How important are network effects in web3?

07/31/2024

Many Internet companies that have durability and market cap also have network effects.

This was probably a little known insight a decade ago, and for those that were particularly astute, their primary thesis for investing or building companies.

Now, network effects, at least the term and the benefits, are widely known. But I'm not sure it's fully understood.

With web3 and blockchains, network effects are bandied about as the key to success.

But there hasn't been further clarity around what network effects look like for the "winners" in web3.

Is the framing right?

One of the powers of network effects was it led to incredible moats for web2 companies (meaning, Internet-first businesses that often leveraged user generated content (UGC).

Network effects primarily benefit the user: as more users join, the value increases for each user.

Two-sided network effects (or cross-side effects) occur when the more of one side, the more value to the other side.

In both cases, however, network effects conveyed value to the platform enabling this dynamics because the users and the data was entirely proprietary to those companies.

Craigslist had all the listings, and so therefore people came; because people came, more listings were added.

Monster had all the job boards, so job seekers came; because that's where job seekers came, more job postings were added to Monster.

eBay had all the auctions, so buyers came; because that's where the buyers were, more auctions and inventory were added.

These marketplace type network effects can be powerful if the demand for the things being supplied is there. People wanted the housing, jobs, and knick-knacks on the respective platforms, so as the supply grew, the demand tended to grow, as well.

However, network effects don't guarantee defensibility.

The "unbundling" of Craigslist was a playbook for venture startups for a reason: by vertically attacking the demand and supply side, often with superior UX, the network effects could be diminished.

Airbnb is a good example, where the founders automated the creation of fake accounts to drive both buyers and listers to Airbnb.

All of the categories of Craigslist could be attacked in a similar way: housing, dating, jobs, used goods.

Despite this, Craigslist is still standing, which is an attribute to the defensibility of network effects.

But...multi-tenancy, or the ability for one side or the other to go to new platforms, doesn't prevent a better run start-up from taking share.

As long as Craigslist continue to provide a decent enough UX for both sides, they are hard to beat.

But Airbnb, for example, through its vertical strategy and superior UX and product set, has likely usurped the position of Craigslist.

For another upstarts to compete against Airbnb and dethrone it, now, would be difficult.

This is what gives Airbnb a winner-take-all position...the end-goal for so many successful web2 start-ups.

But are network effects essential for success?

Here is where I think examining the goals and thinking of a business and market matter.

Network effects aren't the only way to achieve returns (although it is true: if one has a choice between a starting a business with networks versus without, all other things being equal).

Most of the infrastructure companies that have accrued massive value, such as Google Cloud and Amazon, did not have network effects, and yet are successful.

Although Apple did try to make the Apple-to-Apple experience better, those ties were largely weak; they did have a very strong eco-system lock-in effect, and that combined with products that just worked were able to achieve success without dominant network effects.

So what about web3 and blockchain?

Ethereum's network effects

Ethereum had two-sided network effects, which made it an essential platform, because the most successful dapps were being built on Ethereum. This attracted users, which caused transactions. The transactions attracted validators who could earn from transaction fees.

These were largely positive network effects.

But what also happened that negatively impacted users as more users came onto the network.

Congestion and cost.

Each additional user made the experience worse for all the other users.

At the same time, the proof of stake crypto-economic security also had diminishing returns.

At some point, if the amount staked exceeds the threshold to prevent censorship or bribing that is attributable, it doesn't really matter if there's additional economic security.

Having more suppliers in most two-sided marketplaces also can have limited returns, both in web2 and web3. Uber, as an example, has asymptotic returns as there are more drivers. More drivers, alone, doesn't attract more users; perhaps there are marginal improvements at some point in time and we know cost will also asymptotically reach some value, after which, the user doesn't benefit from more suppliers.

Liquidity can be a compelling use case for Ethereum, its L2s, and the related protocols built on top.

The more liquidity in a system, the lower the costs/slippage for the users, which can attract users. And the more users who come to transact, potentially the more fees for those providing liquidity.

Another network effect that we've seen has been those who hold the token when the economics reward more users and more usage by those users.

If the users benefit from the token ownership, they also benefit from more users coming onto the system, because usage should driven token value.

In this regard, Ethereum also has positive network effects.

But a network effect based on token price, which can be subject to exogenous, non-user-based behavior, can send the system into a tailspin. Over financializing a network-effect can also create risks.

But what about UGC-based network effects?

I would argue that many of the purely UGC-based network effects we saw, whether social media posting, content aggregation, and marketplaces (listings by users), were very durable because the data was captive by the platform; even if there were easy multi-tenancy (jobs can be listed in multiple marketplaces, for example), there was durability; but the connections on social media were the most durable source of durability because they lacked portability.

Higher switching costs that could limit multi-tenancy allowed more value to accrue to the platform.

So, in a blockchain world, all data and transactions are public.

Using the marketplace example, if all the job listings, auctions, dating profiles, home listings are on a public ledger, who captures that value? Anyone can come in and stand up an interface.

As long as that data lives on a reputable blockchain with low gas fees, the L2s being the most likely contenders, there is no value capture for an application which performs the read and writes to the blockchain.

The blockchain, on the other hand, does capture value through transaction fees; and if that data is valuable, would want a proliferation of players on both sides of that data market.

In other words, the idea of reading and writing data being a source of network effects accrues to the ledger, not the application platform.

In order to play the game to capture value, the platform must violate web3 principles and keep data siloed.

This means network effects must come from things native and unique to the web3 experience. No data on a blockchain can be the defensible moat.

If that's the case, what can?

Some examples of potentially strong network effects in web3:

Cross-side effects can be very powerful network effects.

As the number of chains continues to grow, the network effect that makes even the long tail chains want to work with a given platform grows. Even the larger chains want to be wherever developers are, and the chains continue to be the primary platform for developers.

So, for example, bridges have network effects: the more chains a bridge is able to bridge, the more likely subsequent bridges want to join; the better the bridge is, the more users are using the bridge.

Given all data is open, software is open source, the network effects must be driven by what benefits users, and really understanding the value chain. The token and incentive/mechanism design, to me, is really where network effects can live.

Are network effects essential?

While the majority of value for Internet companies accrues in a Power Law fashion to those companies with network effects, this makes network effect businesses highly valued if you can build one.

But...is it essential?

In web2, many large businesses that are very successful had limited network effects.

Yes, those generational businesses that successfully leveraged their network effects have sustainable defensiveness.

But it isn't the end all be all.

Network effects doesn't guarantee long-term dominance. It depends on the nature of the network effect.

Network effects also don't guarantee a large TAM.

Network effects can tilt a customer or a business if it contributes to the overall value they extract.

Let's consider L1s.

Why are there still L1's being started and invested into?

Granted, the frenzy has died down.

But if network effects are the winner-take-all mechanism, at a time with Solana and Ethereum winning, why are there new non-EVM's?

The network effects for L1s are tremendous and multi-layered.

All of these have network-type effects.

And yet, other L1's are funded by smart investors and build by smart operators.

Why?