Oddbean new post about login | logout Decoupling sending an event to a pubkey and filtering events for your pubkey in the same connection/IP is key to reducing the power that the relay has over you.
nostr:note1gvrt0qlvku965ylltye2zs9pz32vvrvfv62kx233742zqp0yh97s3d0nu3 DVM's = Decouplers
Applying the conventional wisdom that event queues are decouplers, this would make relays the decouplers. But decoupling is pointless if you don't have actors, which is what DVMs are. It's a beautiful architecture nostr:nprofile1qqs04xzt6ldm9qhs0ctw0t58kf4z57umjzmjg6jywu0seadwtqqc75sprpmhxue69uhhyetvv9ujucm4wfex2mn59en8j6f0qyfhwumn8ghj7mmxve3ksctfdch8qatz9uq3uamnwvaz7tmwdaehgu3dwp6kytnhv4kxcmmjv3jhytnwv46z79z4kue discovered.
"Conventional" 😂 Thanks for reframing that though. Will need more context to fully start grasping this though. As in, how can 1 relay alone be an IP "decoupler"? And agree, Pablo and Gigi have no idea what they unleashed.
I'm just talking in software engineering terms, if you want to decouple two services, you can put a queue in between. "IP decoupling" is a more abstract concept (and sort of a misnomer I'd argue), but if you wanted to apply the metaphor the role of the queue would be played by the onion routing implemented by DVMs. Anyhow 😂
Got it 😉 Thanks man!
Great questions! I'm not the one with answers, at all. DVM's have one powerful difference in my, non-dev, minde though: they have a price.
Maybe there are reasons to do it nostr-natively (the IP obfuscation is not one). I agree we should use existing solutions, tor is just too slow for people to bother with usually, so it's probably more a question of making this stuff accessible with a good UX. I think a harder problem than IP obfuscation is filter analysis anyway. If your client sends 100 filters in a minute, there's a lot the relay can learn about your pubkey.
Damn you! I was hyped until I read this. Why must you think things through all the time!? 🤣