I recently wrote about network effects, which are well described in the book Blitzscalingby Reid Hoffman and Chris Yeh. The ills of their controversial approach – prioritising speed over efficiency in the face of uncertainty – were recently outlined by Tim O’Reilly, who claimed it has gone from “premise to doctrine” in Silicon Valley. The topic of regulating big tech is gaining in public interest and here I outline a few ideas for upholding competition effectively.
To summarise O’Reilly’s objections: network effects result in the winner taking it all, meaning there is a huge incentive for startups to take capricious risks. The winner must not only put their and their investors’ well-founded reservations aside but pay little regard to societal and other ethical concerns.
Andrea Nahles, the General Secretary of Germany‘s Social Democratic Party, proposed a “data for all” policy in the Handelsblatt last August. Justus Haucap, an expert in the field of competition regulation, welcomed the suggestion and stated the challenge in broader terms: you need to open access to the resource that is the cause of monopolistic advantage in the same way transport and telecom networks have been opened up in the past.
Alas, only some network effects are the result of “big data” and I think this is a crucial point, missing from both Nahles’ proposal and Haucap’s evaluation of it. Most network effects are a result of an extremely loyal user base with little incentive to leave but every incentive to stay. Who would favour a WhatsApp clone if no-one else used it, or boycott PayPal if few shops offered an alternative form of payment? Future regulation strategies must take this into account as much as data. Treating access to databases of users like power lines or transmission frequencies will open up a can of worms with data protection written all over it.
But there are things we can do – here are some ideas from an interested non-expert.
0. Do nothing
Yes, this is always an option to be considered.
There was an article in Die Zeit over Christmas outlining Peter Thiel’s philosophy on competition. He says competition is a bad thing because people should be encouraged to covet things others don’t, instead of fighting over limited resources. You could argue, then, that there is no need to encourage competition in the new areas of business that tech startups create. Potential competitors should be encouraged to discover new markets themselves instead of copying what already exists.
While there is some truth to this, the problem with network effects is the type of competition they encourage, not just the competition they suppress. In the beginning, the competitors engage in a scorched-earth battle for dominance – a competition to exclude competition. It’s not necessarily the eccentric free-thinker or innovative first-mover who wins but whoever can raise the most money and execute ruthlessly.
This means moving so fast that authorities can’t uphold regulations, buying up innovative competitors, angering customers, frazzling piles of cash, working inefficiently. These are all features, not bugs, of Blitzscaling and lucrative network effects incentivise them.
But that’s still not to say state regulation is the answer. Even Tim O’Reilly concludes with a catalogue of pleas addressed to the investment community not the state.
After all, companies like Airbnb, Uber, Lyft, Spotify and others have yet to prove their strategy is ultimately a profitable use of investors’ money. Perhaps in ten years it will have proven itself as crazy as it appears. The dotcom boom was enough to make investors wary of tech for tech’s sake and perhaps the perils of pursuing network effects will equally become a lesson from history. It could be the case that the years of high-risk investment are followed by a short period of profitability before users lose interest. This would make Blitzscaling a bad deal for investors.
Nevertheless, let’s do this the European way and get our regulatory rubber gloves on while keeping in mind those two important things: blue-ocean thinking is more innovative and arguably less aggressive than needless competition, and network effects may have a limited shelf life that limits the risk worth taking in pursuit of them.
1. Data for all
This is not my idea, but Andrea Nahles’. It would go some way to addressing those network effects that are a result of access to data. For example, Netflix’ and Amazon’s recommendation engines use big data coupled with machine learning to predict what people will want to watch or buy next. Not having access to that data means you can offer less attractive services, attracting fewer customers, and thereby generating less data to refine your algorithm. Nahles’ proposal is to force a player to give competitors access to a representative sample of anonymised data if their market share goes above a certain size. This presents data-protection challenges and doesn’t address the problem of user loyalty based network effects.
2. Ban acquisitions of competitors
It’s a common gripe that Facebook and Google are becoming overwieldy. But I don’t understand exactly what breaking them up is intended to achieve with regards to network effects. I can only imagine it means breaking up individual services, but also stopping acquisitions of competitors in the way that Facebook acquired Instagram and WhatsApp. The argument for this would be that by acquiring competitors, incumbents can throttle competition. However, I’m not sure that is the case. Arguably Instagram and WhatsApp are “blue ocean” products that go in a different direction from Facebook. Facebook’s strategy appears to be covering all bases in future, so instead of getting overtaken by new innovations, they can channel their existing users into them. If you’re worried about network effects, this is where you need to pay attention, because it enables Facebook to gain long-term value out of its user base, shutting out potential competitors to Instagram and WhatsApp while stacking the odds in favour of their own services. By cloning, for example, SnapChat features in Instagram and Facebook Messenger, Facebook makes innovation less rewarding: Facebook and its acquired or copycat products succeed because of their user-based network effect, not because they are innovative or offer a better service.
Is breaking up Facebook necessary? I don’t think it is the only way to deal with this problem. The real problem is Facebook seamlessly passing on users to new products, whether acquired, copied or even invented itself. So you could ban acquisitions, but large companies could simply build copies of existing competitors and poach their staff, then funnel in existing users to gain an early network effect. This protects it from built-from-scratch competition for users in exactly the same way Microsoft tried to give Internet Explorer preferential treatment by making it the default browser in Windows.
3. Ban user profile sharing
Bearing in mind the above, there is another way. You could say that Facebook, Google and the rest can have as many products as they like, and they can acquire new ones if they want. But they should not be allowed to tip the odds in their favour by using their existing user base to establish a network effect from the beginning. This should include nudging users to connect their existing account to a newly acquired service (such as Instagram) – instead, the competition for users should begin on equal terms. All the users Facebook acquired with Instagram or WhatsApp could, of course, remain on that service but merging them would be a no-no.
This could be a pain in the posterior for users. Imagine a Facebook user having to sign up again and submit all their data and reconnect with all their friends to use Facebook Messenger. I would argue, however, that Messenger was a feature for a long time, and part of Facebook just like groups. But when you start spinning it off into a separate product with its own app, this is where it becomes uncompetitive.
4. A level playing field of user portability
There is a solution for smoothly allowing users to migrate their accounts if platforms think convenience is more important than shutting out the competition.
An exception to idea 3 would be allowed if a service offered an interface where people could port their profile to competitors just as easily, and with just as much nudging and advertising, as the company’s own products. So if Facebook were to give users the opportunity to join WhatsApp with one click (or vice versa), this would only be permissible if a similar opportunity with just as much advertising and nudging were available for all competitors in the product category. They would have to give users the opportunity to migrate to SnapChat (and others, depending on how you define the category) as well as Instragram; Telegram and Threema as well as Messenger.
This way, if a company with an existing network effect tried to take their userbase with them to another product (which can be quite convenient for users who don’t want to sign up several times), then users would immediately be given a choice and the competition would have a big opportunity to compete on equal terms. This would shorten a network effect’s shelf life and make Blitzscaling less likely to pay off.
5. Open standards for mature products/features
I haven’t thought this one through right up to the end, but I think there could be an opportunity around opening up products that have stopped innovating. These products would be standardised in a way that would enable other providers to access users in the same way you have email clients that work to a common standard, but nevertheless innovators such as Gmail who add functionality on top. It would mean I can use Messenger to communicate with people who aren’t on Facebook. I am aware that this is by no means a panacaea – the email standard, because it is in so wide use, hasn’t really been updated for a long time. It certainly doesn’t support native encryption, which was an oversight right at the beginning that hasn’t been satisfactorily rectified even now.
6. A shorter shelf-life for network effects
All in all, I don’t think we are going to be able to take the traditional approach to monopoly aversion, i.e. identifying a resource and making it available to everyone. We can, though, encourage new entrants do truly different things, and accept that they will, for a while, enjoy a user-based network effect. However, reducing the value of these network effects by making them less portable and shortening their shelf-life, is possible without degrading user experience. This would be regulation doing what it is there to do: letting the market do what it does best (which is what we learned from O’Reilly’s approach), while guaranteeing a level playing field for true innovation.
Yes, the winner would still take it all, but only for a limited time. This would devalue network effects, with the pleasant side effect of making so-called Blitzscaling less attractive. It would make the titans a little sluggish and electrify the minnows.