Nostr Web Client

We have the tools with proof of work to mitigate the worst of spam. There is already a NIP to add PoW to notes, but that makes UX for normal use quite degraded. But what if we only had to do it occasionally? Let's say my profile contains a reference to a note on which I included PoW. Clients then could filter out users by which have shown a certain level of PoW, which if set high enough would make these spam attacks unfeasible.

Dikaios1517 6mo ago

That's an interesting idea. It's effectively an attempt to create a distinction between legitimate identities and potential spammers by adding a "cost" to creating an identity, yet without the need for a centralized authority.

However, I don't know how effective it would be. The PoW would have to be high enough that it imposes a significant cost for spammers to have to spin up a new identity once the one they were using is no longer being seen by most users. Yet, that also means it will be a considerable wait for a new user on standard hardware to generate enough PoW to finally be able to post.

Indeed, a spammer could have his GPU working overnight to have multiple Nostr keys ready to go, so that as one is detected as a spammer, he just uses the next one in line, with the PoW note already having been generated.

I think the cost to the new user would be greater than the imposition caused to a determined spammer. Indeed, the more I look at PoW as a possible spam deterrent, the less it makes sense and really just ends up being a deterrent against new users in general. For it to be effective at deterring spam, it needs to be a high enough amount of PoW that it is an annoyance to average users, too. Especially those posting from mobile devices.

Reply to this note

Please Login to reply.

Discussion

plor 6mo ago

I think this is where WoT comes in, as new users generally have some nexus of another user. They could inherit some of the PoW of people who follow them, and the PoW can be some background process that doesn't effect UX too much. The key is making the marginal cost of a new npub higher than spammers are willing to bear. If it costs $0.10 of compute per npub, spammers might make several but will give up when those are easily filtered out.