Blockchain will solve precisely zero problems the Fediverse has for E2EE.

We don't need it.

I will not use it.

@soatok People actually think this? I can only imagine it being used as a really crappy public key store. Literally any other solution is better.

@soatok Nothing helps security like having every single thing written forever on a slow database that government agencies can and will find way to gain access to over time, just ask the people who followed the chaos that was the crypto space for any length of time.

@soatok

That’s pretty good for most proposed uses of blockchains. Typically, they add more problems than they solve so solve a negative number. Solving exactly zero is a big improvement.

@soatok the error of modern tech (the last decade or so) too many "solutions" looking for problems that almost match.

& E2EE doesn't feel a good fit for fedi - you're writing posts to be public. E2EE feels a better match for systems that are intended to be single user to single user, Limited scope programs that can excel at their main feature.

@patterfloof E2EE is for DMs

@patterfloof Also you're ignoring the entire use case of group chats which also benefit from E2EE

@soatok blockchain extension that commits the private keys of cryptocurrency fans

@david_chisnall

@soatok

@zip @david_chisnall @soatok Serious question: Are there any actual use cases for block chains? The idea sounds interesting, but I have not seen any use outside of crypto, which has betrayed its original idea beyond a use for block chains...

@Elrick_Winter @zip @soatok

When people say 'blockchain' they mean two things, which each have separate valid use cases:

There's a data structure that's basically a flattened Merkel tree, which gives you a tamper-proof append-only data structure. This is used in a bunch of places. Git uses it, for example, as does FreeBSD's auditdistd.

Blockchains also come with a consensus protocol. Consensus protocols are generally useful, but typically things that are called blockchains use mean either proof-of-work or proof-of-stake. Proof of work is stupid if your goal is anything other than rate limiting but other consensus protocols are useful.

When I was at Microsoft they open sourced a thing called the Confidential Consortium Framework (CCF) that was called a blockchain while that was a fashionable buzzword and then quietly not called a blockchain when it wasn't. This ran individual nodes inside a trusted execution environment (TEE) of some form (initially SGX, later generalised). It used the append-only log data structure to maintain a distributed, auditable log of changes to a key-value store and replicated that between nodes.

And this actually is a useful thing because it lets you have a fault-tolerant distributed system where some set of code agreed on by members of the consortium (a group of people who don't necessarily trust each other but who agree on a policy) will process inputs, update a key-value store, respond to permitted queries, and where there's an audit log that any node can independently validate.

Etherium and similar try to solve this problem with 'smart contracts' but, because they come from the 'let's build a currency' space they also do two things that are not that useful:

• Put all of these things on the same public audit log.
• Try to build an incentive system on top of some fake coin thing to motivate participants to (very inefficiently) execute the policy programs.

The latter is needed as a consequence of doing the former. If the nodes are each run by entities who care about the results, then you don't need to incentivise them to execute the programs, you just need the consensus protocol to ensure that they're all doing the same thing and something (in the case of CCF, remote attestation from the TEE, in most other cases by some extrinsic trust relationship that already exists).

A lot of the 'blockchain' work is trying to avoid needing a trusted entity, but in practice there often is one that already exists for other reasons. Trying to remove a trusted entity from the banking system is problematic because (as BitCon owners keep discovering) it's really useful in financial systems to be able to roll back fraudulent transactions. And that means you need some entity who has the right to determine whether a transaction is fraudulent, which is typically the legal system that participants agree has jurisdiction. So things like Etherium end up producing a parallel legal system and hoping that the real legal system agrees with it.

@david_chisnall @Elrick_Winter @zip This is the correct response

@ace It's not a blockchain then