[nailer]

Bugs exist in all tech decentralised or otherwise. Banking systems go down, Visa goes down, Twitter goes down, HN goes down. Sol has never gone down for some serious unfixable design flaw: rather a bug gets discovered and fixed. That’s how scalable systems using new architectures are built.

If one of the less active chains handled Solana’s volume they’d occasionally have had issues too.

Also Solana has been pretty damn reliable of late despite doing more volume than ever before: https://status.solana.com/uptime

[virtualritz]

If a bug in a purportedly decentralized system can bring the entire system down/to a halt, this very fact refutes the claim and shows that the system has a – drumrolls – central point of failure.

[nailer]

I don't think you've worked on distributed systems before. Consensus algorithms do not solve bugs.

Say, being human, and a software author, you didn't anticipate a node with a single ID producing two valid blocks (someone's blue/green setup went active/active).

Yes, one could say that you should have anticipated a single node producing two valid blocks and programmed defensively. I would argue that many people would not have, and even if they did, they would not anticipate other potential bugs. This does not mean the design is centralised: it's just that it's a peer-to-peer design with bugs. Like all software, ever.

You can read cause and effects of the outages, including the one above, here: https://status.solana.com/history

[kerkeslager]

> Consensus algorithms do not solve bugs.

Not yet, but that's definitely an open problem that needs good solutions, and bugs aren't an excuse for allowing centralization.

The "solution" to bugs being used by Ethereum is simply: move very slowly and try very hard to not create bugs in the first place. But others, i.e. Polkadot, are trying to come up with ways of reaching consensus around reverting or patching bugs.

> I would argue that many people would not have, and even if they did, they would not anticipate other potential bugs.

Well, tough shit. This isn't a space with room for errors, certainly not ones this big.

> This does not mean the design is centralised: it's just that it's a peer-to-peer design with bugs. Like all software, ever.

It does mean it's centralized.

If you are skeptical that it's even possible to avoid bugs, well yeah, me too. That's why crypto is a high-risk space: bugs aren't really acceptable, but so far we really haven't found a way to completely avoid them.

[nailer]

> This isn't a space with room for errors, certainly not ones this big.

It is. Comparable systems can and do have errors. You can attach a bunch of nines as an SLA there and Solana would do pretty well in the last months. Especially because it’s only competition in the scalable blockchain space is itself.

> > This does not mean the design is centralised: it's just that it's a peer-to-peer design with bugs. Like all software, ever.

> It does mean it's centralized.

ok.

[kerkeslager]

> It is. Comparable systems can and do have errors. You can attach a bunch of nines as an SLA there and Solana would do pretty well in the last months.

Uptime isn't the issue, security is. If a centralized group of developers can push out a patch and validators will pull it without the time to properly validate it, thus restarting the entire chain, that's a major security problem.

> Especially because it’s only competition in the scalable blockchain space is itself.

Oh give me a break. There are literally hundreds of cryptos in the scalable blockchain space, and Solana isn't close to being the fastest, most scalable, or most stable. You've drunk too much Kool-aid.