Why everyone is converging on the same path to scaling blockchains

Take a look at what's happening across many of the differing ecosystems. If you do, you'll find one thing in common. They are all converging on the same path toward scaling - in one form or another. Many are working towards rollups. Some are integrating validity proofs, while others are working on fraud proofs. And more effort is getting spent on making light clients secure.

So, how did everyone arrive down the same path?

It is impossible to scale with low hardware requirements

Initially, there was a widely held belief that blockchains could scale with low validator hardware requirements. The most prominent trend in scaling through this approach was execution sharding. The idea behind sharding is that a single blockchain gets split into multiple chains (shards), with each validator only running on a single shard – doing a small subset of the total work. As the blockchain’s capacity grows by adding more shards and validators, the overall hardware requirements per validator would remain low.

But sharding brought with it more problems than solutions. The main issue with sharding is that it reduces security, causing issues like data withholding attacks and the fisherman’s dilemma, among many others. These trade-offs with sharding made it an unviable scaling solution, leading to the search for alternatives.

Philosophically, keeping validator hardware requirements low was considered critical to a decentralized blockchain. If anyone could run a validator on a low-cost machine, like a consumer laptop, the blockchain would be resilient to centralization. This assumption was incorrect for two reasons:

  1. Stake naturally concentrates to a small number of validators. More stake leads to higher rewards, causing top validators to compound their share of stake faster than smaller validators. Naturally, a power law forms where a small minority of validators hold a majority of the stake - seen across virtually every blockchain, including Bitcoin and its hash rate distribution. Now, imagine a blockchain with 100k validators. If it only takes five validators to halt the chain because they control a majority of the stake, decentralization doesn’t really help security.
  2. MEV causes stronger centralization tendencies. A few validators will possess more skills to extract MEV than the rest. This allows them to compound their stake at an even faster rate than they would be able to otherwise – speeding up the centralization process of stake distribution.

Knowing this, there were then two core problems that stood in the way of scaling. Blockchains can’t scale with low validator requirements, and blockchains and decentralized validator sets don’t guarantee all their nice security properties.

Verifiability takes center stage

After realizing this, many went back to the drawing board. Philosophically, what is the essential requirement for blockchains if validator decentralization doesn’t guarantee security? It's verifiability.

If a blockchain can’t stop its validator set from naturally centralizing, users need the means to cheaply verify if those validators follow the rules. It is verification that allows the community to punish malicious validators accordingly. Punishment could be programmatic (automatic slashing) or community-driven (fork & social slash).

Whatever the case, the users verifying the chain keep it secure, even in the face of attacks.

The convergent path

A blockchain can’t scale with low validator hardware requirements, and its users should have cheap access to verify the chain. That only leaves one solution. Blockchains where validators have high hardware requirements but verifying nodes have low hardware requirements.

Low hardware for verifying nodes comes down to making the verification process efficient. On the execution side, nodes need to verify block validity without re-executing every transaction. This can be done with fraud or validity proofs - commonly used by rollups. As for data, nodes need to verify data availability without downloading every transaction – solved with data availability sampling.

Mustafa outlined an idea for a scalable blockchain centered around secure, low-cost verification with the LazyLedger whitepaper, now Celestia, in 2019. Ethereum has been indirectly moving towards this path since the abandonment of execution sharding in favor of rollups and directly moving towards this since switching from data sharding to danksharding. Solana has also started making efforts in the same direction.

Those are just a few examples of many projects converging on the ‘same’ solution to scaling. The only difference between them will become their implementation.