Thread by @hus_qy

@buonogaston @BarryFried1 @Aleph__Zero AlephZero is most probably very close to the best totally ordered DLT you could possibly ever build and it is able to totally order Transactions at almost real time (so I agree that compared to the named competitors it is superior).

I do however think that the future of DLT ...

@buonogaston @BarryFried1 @Aleph__Zero ... lies in causally ordered DLTs because they allow you to parallelize smart contract execution on multiple cores to the maximum degree (according to admahls law).

If you look at contemporary DLTs then you will see that the bottleneck is not how fast you can order things ...

@buonogaston @BarryFried1 @Aleph__Zero ... but how fast you can actually compute things.

Hedera supports 30 smart contract calls per second. Avalanche supports 50 smart contract calls per second and AlephZero will be in the same ballpark if they want to have EVM compatibility on reasonable hardware.

You can get ...

@buonogaston @BarryFried1 @Aleph__Zero ... some additional performance if you use a compiled smart contract language (like in Solana) that achieves around an order of magnitude more performance but even Solana can only execute less than 500 smart contracts calls per second (the majority of "transactions" in Solana ...

@buonogaston @BarryFried1 @Aleph__Zero ... are consensus messages and actually not real "transactions").

Most DLTs that claim high throughput only mean actual value transfers and not smart contract executions for marketing reasons (even Alephzero claims TPS that are completely unrealistic for smart contract calls).

@buonogaston @BarryFried1 @Aleph__Zero To really improve the performance of contemporary DLTs you need to support multi-threaded execution of non-causally related transactions.

And there is another aspect that is relevant in the context of multi-threaded execution and that is the fact, that if you can parallelize ...

@buonogaston @BarryFried1 @Aleph__Zero ... the execution of independent transactions on multiple cores, then you can also parallelize it on multiple machines, which means that you can reach "infinite scalability" on L1.

Since single threaded execution is not going to get much faster (we have reached the physical ...

@buonogaston @BarryFried1 @Aleph__Zero ... limits of how small transistors can get), the only way to further speed up smart contract execution is to go multi-threaded.

It is possible to build a causally ordered consensus with the very same messaging complexity and finality times as AlephZero and that is what ...

@buonogaston @BarryFried1 @Aleph__Zero ... IOTA is working on (the optimal causally ordered consensus).

I am a big fan of AlephZero because it is very close to the optimal totally ordered consensus and I like when people strive for perfection but I believe that causally ordered consensus mechanism will ultimately ...

@buonogaston @BarryFried1 @Aleph__Zero ... prove to be superior for a L0/L1 settlement layer as they give you more computational throughput and are also inherently more decentralized because you can tap into social consensus to secure the DLT and don't have to limit your validator set to a few dozen - hundred nodes.

@buonogaston @BarryFried1 @Aleph__Zero Nevertheless, there is still a need for totally ordered smart contracts especially in the setting of rollups and I would really like to use AlephZero's tech for L2 totally ordered smart contracts on IOTA at some point.

Currently, there are only two projects working on ...

@buonogaston @BarryFried1 @Aleph__Zero ... causally ordered DLTs:

- IOTA
- SUI

and both claim to ultimately be able to process an infinite amount of transactions per second (which is an almost natural consequence of extending multithreading to execution on multiple machines).

I don't just see SUI as the first ...

@buonogaston @BarryFried1 @Aleph__Zero ... real competitor to IOTA but I also see it as a validation of our line of thought. If you want to build something that is faster than anything that exists today, then you need a different execution model and not a way to totally order things faster.

So TL;DR: AlephZero ...

@buonogaston @BarryFried1 @Aleph__Zero ... is amazing but I "obviously" still believe in the design principles behind IOTA (namely the causally ordered consensus based on CRDT's).