fiatjaf on Nostr: I have tested Pkarr, I've seen it working, it seems to work great. I have also used ...
I have tested Pkarr, I've seen it working, it seems to work great. I have also used multiple Nostr clients with different outbox model implementations and they always work too.
In other words, if we're talking about the practical world I understand that there are conditions under which the outbox model can fail, I don't deny that. I also think Pkarr can fail. The actual exact numbers and specifications of such failure conditions and their likelihood in the real world are unknown to both of us.
If you think this is false equivalence, then I understand you, but I think I am only saying true things here. And I am not trying to claim outbox model and DHTs are equivalent, I am not trying to claim I know these numbers at all. I just think DHTs will fair better than outbox in some conditions and vice-versa, and they could be complimentary, but this is all irrelevant if I am right and DHTs prove to be unnecessary in practice (even if they would be better if you considered a specific theoretical framework).
Sorry, I don't remember what my point was anymore.
Oh, also, this is doubly irrelevant if we can't use Pkarr for Nostr because of the crypto mismatch. If you are right and we definitely need a DHT then yes we should start making our own DHT immediately, but I am not convinced of that so I'll not start working on the DHT immediately.
In other words, if we're talking about the practical world I understand that there are conditions under which the outbox model can fail, I don't deny that. I also think Pkarr can fail. The actual exact numbers and specifications of such failure conditions and their likelihood in the real world are unknown to both of us.
If you think this is false equivalence, then I understand you, but I think I am only saying true things here. And I am not trying to claim outbox model and DHTs are equivalent, I am not trying to claim I know these numbers at all. I just think DHTs will fair better than outbox in some conditions and vice-versa, and they could be complimentary, but this is all irrelevant if I am right and DHTs prove to be unnecessary in practice (even if they would be better if you considered a specific theoretical framework).
Sorry, I don't remember what my point was anymore.
Oh, also, this is doubly irrelevant if we can't use Pkarr for Nostr because of the crypto mismatch. If you are right and we definitely need a DHT then yes we should start making our own DHT immediately, but I am not convinced of that so I'll not start working on the DHT immediately.