Personal relays (even running on the phone) are the future :)
Stats for wss://chorus.mikedilger.com:444/ running the chorus relay software https://github.com/mikedilger/chorus
Host System: AMD EPYC 2313 16-Core Processor, 3 GHz, 512K Cache
Guest System (VPS): 2-core, 3GB ram, 30GB disk
Uptime of chorus: 8 days
CPU time consumed (9 minutes 55 seconds, or less than 1/1000)
Memory resident (38.424 MB)
Clients being served: 51 connections currently open long-term, no data on short-term.
This 'personal' relay isn't getting much traffic. But it is good to see that it is using less than 1/1000 of the CPU of a cheap VPS. It has also been rock-solid reliable, not just in never crashing but also in serving the events it is supposed to serve.
Discussion
I like the idea of a personal relay, maybe encrypted and hosted which means I can access it anywhere, maybe also backed up locally so I don’t have to trust the hosting service with my data.
This would interface nicely with the Grapevine. There will be a csv file that stores an array of Influence Scores for each user & for each context I care about. A hosted relay could run scripts to update Influence Scores in real time even if I’m offline. Or if I don’t want to trust a hosted relay, I can run scripts locally to update Influence Scores and store them locally in my relay when I’m online.
If my Influence Score csv file gets too big, I delete an ignore Influence Scores in contexts I don’t care about or for any score whose certainty is below some threshold (no sense recording that Bob has a high trust score in some context if my certainty in that score is minuscule bc it’s based on scant data.)