Last week, I wrote about Elizabeth Warren’s big plan to break up big tech and why I thought her plan (a) would not work and (b) was based on a fairly shocking number of factual errors. Not everyone agreed (indeed, many people have disagreed). Many of those who disagreed, though, seemed to only do so because they hate the big internet companies, and thus they seemed happy about any attack on them, no matter how pointless. Others attacked me personally, insisting that my detailed explanation of why I found Warren’s plan laughably naive was really just because I “love big tech.” Finally, some demanded to know what my plan would be. And while I think it’s somewhat silly to imply that you cannot critique a bad plan if you can’t come up with another plan (sometimes, doing nothing is the best plan), I’ve been meaning to write some more about this anyway, and here’s a good opportunity.
Contrary to the strawman beliefs some insist I have, I am quite worried about the market power of many large companies these days, and how that might be stifling competition. As I’ve argued for over twenty years on this site, the single biggest driver of innovation is competition. And I want to see more competition to get more innovation. My issue is that doing so through regulatory means is fraught with significant risks — ones that could very much do the opposite. Highly regulated industries are not known for being competitive and innovative for the most part. They tend to enable only big entities — who can deal with the regulations — to exist and crowd out startups. On top of that, thanks to regulatory capture and the crony nature of our political system these days, you also end up with just a few big companies who now focus on what we’ve referred to in the past as political innovation rather than technological or entrepreneurial innovation. It’s a recipe for stagnation, not innovation and competition.
My second big concern with the plans people have been floating is that they ignore the reality of why some of the tech companies have gotten so big and so successful. For the most part, they’re in highly networked industries, where it’s not just “winner takes all” but in many ways size and dominance of the network is fundamental to their operation. Network effects can lead to dominant positions, for the fairly obvious reason that the bigger they are, the better they are for everyone involved. For all of Warren’s talk of breaking up companies, note that she was only talking about chipping off a few of their peripheral acquisitions: not taking an axe to their core business.
And that’s because she recognizes that as much as people scream to “break up big tech,” there’s no reasonable way to do that without making the overall offerings a lot less useful for the public. How do you break up Facebook’s social network? Do you say half the world can’t use it and have to use the BookFace spinoff instead? You could, of course, cleave off Instagram and Whatsapp, but that doesn’t really change Facebook’s overall global dominance. The reason Facebook is so powerful is that it connects the entire globe. There is no place to make a reasonable cut to split that up. Google is powerful because of its search engine. How do you break that up? Do you say for searches on topic X you use Google, but for searches on topic Y you have to use Elgoog? You could cut off Doubleclick from Google, but then you still have a massive search engine and a massive internet ads company. And while I guess you could cut off Amazon’s web services piece from it store, that doesn’t change the main “competition” complaint most people have about Amazon, which is the size of its footprint in e-commerce. But again, it got there not through predatory practices, but because it’s so convenient and easy for most people that they actually get tremendous benefit from it.
But, that presents a dilemma. And while lots of people seem to think there are easy answers to this (just like they think there are easy answers to “content moderation”) there are not. This is a really complex issue, and like nearly all super complex issues, the easy solutions tend to look appealing, while actually making everything a hell of a lot worse.
So I will make a suggestion for how I’d like to “break up” big tech, while admitting that since this is a complex topic with no easy answer, I could be wrong. But so could everyone else. And I’ve been digging through the details on this stuff for many years now, and I do think my plan makes the most sense. Later this year, I have a big academic paper on this topic coming out with a lot more details, so in the meantime expect a bunch more posts on this topic leading up to that.
The idea goes back to one I raised back in 2015 in the context of content moderation: that we need to move to a world of protocols, not platforms. This is the world of the earlier internet, dominated by open protocols with a variety of competitive apps built on top. Instead of Twitter, there was IRC. Instead of Reddit, there was Usenet. And you had a choice of clients and servers and could move around if you didn’t like the policies of one or the other.
In the world of protocols, you still get the global connectivity benefit, but without the lockdown control and silos (and, potentially, the questionable privacy practices). In a world of protocols, there may be a global network, but you get competition at every other level. You can have competitive servers, competitive apps and user interfaces, competitive filters, competitive business models, and competitive forms of data management. If you don’t like how one app provider handles privacy, you move to another — but because you’re using the same protocol, you don’t lose everything you’re doing with it, you’re just entering through a new door that you like better. If you don’t like the way one provider handles content moderation, you change it or move to another.
And, yes, I noted competition at the business model level as well — because that’s important. We could see lots of interesting attempts at creating different services with different business models that go beyond the limited options (pay with your data, freemium, advertising, etc.) today. One option might be in the form of cryptocurrency or token tied to the protocol. While I can already hear half of you rolling your eyes, this is a model that is at least worth exploring. A cryptocurreny or token tied to a protocol takes away much of the incentive for the really terrible business models everyone complains about. You don’t need to spy on everyone if just getting more usage in general increases the value of the currency. And encouraging business models that don’t require collecting data on everyone is something we should celebrate, not mock. But, cryptocurrency isn’t the only such solution either. I’ve been playing around with a few attempts at new protocol-based systems these days that purposely eschew the cryptocurrency/token model, and are exploring other models instead. The point is that there are other ways of making this work, and more options is better.
However, if we were in a world where the major services and functions we used online were protocols instead of platforms, it would move the power and control out to the ends of the network, rather than centralizing it on the servers of a few giant companies. We’d still get the benefits of the network effects of the systems, but without the centralized control. We’d still be able to get innovation at various levels, but without relying on a single entity to determine what’s best. We’d still get the convenience of powerful services, but without the opaque decision making of a single entity. It’s an approach that could actually work.
That still leaves the question of how do we get there from here. And there are a lot of challenges in that. But I don’t think declaring large platforms as “platform utilities” gets us any closer to that vision — and if anything seems to drive us away from it. I’ll be writing some more posts on how we get towards a world of protocols instead of platforms, and the many hurdles in the way, in the coming weeks and months. However, there are two key approaches to making it happen: either bottom up or top down. And both could work — but both could be difficult.
The bottom up approach is people designing new protocols from scratch and building a new userbase. That presents a huge number of challenges in terms of building up the userbase, but it’s not impossible. New startups pop on the scene all the time, and some of them even succeed. And I already know of at least 6 or 7 attempts at building these kinds of protocols from scratch. And while they’re all fairly small, some are building up at least some traction and are interesting to follow and experiment with. The benefit to this approach is you have no legacy to deal with, making things easier to design and the entire setup more nimble. The cons, obviously, are the lack of a userbase and the basic “empty room” problem: how do you get someone to use a social application when there’s nothing to do there and no one to connect with?
The “top down” approach would be to convince an existing internet giant to move towards such a world. It’s unlikely that any company today would agree to flip the switch entirely and open up their platform into an open protocol. But I would argue that it’s not as far out and unrealistic an idea as many assume. In the last six months, I’ve had in-depth conversations with four large internet companies about this approach, and they were surprisingly more open to at least considering what it would mean than I initially expected. And while I may go into more detail in later posts, I’ll give three quick reasons why the big tech firms may actually decide it makes sense to give up their silos in the long run:
It gets them away from proposals like Warren’s to “break up” their business, which they know would create a giant mess. By opening up their core code for anyone to use and pushing the power out to the ends of the network, the companies are effectively “breaking themselves up.
It takes them off the hook for all of the platform liability that is getting dumped on them these days. There are all sorts of competing and impossible demands on these platforms concerning how they moderate content, with tons of people being upset no matter what decision is made, and various people and (especially) politicians now looking to dump massive liability on the platforms themselves. If they turn their system into a protocol, they actually get rid of much of that headache. There will still be questions about content moderation, but it becomes a very different problem — one to where there can be many different approaches, with competition at the filter level. We didn’t used to blame email providers for spam, but thanks to the open protocols of email, it allowed all sorts of innovative spam filtering services to spring up. Given how much of a headache content moderation and platform liability has become of late — and the fact it’s likely to become an even bigger headache over time — at some point, companies are going to realize an open protocols approach is a solution to that problem.
The possibility of new business models. This is important. Lots of people insist that no giant platform would willingly give up so much control, because it’s entirely antithetical to business models built off of collecting all the data. And that’s true. But, that’s why I’m eager to see how new business models shake out in this space, because it might lessen the “we collect all your data” business model reliance and open up new opportunities that could even be more lucrative (especially if they create opportunities where people start abandoning services that don’t respect their privacy).
There still are plenty of reasons why the big platforms are unlikely to do this in the near future (and I’m pretty sure a clueless Wall St. would punish any company that tried), but as various trends continue to move along, it may become more and more appealing.
Notice that all of that could very well happen without a massive policy intervention from the government. However, there are certainly policy ideas that could help move this along a lot faster. Of course, moving these policy proposals forward would require a more detailed understanding of how technology and technology innovation really works, because the key isn’t enforcing breakups or trying to codify “fairness” guidelines with new rules — it’s taking away some of the existing legal and regulatory tools that make it easy and appealing to build vertically-integrated walled gardens, which aren’t always obvious to people who aren’t immersed in the tech policy space since they are presented as having other purposes:
Get rid of most patents (if we’re talking real antitrust, we’d start by getting rid of these government-granted monopolies). Perhaps you could start with “software” ones, though defining what is a “software” patent is hard enough. Patents hinder competition and would create huge problematic thickets in a world of protocols where we want widespread competition at every other level. Without patent thickets and patent trolls, we’d see a lot more competition, especially in taking different approaches to providing interfaces and filters for the same kinds of content.
Dump the ridiculous CAFC ruling on API copyrights and make it clear that there is not copyright on interfaces, meaning that you could actually have much more interoperability and people building new systems to connect to third party products.
Get rid of Section 1201 of the DMCA and its anti-circumvention rules. Again, this blocks competition and interoperability. Indeed, Section 1201 is a huge barrier to this kind of future, in that it would allow companies to effectively block out competitive third party services through “technological protection measures.” Without 1201, it would make it much easier for almost anyone to create a new competitive interface for an existing protocol-based service.
Make sure that full encryption is supported as a fundamental right of end users. This becomes important, as a protocol-based system should also lead to a booming new market for databanks that will store your encrypted data, giving you total control over it and who it is shared with (and for what purposes). But if there are mandated backdoors or some other crap, you completely destroy such a market. Encryption makes this possible.
Get the SEC to stop thinking of cryptocurrency as just a financial tool and to recognize it’s something entirely different. Sure, you need them to deal with the out-and-out scams, but as it stands right now, the SEC is talking about crazy ideas like how any crypto system should need to go through “clinical trials” like drugs. Ideas like that would simply snuff out any possibility of the above ever happening. People don’t realize how much of this future may be held back by a myopic SEC who is viewing all of this through a single lens.
There’s more, but those steps would open up a lot of opportunity for a protocols-based system to take hold. And when it really happens, no one’s going to worry about “breaking up” big tech anymore, because it won’t even be necessary.
Read more: techdirt.com