The Internet is Closed for Business

A use case of Google and YouTube building a private Internet

Because of the app I was building designed to make it easier for users, like myself, who have to keep up with multiple conferences and research talks, I built an audio-based summarizer for YouTube videos.

But YouTube blocks this programmatic behavior. If you go to their app, you are able to choose from my transcript and summarizing extensions.

On the other hand, if building something on top of YouTube, treating it more as a platform, they impose limits. Looking at the YouTube Data v3 API, the content is **primarily** constrained to your own content; which makes sense. They want to automate the process by which content creators add more content.

But if you want to use the content, it’s limited. It’s treated as YouTube’s own content.

In the end, this does make sense. Brian Balfour lays out in great detail a playbook used by all big tech:

1. Discover the moat (YouTube’s is the definitive trove of content)

2. Attract developers (Tools, distribution, generous payouts)

3. Close the Platform (it’s already here - no one else can build on it…if they ever could).

In many ways, YouTube built a great set of applications; and the content itself was the platform/moat. Unlike developer platforms where the platforms are the underlying compute resources and services, YouTube really offers content (and scalable systems to store and service this) and on top an application to record, upload, monetize.

YouTube is able to monetize, disrupting the upfront markets of traditional television and OTP streaming services because they have the audience (the most important part of a platform business: distribution).

But what this means is content views and distribution **must** go through YouTube for it to exist in the real sense of discovery and consumption.

And there is no way for builders to build upon it.

This is why podcasters have switched to YouTube: YouTube has a much better recommendation engine and ultimately, a better form of distribution. It can give away this distribution for free (for now) to content creators because it is capturing the value.

Will content creators eventually need to pay for distribution the way Facebook reduced the virality and discoverability for its own content creators?

But what about web3?

This makes me wonder if the web3 platforms won’t eventually employ this playbook, as well.

The promise of web3 has been the open, composable infrastructure opposing BigTech’s closed, rent-seeking behavior.

But what if the very act of survival requires this?

I’ve shared that platforms (L1s or base-layer DeFi protocols) need products that attract more users. This is akin to what Apple, Microsoft, Facebook have done to gain initial traction.

L1 generosity has been to hand out money. Much of it poorly deployed (there’s probably an analogy to government handing out cash indiscriminately here, but I would need to research this).

The goal is the same: products that succeed, driving distribution and users.

L1’s wouldn’t need to provide more grants once there is take-off velocity.

And the belief is that gas, paying for transactions, would support L1 growth.

But as gas transaction costs go down, and as those applications (including the “L2’s”) build their own distribution, what is the terminal state for these infrastructure platforms?

If they primarily provide commodity infrastructure (compute, storage), they must compete for providing the lowest costs possible, down to the marginal cost of production for the resource. A race to the bottom.

Any kind of taxation or attempt to retain surplus would incentivize those products or L2’s on top to seek greener pastures: do it themselves or find a more generous foundation layer.

The largest counter-argument has been liquidity: that if an L1 or base protocol has the most liquidity, provided by more LPs such that swap fees have the least slippage, then traders would stay. This is a different domain from raw compute and is about the cost of capital.

Resources like compute and storage, which seem tied to physical machines, on the surface should be less liquid; but we see how fungible these resources can be, especially for the right incentives. A resource provider can shift which networks they provide this resource too fairly easily (and the networks are motivated to make this as easy as possible).

Capital has typically been seen as highly fungible; and we also see a similar behavior with strong incentives driving capital holders to go from one protocol to the other.

Both are seeking yield and will behave as such.

But if that’s the case, do the base layers really have the ability to lock in profits or prevent applications or products from competing? Presumably they don’t have the same ability as, say, Facebook or Twitter, centrally shutting off APIs. This behavior has been painted as pure evil, rent-seeking. And it does make sense for these protocols to meet the growth and defensiveness goals against public market expectations.

But don’t those similar dynamics *also* apply to the L1s and base protocols which have been richly rewarded by their own public markets?

Will the governance bodies see this trend and, while still an open environment, behave in the same as a shareholders do in traditional public markets and make decisions to capture more value?

After all, isn’t that what shareholders in both cases still want? Are web3 shareholders immune to the yield-seeking objective function?

What is different, then?

Perhaps the control is more democratic and open, which is a good thing.

But if voting rights become more concentrated, it doesn’t matter if the governance **process** is more open or not. Yield-seeking decisions will still prevail.

It seems like we can’t get our cake and eat it too.

One the one hand, we’re hyper-financializing infrastructure access, meaning more parties can participate in the economic up-side.

On the other hand, economic upside is rational and does drive behavior to grow and capture value, which can strategically put the platforms at odds with products that run counter to it, resulting in the end of the honeymoon.

Apple → distribution → 30% lock in

Google → distribution →closing discovery capturing revenue

Facebook → distribution → pay to play

Why would the incentives be different?