Rollup vs Parachain

Tech Talk
7

I gave a presentation at ETH Denver on the topic “What is Shared Security?”

You can see a recording of that talk here:

And the presentation:

Before describing how Polkadot achieves shared security, I touch on some other “shared security” models:

Different Forms of Shared Security Today

  • Native: Native shared security is implemented at the protocol level, and is represented as a Layer 0 blockchain, working underneath Layer 1 chains.
  • Rollups: Optimistic and zero knowledge rollups use a settlement layer to provide security and finality to their transactions.
  • Re-Staking: Some protocols allow the use of already staked tokens to secure another network, usually through the creation of a derivative token.

but these different forms are not equal…

Then, at the end, I do a quick pros and cons of each of these models:

Re-Staking Solutions

Pros

  • Seems to be protocol agnostic, and can be “backported” to new and existing chains.
  • Smaller / newer chains can rely on more valuable and stable economies.

Cons

  • As tokens are continually re-staked, the economic “costs” needed to attack secured chains decreases.
  • No real computational verification or protection provided by these systems.
  • Seems to ultimately fall back on centralized sources of trust.

Optimistic Rollups

Pros

  • Not limited by the complexity of the on-chain VM.
  • Can be parallelized.
  • They can stuff a lot of data in their STF.
  • They can use compiled code native to modern processors.

Cons

  • Concerns around centralization and censorship of sequencers.
  • Long time to finality due to challenge periods. (could be days)
  • Settlement layers could be attacked, interfering with the optimistic rollup protocols.
  • Suffers from the same problems allocating blockspace as on-chain transactions.
    • On-chain costs to perform the interactive protocol.
    • Congestion of the network.

Zero Knowledge Rollups

Pros

  • Honestly, they are pretty great.
  • Proven trustlessly.
  • Minimal data availability requirements.
  • Instant Finality (at high costs).
  • Exciting future if recursive proofs work out.

Cons

  • Concerns around centralization of sequencers and provers.
  • Challenging to write ZK Circuits.
    • Turing complete, but usually computationally complex.
    • Hard to bound complexity of circuits when building apps.
  • Suffers from the same problems allocating blockspace as on-chain transactions.
    • On-chain costs to submit and execute proofs on settlement layer.
    • Congestion of the network.

Polkadot Native Shared Security

Pros

  • Protocol level handling of sharding, shared security, and interoperability.
  • Easy to develop STF.
    • Anything that compiles to Wasm.
  • Probably the best time to finality, usually under a minute.
  • Data availability provided by the existing validators.
  • Much less concern of centralization from collators vs sequencers and provers.

Cons

  • Wasm is unfortunately still 2x slower than native compilation.
  • Requires lot of data being provided and available in PoV.
  • Certain limitations enforced to keep parachains compatible with the parachains procol.

I wanted to a twitter thread with this same information, but also get feedback from the community before I start sharing it further.

What do you all think of what I wrote? What did I get right? What did I get wrong? What did I miss all together?

cc @pepyakin @rphmeier @burdges

Related: ZK Rollups as Parachains

10 Likes