Nostr is not decentralized nor censorship-resistant

Peter Todd has been saying this for a long time and all the time I’ve been thinking he is misunderstanding everything, but I guess a more charitable interpretation is that he is right.

Nostr today is indeed centralized.

Yesterday I published two harmless notes with the exact same content at the same time. In two minutes the notes had a noticeable difference in responses:

The top one was published to wss://nostr.wine, wss://nos.lol, wss://pyramid.fiatjaf.com. The second was published to the relay where I generally publish all my notes to, wss://pyramid.fiatjaf.com, and that is announced on my NIP-05 file and on my NIP-65 relay list.

A few minutes later I published that screenshot again in two identical notes to the same sets of relays, asking if people understood the implications. The difference in quantity of responses can still be seen today:

These results are skewed now by the fact that the two notes got rebroadcasted to multiple relays after some time, but the fundamental point remains.

What happened was that a huge lot more of people saw the first note compared to the second, and if Nostr was really censorship-resistant that shouldn’t have happened at all.

Some people implied in the comments, with an air of obviousness, that publishing the note to “more relays” should have predictably resulted in more replies, which, again, shouldn’t be the case if Nostr is really censorship-resistant.

What happens is that most people who engaged with the note are following me, in the sense that they have instructed their clients to fetch my notes on their behalf and present them in the UI, and clients are failing to do that despite me making it clear in multiple ways that my notes are to be found on wss://pyramid.fiatjaf.com.

If we were talking not about me, but about some public figure that was being censored by the State and got banned (or shadowbanned) by the 3 biggest public relays, the sad reality would be that the person would immediately get his reach reduced to ~10% of what they had before. This is not at all unlike what happened to dozens of personalities that were banned from the corporate social media platforms and then moved to other platforms – how many of their original followers switched to these other platforms? Probably some small percentage close to 10%. In that sense Nostr today is similar to what we had before.

Peter Todd is right that if the way Nostr works is that you just subscribe to a small set of relays and expect to get everything from them then it tends to get very centralized very fast, and this is the reality today.

Peter Todd is wrong that Nostr is inherently centralized or that it needs a protocol change to become what it has always purported to be. He is in fact wrong today, because what is written above is not valid for all clients of today, and if we drive in the right direction we can successfully make Peter Todd be more and more wrong as time passes, instead of the contrary.