[everfree]

> Ugh. Satoshi wrote about this.

I don't believe he did, and I'm familiar with a lot of his writings. Do you have a source?

> Single Client = Good. multiple client = menace.

Two major bugs took 100% of the Bitcoin network down on two occasions.

In contrast, several major bugs temporarily took out small subsets of Ethereum validators several times, but the network remained operational throughout. Due to EIP-1559, the network was not even degraded at all in terms of performance or throughput.

Apparently you're right that I can't convince you otherwise. One network's major bugs have caused major outages, and another network's major bugs resulted in 100.0% uptime still being maintained, and you still think the one with the major outages has a better strategy to defend against bugs.

[proxynoproxy]

Source: https://bitcointalk.org/index.php?topic=195.msg1611#msg1611

It’s not an apples to apples comparison — I was there in 2013 with the 0.4/0.5 bdb issue happened, it was a split (not downtime), and the community went with 0.4 until 0.5 was patched. The community was much smaller. There was no downtime. There could have been loses on the centralized exchange side for the few hours of ambiguity. maybe there was 1 public report of loss at the time. This is the lesson that ETH people were not around for. More moving parts; more failure cases. More client, more moving parts.

[everfree]

> it was a split (not downtime)

There were several significant double-spends. People were able to create fake transactions and scam each other.

Preventing people from sending fake transactions is Bitcoin's one reason for existence. People made out with a bunch of stolen money.

If everyone being able to send fake Bitcoins around and scam people doesn't count as downtime, I don't know what counts as downtime.

Showing fake bitcoins is actively worse than if the network refused to process transactions at all.

As a side note, due to its slashing system, Ethereum "fails closed" like this and refuses to confirm transactions if the network were majorly disrupted.

> There could have been loses on the centralized exchange side for the few hours of ambiguity. maybe there was 1 public report of loss at the time.

At the time, I remember several reports of relatively large losses (and gains, by the scammers). But the private losses are probably larger and they're just as important. When people (and exchanges) get scammed, they're generally incentivized to stay quiet about it.

> This is the lesson that ETH people were not around for.

The "ETH people" that matter - the protocol researchers and client developers and exchange CEOs - were for the most part all around during the early Bitcoin days.

> More moving parts; more failure cases. More client, more moving parts.

Ten cars have "more moving parts" than one car, and the fleet is much more prone to a failure of one of the parts in one of its cars.

When two parts break at once, it's better to have a fleet of eight working cars than one broken car in need of two repairs.

Thanks for the source, by the way. I think Satoshi was just as wrong there as he was when he thought Bitcoin would become a "peer to peer electronic cash system".

[proxynoproxy]

The fleet analogy doesn’t hold. I used the term “lockstep” for a reason. Fleets of cars are not in lockstep. A degraded fleet is worse than no fleet here, there is no 2 are down, 8 are up here.

A better yet strained analogy is 10 cars with different parts that are not interchangeable, work differently and need different drivers. Maintenance nightmare. Contrast with say, an Airline fleet all of the same type of plane, interchangeable parts and pilots qualified in that model.

Anyway, we should leave it here. This debate will be referenced in n years as other debates of in previous cycles on here. I am getting more concerned about physical threats after the recent incidents. I’m going to burn this identity now.

[everfree]

> A degraded fleet is worse than no fleet here, there is no 2 are down, 8 are up here.

Ethereum runs at 100% throughput with 100% of its security guarantees, all the way up to a 33% outage of its validator set. So it stays running at 100% capacity even in a 3 down, 7 up scenario. This has been the case since EIP-1559 was implemented in 2021, and it has been empirically tested.

That's more favorable than Bitcoin, which degrades in capacity (though not security) during partial outages. A 30% miner outage means a 30% reduction in transaction throughput.

The entire point of blockchains is that they are able to thrive and run robustly in a "2 down, 8 up" scenario. That's the sole property they sacrifice so much to achieve compared to centralized database software.

> A better yet strained analogy is 10 cars with different parts that are not interchangeable, work differently and need different drivers. Maintenance nightmare.

Every client is drop-in interchangeable. You don't need different skills to run each one, any more than you need different driving skills to drive a Honda than to drive a Toyota. The only difference is their internals.

To continue with your analogy, you start reaping the rewards of diversity when one car is revealed to have a manufacturing defect and the whole world needs a specific replacement part all at once. This isn't that uncommon. What happens is that the part goes out of stock.

Car diversity causes no maintenance nightmare, because good mechanics are everywhere and they can fix the 500 most popular car models very cheaply despite their varying internals. With Ethereum client diversity, if one of the ten clients breaks down, the fix arrives within hours, for free over the internet - no mechanic trip needed.

> Contrast with say, an Airline fleet all of the same type of plane, interchangeable parts and pilots qualified in that model.

If an airline bought all Boeing 737 Max, which didn't seem like a bad choice at the time, their entire fleet would have been grounded from March 2019 to November 2020.

Southwest suffered a $435 million loss. Everyone who bet on a single aircraft, lost. Everyone who diversified their fleet pulled ahead.

> I am getting more concerned about physical threats after the recent incidents. I’m going to burn this identity now.

It's never a bad idea to burn an identity. But after these conversations, I'm not convinced of the reason.