Thread by @bramcohen

Here's Vitalik's opinions about the state of technology around cryptocurrencies. As you might expect, I think much of this is wrong-headed, and will explain (thread) vitalik.ca/general/2019/1…

For 'Blockchain Scalability' he talks about on-chain scaling with sharding as the only option. Of course it isn't the only option, payment channel networks are much more appealing in many ways and are becoming a real thing.

This is a common take mostly because, to put it bluntly, a lot more people (think they) understand sharding than payment channels, and people like to think they're smart so they advocate for solutions they understand.

Here's the truth about payment channels: They trade off fundamentally weakening the security of the system because you're trusting a subset of all peers for the integrity of each shard in exchange for a very small improvement in scaling. How small? Well...

When you run numbers a factor of three is entirely doable, a factor of ten is on the outside of plausible, and a factor of a hundred is a complete joke. This is a lot of complexity for very little benefit even in the best case.

What Ethereum's proposing for sharding further breaks this by essentially requiring miners to have all the shards, which is... not sharding, it's just further redefining 'full node' to mean less than it did before.

And value won't be able to move across shards atomically, which might be unavoidable given what a mess the semantics of the EVM are, but creates a giant mess of incompatibility in exchange for what is, again, a very small benefit.

Moving on, there's the question of 'timestamping', which... works fine already? The claim of needing precision to within 20 seconds is strange. Bitcoin already can do it based on block height or block timestamp

and for what it's worth in proof of space and time you have the extra option of total VDF operations, which has much better precision in some circumstances.

Moving on, for 'Arbitrary Proof of Computation' there is a lot of very exciting stuff going on. So fast-moving and exciting that I'm not going to dare looking into it seriously for a few more years until everything has settled down. But eventually it will be be amazing.

For 'Code Obfuscation' I don't know why you'd want that. ZK techniques fulfill almost all the use cases in practice. It would be amazing to have a protocol for taking any trusted setup and turning it into untrusted, but that's still well into theoretical pipe dream territory

Although to be fair recent improvements have gotten it from something like 2^(10^4) to 2^(10^3) which is progress

For 'Hash-Based Cryptography' if we assume that this is mostly referring to hash based signatures we have a good understanding of what they do and their limitations and no reason to believe there will be meaningful improvements

ASIC-resistant proof of work is both a pipe dream and a bad idea. Being ASIC-friendly to make the hardware more commodity is a much better idea, because ASIC resistance just creates more centralization around manufacture when it inevitably fails

Just FYI, if you want to be ASIC friendly the best approach is to use SHA-3 iterated 100 times. If you really wanted to go thermonuclear you could use proofs of space and time, if only someone were working on that technology...

'Useful Proof of Work' is even more profoundly busted than ASIC resistant, and the biggest attack vector isn't even mentioned here, which is that if the problems are user-suggested then an attacker can put in problems which they happen to have good solutions to

A reasonable bar to clear whether the proposed problem is better than dick pic coin, which is powered by high quality dick pics as evaluated by a pretrained neural network. Most of the useful proof of work proposals I've seen don't clear this extremely low bar.

Proof of stake continues to be a bad idea. It starts with a fundamentally weakened security model and runs into a whole host of deep technical issues on top of that. Progress has been made, but it's more about making decent BFT than real proof of stake.

Proof of space and time offers real practical improvements on the decentralization and wastefulness of proof of work. Maybe somebody should get on that

Proof of storage is an old canard I wish would die already. It still has the same fundamental issues it did back when Zooko and I were working on Mojo Nation in 2001. The problem is that everyone wants their data archived \

which means that it will have to be around for years, during which time all currently available hardware will probably be dead and recycled and the tech stacks people are using today will probably have been completely replaced and the relative supply and demand of storage \

may have shifted so much that the old stuff is getting deleted to make way for the new.

No amount of storing duplicate copies with more error correcting codes will help with this. In fact redundancy makes them worse by reducing overall capacity and increasing costs

If you have a bunch of humans in a room the chances that all of them will be dead a year from now goes down exponentially as the number of humans goes up. For 500 years from now not so much

For 'Stable-value cryptoassets' stable coins are a thing and their problems have to do with the legal and financial status of the entities doing the backing. Non-backed stablecoins are not and never will be a thing for reasons which are too obvious to be worth explaining

The rest of the points under 'economics' are profoundly wrong-headed. The whole point of Nakamoto consensus is to avoid human governance. Working on governance is taking a step backwards, not forwards

And almost all the proposals being put forth about how to run things are literally out of the dark ages. Here's an idea: How about we have a 'company' with 'equity stakeholders' and a 'business model' which makes 'profits' and have part of its \

ongoing operations include the development and maintenance of open source software. Such obscure and esoteric examples as Redhat, IBM, Microsoft, Google, and Facebook have been known to do this in significant amounts, maybe you've heard of them?