issue is Google contractually foreclosing OEMs from selling devices with non-Google versions of Android; the only way to undo that harm in 2018, though, would be to make Google Play Services available to any Android fork.

The Commission’s Remedies

To be sure, that’s not exactly what the European Commission ordered (in fact, “Google Play Services” does not appear a single time in the press release); the Commission seems to feel that the three issues do stand alone. That means that Google has to respond to each individually:

  • Google has to untie the Play Store from Search and the Chrome browser
  • Google has already stopped paying OEMs for portfolio-wide search exclusivity
  • Google can no longer stop OEMs from selling devices with Android forks

The most momentous by far is the first (despite the fact it is the weakest allegation, in my estimation). Samsung, or any other OEM, could in 90 days sell a device with Bing search only and the Google Play Store (where of course Google Search could be downloaded). This will likely accrue to consumers’ benefit: Microsoft, Google, and other providers will soon be bidding to be the default search option, and, given the commoditized nature of Android devices, it is likely that most of what they are willing to pay will go towards lower prices.

Still, it is an unsatisfying remedy: Google built Android for the express purpose of monetizing search, and to be denied that by regulatory edict feels off; Google, though, bears a lot of the blame for going too far with its contracts.

More broadly, the European Commission continues to be a bit too cavalier about denying companies — well, Google, mostly — the right to monetize the products they spend billions of dollars at significant risk to develop; this was my chief objection to last year’s Google Shopping case. In this case I narrowly come down on the Commission’s side almost by accident: I think Google acted illegally by contractually foreclosing Android competitors at a time when it might have made a difference, but I am concerned that the Commission’s publicly released reasoning doesn’t seem to grasp exactly how Android has developed, the choices Google made, and why.

That noted, I highly doubt Google would do anything differently: when it comes to the company’s goals, Android could not be a bigger success — if anything, this ruling is evidence of just how successful the product was.

  1. The exact date of the acquisition is unknown [↩︎]
  2. For those wondering, then-Google CEO Eric Schmidt didn’t join Apple’s Board of Directors until August 2006 [↩︎]

Intel and the Danger of Integration

Last week Brian Krzanich resigned as the CEO of Intel after violating the company’s non-fraternization policy. The details of Krzanich’s departure, though, ultimately don’t matter: his tenure was an abject failure, the extent of which is only now coming into view.

Intel’s Obsolete Opportunity

When Krzanich was appointed CEO in 2013 it was already clear that arguably the most important company in Silicon Valley’s history was in trouble: PCs, long Intel’s chief money-maker, were in decline, leaving the company ever more reliant on the sale of high-end chips to data centers; Intel had effectively zero presence in mobile, the industry’s other major growth area.

Still, I framed the situation that faced Krzanich as an opportunity, and drew a comparison to the challenges that faced the legendary Andy Grove three decades ago:

By the 1980s, though, it was the microprocessor business, fueled by the IBM PC, that was driving growth, while the DRAM business was fully commoditized and dominated by Japanese manufacturers. Yet Intel still fashioned itself a memory company. That was their identity, come hell or high water.

By 1986, said high water was rapidly threatening to drag Intel under. In fact, 1986 remains the only year in Intel’s history that they made a loss. Global overcapacity had caused DRAM prices to plummet, and Intel, rapidly becoming one of the smallest players in DRAM, felt the pain severely. It was in this climate of doom and gloom that Grove took over as CEO. And, in a highly emotional yet patently obvious decision, he once and for all got Intel out of the memory manufacturing business.

Intel was already the best microprocessor design company in the world. They just needed to accept and embrace their destiny.

Fast forward to the challenge that faced Krzanich:

It is into a climate of doom and gloom that Krzanich is taking over as CEO. And, in what will be a highly emotional yet increasingly obvious decision, he ought to commit Intel to the chip manufacturing business, i.e. manufacturing chips according to other companies’ designs.

Intel is already the best microprocessor manufacturing company in the world. They need to accept and embrace their destiny.

That article is now out of date: in a remarkable turn of events, Intel has lost its manufacturing lead. Ben Bajarin wrote last week in Intel’s Moment of Truth:

Not only has the competition caught Intel they have surpassed them. TSMC is now sampling on 7nm and AMD will ship their architecture on 7nm technology in both servers and client PCs ahead of Intel. For those who know their history, this is the first time AMD has ever beat Intel to a process node. Not only that, but AMD will likely have at least an 18 month lead on Intel with 7nm, and I view that as conservative.

As Bajarin notes, 7nm for TSMC (or Samsung or Global Foundries) isn’t necessarily better than Intel’s 10nm; chip-labeling isn’t what it used to be. The problem is that Intel’s 10nm process isn’t close to shipping at volume, and the competition’s 7nm processes are. Intel is behind, and its insistence on integration bears a large part of the blame.

Intel’s Integrated Model

Intel, like Microsoft, had its fortunes made by IBM: eager to get the PC an increasingly vocal section of its customer base demanded out the door, the mainframe maker outsourced much of the technology to third party vendors, the most important being an operating system from Microsoft and a processor from Intel. The impact of the former decision was the formation of an entire ecosystem centered around MS-DOS, and eventually Windows, cementing Microsoft’s dominance.

Intel was a slightly different story; while an operating system was simply bits on a disk, and thus easily duplicated for all of the PCs IBM would go on to sell, a processor was a physical device that needed to be manufactured. To that end IBM insisted on having a “second source”, that is, a second non-Intel manufacturer for Intel’s chips. Intel chose AMD, and licensed first the 8086 and 8088 designs that were in the original IBM PC, and later, again under pressure from IBM, the 80286 design; the latter was particularly important because it was designed to be upward compatible with everything that followed.

This laid the groundwork for Intel’s strategy — and immense profitability — for the next 35 years. First off, the dominance of Intel’s x86 design was assured thanks to its integration with DOS/Windows: specifically, DOS/Windows created a two-sided market of developers and PC users, and DOS/Windows ran on x86.

Microsoft and Intel were integrated in the PC value chain

However, thanks to its licensing deal with AMD, Intel wasn’t automatically entitled to all of the profits that would result from that integration; thus Intel doubled-down on an integration of its own: the design and manufacture of x86 chips. That is, Intel would invest huge sums of money into creating new and faster designs (the 386, the 486, the Pentium, etc.), and also invest huge sums of money into ever smaller and more efficient manufacturing processes that would push the limits of Moore’s Law. This one-two punch would ensure that, despite AMD’s license, Intel’s chips would be the only realistic choice for PC makers, allowing the company to capture the vast majority of the profits created by the x86’s integration with DOS/Windows.

Intel was largely successful. AMD did take the performance crown around the turn of the century with the Athlon 64, but the company was unable to keep up with Intel financially when it came to fabs, and Intel illegally leveraged its dominant position with OEMs to keep them buying mostly Intel parts; then, a few years later, Intel not only took back the performance lead with its Core architecture, but settled into the “tick-tock” strategy where it alternated new designs and new manufacturing processes on a regular schedule. The integration advantage was real.

TSMC’s Modular Approach

In the meantime there was a revolution brewing in Taiwan. In 1987, Morris Chang founded Taiwan Semiconductor Manufacturing Company (TSMC) promising “Integrity, commitment, innovation, and customer trust”. Integrity and customer trust referred to Chang’s commitment that TSMC would never compete with its customers with its own designs: the company would focus on nothing but manufacturing.

This was a completely novel idea: at that time all chip manufacturing was integrated a la Intel; the few firms that were only focused on chip design had to scrap for excess capacity at Integrated Device Manufacturers (IDMs) who were liable to steal designs and cut off production in favor of their own chips if demand rose. Now TSMC offered a much more attractive alternative, even if their manufacturing capabilities were behind.

In time, though, TSMC got better, in large part because it had no choice: soon its manufacturing capabilities were only one step behind industry standards, and within a decade had caught-up (although Intel remained ahead of everyone). Meanwhile, the fact that TSMC existed created the conditions for an explosion in “fabless” chip companies that focused on nothing but design. For example, in the late 1990s there was an explosion in companies focused on dedicated graphics chips: nearly all of them were manufactured by TSMC. And, all along, the increased business let TSMC invest even more in its manufacturing capabilities.

Integrated intel was competing with a competitive modular ecosystem

This represented into a three-pronged assault on Intel’s dominance:

  • Many of those new fabless design companies were creating products that were direct alternatives to Intel chips for general purpose computing. The vast majority of these were based on the ARM architecture, but also AMD in 2008 spun off its fab operations (christened GlobalFoundries) and became a fabless designer of x86 chips.
  • Specialized chips, designed by fabless design companies, were increasingly used for operations that had previously been the domain of general purpose processors. Graphics chips in particular were well-suited to machine learning, cryptocurrency mining, and other highly “embarrassingly parallel” operations; many of those applications have spawned specialized chips of their own. There are dedicated bitcoin chips, for example, or Google’s Tensor Processing Units: all are manufactured by TSMC.
  • Meanwhile TSMC, joined by competitors like GlobalFoundries and Samsung, were investing ever more in new manufacturing processes, fueled by the revenue from the previous two factors in a virtuous cycle.

Intel’s Straitjacket

Intel, meanwhile, was hemmed in by its integrated approach. The first major miss was mobile: instead of simply manufacturing ARM chips for the iPhone the company presumed it could win by leveraging its manufacturing to create a more-efficient x86 chip; it was a decision that evinced too much knowledge of Intel’s margins and not nearly enough reflection on the importance of the integration between DOS/Windows and x86.

Intel took the same mistaken approach to non general-purpose processors, particularly graphics: the company’s Larrabee architecture was a graphics chip based on — you guessed it — x86; it was predicated on leveraging Intel’s integration, instead of actually meeting a market need. Once the project predictably failed Intel limped along with graphics that were barely passable for general purpose displays, and worthless for all of the new use cases that were emerging.

The latest crisis, though, is in design: AMD is genuinely innovating with its Ryzen processors (manufactured by both GlobalFoundries and TSMC), while Intel is still selling varations on Skylake, a three year-old design. Ashraf Eassa, with assistance from a since-deleted tweet from a former Intel engineer, explains what happened:

According to a tweet from ex-Intel engineer Francois Piednoel, the company had the opportunity to bring all-new processor technology designs to its currently shipping 14nm technology, but management decided against it.

my post was actually pointing out that market stalling is more troublesome than Ryzen, It is not a good news. 2 years ago, I said that ICL should be taken to 14nm++, everybody looked at me like I was the craziest guy on the block, it was just in case … well … now, they know

— François Piednoël (@FPiednoel) April 26, 2018

The problem in recent years is that Intel has been unable to bring its major new manufacturing technology, known as 10nm, into mass production. At the same time, the issues with 10nm seemed to catch Intel off-guard. So, by the time it became clear that 10nm wouldn’t go into production as planned, it was too late for Intel to do the work to bring one of the new processor designs that was originally developed to be built on the 10nm technology to its older 14nm technology…

What Piednoel is saying in the tweet I quoted above is that when management had the opportunity to start doing the work to bring their latest processor design, known as Ice Lake (abbreviated “ICL” in the tweet), [to the 14nm process] they decided against doing so. That was likely because management truly believed two years ago that Intel’s 10nm manufacturing technology would be ready for production today. Management bet incorrectly, and Intel’s product portfolio is set to suffer as a result.

To put it another way, Intel’s management did not break out of the integration mindset: design and manufacturing were assumed to be in lockstep forever.

Integration and Disruption

It is perhaps simpler to say that Intel, like Microsoft, has been disrupted. The company’s integrated model resulted in incredible margins for years, and every time there was the possibility of a change in approach Intel’s executives chose to keep those margins. In fact, Intel has followed the script of the disrupted even more than Microsoft: while the decline of the PC finally led to The End of Windows, Intel has spent the last several years propping up its earnings by focusing more and more on the high-end, selling Xeon processors to cloud providers. That approach was certainly good for quarterly earnings, but it meant the company was only deepening the hole it was in with regards to basically everything else. And now, most distressingly of all, the company looks to be on the verge of losing its performance advantage even in high-end applications.

This is all certainly on Krzanich, and his predecessor Paul Otellini. Then again, perhaps neither had a choice: what makes disruption so devastating is the fact that, absent a crisis, it is almost impossible to avoid. Managers are paid to leverage their advantages, not destroy them; to increase margins, not obliterate them. Culture more broadly is an organization’s greatest asset right up until it becomes a curse. To demand that Intel apologize for its integrated model is satisfying in 2018, but all too dismissive of the 35 years of success and profits that preceded it.

So it goes.

AT&T, Time Warner, and the Need for Neutrality

The first thing to understand about the decision by a federal judge to approve AT&T’s acquisition of Time Warner, over the objection of the U.S. Department of Justice, is that it is very much in-line with the status quo: this is a vertical merger, and both the Department of Justice and the courts have defaulted towards approving such mergers for decades.1

Second, that there is an explosion of merger activity in and between the television production and distribution space is hardly a surprise: the Multichannel Video Programming Distributor (MVPD) business — that is, television distributed by cable, broadband, or satellite — has been shrinking for years now, and in a world where the addressable market is decreasing, the only avenues for growth are winning share from competitors, acquiring competitors, or vertically integrating.

Third, that last paragraph overstates the industry’s travails, at least in terms of television distribution, because most TV distributors are also internet service providers (ISPs), which means they are getting paid by consumers using the services disrupting MVPDs, including Netflix, Google, Facebook, and the Internet generally.

What was both unsurprising and yet odd about this case was the degree to which it was fought over point number two, with minimal acknowledgement of point number three. That is, it seems clear to me that AT&T made this acquisition with an eye on point number three, yet the government’s case was predicated on point number two; to that end, the government, in my eyes, rightly lost given the case they made. Whether they should have lost a better case is another question entirely.

Why AT&T Bought Time Warner

What is the point of a merger, instead of a contract? This is a question that always looms large in any acquisition, particularly one of this size: AT&T is paying $85 billion for Time Warner, and that’s an awfully steep price to simply hang out with movie stars.

The standard explanation for most mergers is “synergies”, the idea that there are significant cost savings from combining the operations of two companies; the reason this explanation is popular is because saving money is not an issue for antitrust, while the corresponding possibility — charging higher prices by achieving a stronger market position through consolidation — is. Such an explanation, though, is usually applied in the case of a horizontal merger, not a vertical one like AT&T and Time Warner.

To that end, AT&T was remarkably honest in its press release announcing the merger back in 2016:2

“With great content, you can build truly differentiated video services, whether it’s traditional TV, OTT or mobile. Our TV, mobile and broadband distribution and direct customer relationships provide unique insights from which we can offer addressable advertising and better tailor content,” [AT&T CEO Randall] Stephenson said. “It’s an integrated approach and we believe it’s the model that wins over time…

AT&T expects the deal to be accretive in the first year after close on both an adjusted EPS and free cash flow per share basis…Additionally, AT&T expects the deal to improve its dividend coverage and enhance its revenue and earnings growth profile.

Start with the second point: as I noted at the time, it’s not very sexy, but it matters to AT&T, a 34-year member of the Dividend Aristocrats, that is, a company in the S&P 500 that raised its dividend for 25 years straight or more. It’s a core part of AT&T’s valuation, but the company’s free cash flow has been struggling to keep up with its rising dividends. Time Warner will help significantly in this regard, as did the previous acquisition of DirecTV.

It is the first point, though, that is pertinent to this analysis: how exactly might Time Warner allow AT&T to “build truly differentiated video services”?

The Government’s Case

While the AT&T press release noted that those “truly differentiated video services” could be delivered via traditional TV, OTT, or mobile, the government’s case was entirely concerned with traditional TV. The original complaint stated:

Were this merger allowed to proceed, the newly combined firm likely would — just as AT&T/DirecTV has already predicted — use its control of Time Warner’s popular programming as a weapon to harm competition. AT&T/DirecTV would hinder its rivals by forcing them to pay hundreds of millions of dollars more per year for Time Warner’s networks, and it would use its increased power to slow the industry’s transition to new and exciting video distribution models that provide greater choice for consumers. The proposed merger would result in fewer innovative offerings and higher bills for American families.

The idea is that AT&T could leverage its ownership of DirecTV to demand higher prices for Turner networks from other MVPDs, because if the MVPDs refused to pay customers would be driven to switch to DirectTV. The problem is that, as was easily calculable, this makes no economic sense: the amount of money AT&T would lose by blacking out Turner would almost certainly outweigh whatever gains it might accrue. The judge agreed, and that was that.

AT&T’s Real Goals

Remember, though, that AT&T did not limit its options to traditional TV: what is far more compelling are the possibilities Time Warner content presents for OTT and mobile. The question is not what AT&T can do to increase the revenue potential of Time Warner content (which was the government’s focus), but rather what Time Warner content can do to increase the potential of AT&T’s services, particularly mobile.

Forgive the long excerpt, but I covered this angle at length in a Daily Update when the deal was announced:

AT&T’s core wireless business is competing in a saturated market with few growth prospects. Apple’s gift to the wireless industry of customers demanding high-priced data plans has largely run its course, with AT&T perhaps the biggest winner: the company acquired significant market share even as it increased its average revenue per user for nearly a decade, primarily thanks to the iPhone. Now, though, most everyone has a smartphone and, more pertinently, a data plan…

The implication of a saturated market is that growth is increasingly zero sum, which presents both a problem and an opportunity for AT&T. The problem is primarily T-Mobile: fueled by the massive break-up fee paid by AT&T for the aforementioned failed acquisition, T-Mobile has embarked on an all-out assault against the incumbent wireless carriers, and AT&T has felt the pain the most, recording a negative net change in postpaid wireless customers for eight straight quarters. Unable or unwilling to compete with T-Mobile on price, AT&T needs a differentiator, ideally one that will not only forestall losses but actually lead to gains.

At first glance this doesn’t explain the Time Warner acquisition either: per my point above these are two very different companies with two very different strategic views of content. A distributor in a zero-sum competition for subscribers (like AT&T) has a vertical business model: ideally there should be services and content that are exclusive to the distributor, thus securing customers. Time Warner, though, is a content company, which means it has a horizontal business model: content is made once and then monetized across the broadest set of potential customers possible, taking advantage of content’s zero marginal cost. The assumption of this sort of horizontal business model underlay Time Warner’s valuation; to suddenly make Time Warner’s content exclusive to AT&T would be massively value destructive (this is a reality often missed by suggestions that Apple, for example, should acquire content companies to differentiate its hardware).

AT&T, however, may have found a loophole: zero rating. Zero rating is often conflated with net neutrality, but unlike the latter, zero rating does not entail the discriminatory treatment of data; it just means that some data is free (sure, this is a violation of the idea of net neutrality, but this is why I was critical of the narrow focus on discriminatory treatment of data by net neutrality advocates). AT&T is already using zero rating to push DirecTV:

This is almost certainly the plan for Time Warner content as well: sure, it will continue to be available on all distributors, but if you subscribe to AT&T you can watch as much as you want for free; moreover, this offering is one that is strengthened by secular trends towards cord-cutting and mobile-only video consumption. If those trends continue on their current path AT&T will not only strengthen the moat of its wireless service against T-Mobile but maybe even start to steal share.

That this point never came up in the government’s case, and, by extension, the judge’s ruling, is truly astounding.

That noted, it is very fair to wonder why exactly the Department of Justice sued to block this acquisition: President Trump was very outspoken in his opposition to this deal and even more outspoken in his antipathy towards Time Warner-owned CNN. At the same time, Makan Delrahim, the Assistant Attorney General for Antitrust who led the case, didn’t see a problem with the merger before his appointment. That the government’s complaint rested on both the most obvious angle and, from AT&T’s perspective, the least important, suggests a paucity of rigor in the prosecution of this case; it is very reasonable to wonder if the order to oppose the merger came from the top, and that the easiest case was the obvious out.

The Neutrality Solution

Thus we are in the unfortunate scenario where a bad case by the government has led to, at best, a merger that was never examined for its truly anti-competitive elements, and at worst, bad law that will open the door for similar tie-ups. To be sure, it is not at all clear that the government would have won had they focused on zero rating: there is an obvious consumer benefit to the concept — that is why T-Mobile leveraged it to such great effect! — and the burden would have been on the government to show that the harm was greater.

The bigger issue, though, is the degree to which laws surrounding such issues are woefully out-of-date. Last fall I argued that Title II was the wrong framework to enforce net neutrality, even though net neutrality is a concept I absolutely support; I came to that position in part because zero rating was barely covered by the FCC’s action.3

What is clearly needed is new legislation, not an attempt to misapply ancient regulation in a way that is trivially reversible. Moreover, AT&T has a point that online services like Google and Facebook are legitimate competitors, particularly for ad dollars; said regulation should address the entire sector. To that end I would focus on three key principles:

  • First, ISPs should not purposely slow or block data on a discriminatory basis. I am not necessarily opposed to the concept of “fast lanes”, as I believe that offers significant potential for innovative services, although I recognize the arguments against them; it should be non-negotiable, though, that ISPs cannot purposely disfavor certain types of content.
  • Second, and similarly, dominant internet platforms should not be allowed to block any legal content from their services. At the same time, services should have discretion in monetization and algorithms; that anyone should be able to put content on YouTube, for example, does not mean that one has a right to have Google monetize it on their behalf, or surface it to people not looking for it.
  • Third, ISPs should not be allowed to zero-rate their own content, and platforms should not be allowed to prioritize their own content in their algorithms. Granted, this may be a bit extreme; at a minimum there should be strict rules and transparency around transfer pricing and a guarantee that the same rates are allowed to competitive services and content.

The reality of the Internet, as noted by Aggregation Theory, is increased centralization; meanwhile, the impact on the Internet on traditional media is an inexorable drive towards consolidation. Our current laws and antitrust jurisprudence are woefully unprepared to deal with this reality, and a new law guaranteeing neutrality is the best solution.

  1. Whether or not the presumption that vertical mergers are not anti-competitive is a worthwhile, albeit separate, discussion [↩︎]
  2. To be fair, the company also mentioned synergies, but it was hardly the point of the press release. [↩︎]
  3. The FCC said it would take it case-by-case, and did argue in the waning days of the Obama administration that zero rating one’s own services as AT&T is clearly trying to do was a violation, but that was never tested in court and was quickly rolled back [↩︎]

The Scooter Economy

As I understand it, the proper way to open an article about electric scooters is to first state one’s priors, explain the circumstances of how one came to try scooters, and then deliver a verdict. Unfortunately, that means mine is a bit boring: while most employing this format wanted to hate them,1 I was pretty sure scooters would be awesome — and they were!2

For me the circumstances were a trip to San Francisco; I purposely stayed at a hotel relatively far from where most of my meetings were, giving me no choice but to rely on some combination of scooters, e-bikes, and ride-sharing services. The scooters were a clear winner: fast, fun, and convenient — as long as you could find one near you. The city needs five times as many.

So, naturally, San Francisco banned them, at least temporarily: companies will be able to apply for their share of a pool of a mere 1,250 permits; that number may double in six months, but for now the scooter-riding experience will probably be more of a novelty, not something you can rely on. In fact, by the end of my trip, if I were actually in a rush, I knew to use a ride-sharing service.

It’s no surprise that ride-sharing services have higher liquidity: San Francisco is a car-friendly town. The city has a population of 884,363 humans and 496,843 vehicles, mostly in the city’s 275,000 on-street parking spaces. Granted, most of the Uber and Lyft drivers come from outside the city, but there is no congestion tax to deter them.

The result is an urban area stuck on a bizarre local maxima: most households have cars, but rarely use them, particularly in the city, because traffic is bad and parking is — relative to the number of cars — sparse; the alternative is ride-sharing, which incurs the same traffic costs but at least doesn’t require parking. And yet, San Francisco, for now anyways, will only allow about 60 parking spaces-worth of scooters onto the streets.

Everything as a Service

This is hardly the forum to discuss the oft-head-scratching politics of tech’s de facto capital city, and I can certainly see the downside of scooters, particularly the haphazard way with which they are being deployed; in an environment built for cars scooters get in the way.

It’s worth considering, though, just how much sense dockless scooters make: the concept is one of the purest manifestations of what I referred to in 2016 as Everything as a Service:

What happens, though, if we apply the services business model to hardware? Consider an airplane: I fly thousands of miles a year, but while Stratechery is doing well, I certainly don’t own my own plane! Rather, I fly on an airplane that is owned by an airline that is paid for in part through some percentage of my ticket cost. I am, effectively, “renting” a seat on that airplane, and once that flight is gone I own nothing other than new GPS coordinates on my phone.

Now the process of buying an airplane ticket, identifying who I am, etc. is far more cumbersome than simply hopping in my car — there are significant transaction costs — but given that I can’t afford an airplane it’s worth putting up with when I have to travel long distances. What happens, though, when those transaction costs are removed? Well, then you get Uber or its competitors: simply touch a button and a car that would have otherwise been unused will pick you up and take you where you want to go, for a price that is a tiny fraction of what the car cost to buy in the first place. The same model applies to hotels — instead of buying a house in every city you visit, simply rent a room — and Airbnb has taken the concept to a new level by leveraging unused space.

The enabling factor for both Uber and Airbnb applying a services business model to physical goods is your smartphone and the Internet: it enables distribution and transactions costs to be zero, making it infinitely more convenient to simply rent the physical goods you need instead of acquiring them outright.

What is striking about dockless scooters — at least when one is parked outside your door! — is that they make ride-sharing services feel like half-measures: why even wait five minutes, when you can just scan-and-go? Steve Jobs described computers as bicycles of the mind; now that computers are smartphones and connected to the Internet they can conjure up the physical equivalent as well!

Indeed, the only thing that could make the experience better — for riders and for everyone else — would be dedicated lanes, like, for example, the 900 miles worth of parking spaces in San Francisco. To be sure, the city isn’t going to make the conversion overnight, or, given the degree to which San Francisco is in thrall to homeowners, probably ever, but that is particularly a shame in 2018: venture capitalists are willing to fund the entire thing, and I’m not entirely sure why.

Missing Moats

Late last month came word that Sequoia Capital was leading a $150 million funding round for Bird, one of the electric scooter companies, valuing the company at $1 billion; a week later came reports that GV was leading a $250 million investment in rival Lime.

One of the interesting tidbits in Axios’s reporting on the latter was that each Lime scooter is used on average between 8 and 12 times a day; plugging that number into this very useful analysis of scooter-sharing unit costs suggests that the economics of both startups are very strong (certainly the size of the investments — and the quality of the investors — suggests the same).

The key word in that sentence, though, is “both”: what, precisely, might make Bird and Lime, or any of their competitors, unique? Or, to put it in business parlance, where is the moat? This is where the comparison to ride-sharing services is particularly instructive; I explained back in 2014 why there was more of a moat to be had in ride-sharing than most people thought:

  • There is a two-sided network between drivers and riders
  • As one service gains share, its increased utility of drivers will restrict liquidity on the other service, favoring the larger player
  • Riders will, all things being equal, use one service habitually

This leads to winner-take-all dynamics in a particular geographic area; then, when it comes times to launch in new areas, travelers and brand will give the larger service a head start.

To be sure, these interactions are complicated, and not everything is equal (see, for example, the huge amounts of share Lyft took last year thanks to Uber’s self-inflicted crises). It is that complication, though, and the fact it is exponentially more difficult to build a two-sided network (instead of, say, plopping a bunch of scooters on the street), that creates the conditions for a moat: the entire point of a moat is that it is hard to build.

Uber’s Self-Driving Mistake

This is why I have long maintained that the second-biggest mistake3 former Uber CEO Travis Kalanick made was the company’s head-first plunge into self-driving cars. On a surface level, the logic is obvious: Uber’s biggest cost is the driver, which means getting rid of them is an easy route to profitability — or, should someone else deploy self-driving cars first, then Uber could be undercut in price.

The mistake in Kalanick’s thinking is two-fold:

  • First, up-and-until the point that self-driving cars are widely available — that is, not simply invented, but built-and-deployed at scale — Uber’s drivers are its biggest competitive advantage. Kalanick’s public statements on the matter hardly evinced understanding on this point.
  • Second, bringing self-driving cars to market would entail huge amounts of capital investment. For one, this means it would be unlikely that Google, a company that rushes to reassure investors when it loses tens of basis points in margin, would do so by itself, and for another, whatever companies did make such an investment would be highly incentivized to maximize utilization of said investment as soon as possible. That means plugging into the dominant transportation-as-a-service network, which means partnering with Uber.

My contention is that Uber would have been best-served concentrating all of its resources on its driver-centric model, even as it built relationships with everyone in the self-driving space, positioning itself to be the best route to customers for whoever wins the self-driving technology battle.

Uber’s Second Chance

Interestingly, scooters and their closely-related cousin, e-bikes, may give Uber a second chance to get this right. Absent two-sided network effects, the potential moats for, well, self-riding scooters and e-bikes are relatively weak: proprietary technology is likely to provide short-lived advantages at best, and Bird and Lime have plenty of access to capital. Both are experimenting with “charging-sharing”, wherein they pay people to charge the scooters in their homes, but both augment that with their own contractors to both charge vehicles and move them to areas with high demand.

What remains under-appreciated is habit: your typical tech first-adopter may have no problem checking multiple apps to catch a quick ride, but I suspect most riders would prefer to use the same app they already have on their phone. To that end, there is certainly a strong impetus for Bird and Lime to spread to new cities, simply to get that first-app-installed advantage, but this is where Uber has the biggest advantage of all: the millions of people who already have the Uber app.

To that end, I thought Uber’s acquisition of Jump Bikes was a good idea, and scooters should be next (an acquisition of Bird or Lime may already be too pricey, but Jump has a strong technical team that should be able to get an Uber-equivalent out the door soon). The Uber app already handles multiple kinds of rides; it is a small step to handling multiple kinds of transportation — a smaller step than installing yet another app.

More Tech Surplus

More generally, in a world where everything is a service, companies may have to adapt to shallower moats than they may like. If you squint, what I am recommending for Uber looks a bit like a traditional consumer packaged goods (CPG) strategy: control distribution (shelf-space | screen-space) with a few dominant products (e.g. TIDE | UberX) that provide leverage for new offerings (e.g. Swiffer | Jump Bikes). The model isn’t nearly as strong, but there may be other potential lock-ins, particularly in terms of exclusive contracts with cities and universities.

Still, that is hardly the sort of dominance that accrues to digital-only aggregators like Facebook or Google, or even Netflix; the physical world is much harder to monopolize. That everything will be available as a service means a massive increase in efficiency for society broadly — more products will be available to more people for lower overall costs — even as the difficulty in digging moats means most of that efficiency becomes consumer surplus. And, as long as venture capitalists are willing to foot the bill, cities like San Francisco should take advantage.

I wrote a follow-up to this article in this Daily Update.

  1. That article is perhaps more revealing than the author appreciated [↩︎]
  2. Note: this article is going to focus on San Francisco for simplicity’s sake, although the broader points have nothing to do with San Francisco specifically; I am aware that the transportation situation is different in different cities — I do live in a different country, after all, in a city with fantastic public transportation and a plethora of personal transportation options. [↩︎]
  3. The first was not buying Lyft [↩︎]