Want to know which of your followers you might want to follow back?
Rank your followers by their WoT scores and see which high quality pubkeys you might have overlooked!
Discussion
That page links to primal (straycat) and primal wants me to sign something even though I don't want to use primal. That's annoying. I allowed it to use my pubkey from the browser extension but it immediately went on to ask me to sign something which I don't think is legitimate at that point.
Which page links to primal? You don’t mean my site (https://grapevine-brainstorm.vercel.app) … ?
I do.
Ahh I forgot I put that in the footer. I could change that to another site — njump perhaps?
I’ve changed it so it links out to my profile at njump.me rather than primal. Which makes sense, bc from njump.me you can then link out to any one of 7 or 8 apps, primal being one.
Sadly appears Alby's extension has stopped working for me and I can't login and don't know how to fix it. Would love to try this.
Have you tried nos2x? I think it ought to work
nice. i wonder who ive missed.
i think we broke your site lol
everyone is using it at the same time.
We broke it boys 🤝
Oh hold on we're back
Haha I already had to upgrade the processor once last week 😂
Neo4j is a beast, and has a big resource appetite 😂
Yup, but it works. Step 1: do stuff no one has never done before. Step 2: optimization
Absolutely. But it's the optimization step where the real work begins 😅
Well, there are many pieces to the puzzle and problems to solve! We need to know what we want our graph dbs to do before we can optimize them! Is PageRank what we want? I think we need a whole new class of centrality algos. GrapeRank is based on my ideas of how to do that. But the only way to test if I’m right is to implement GrapeRank and demonstrate its utility.
One think I think would be neat and useful personally would be a list of npubs that are widely followed by my follows, but not by me.
That way, you can pick 10 users you recognize on to follow on sign up and then generate a list of a dozen npubs that are the most followed and least muted among them.
Right now on my site you can go to the main WoT table, filter for hops 2 or more to see pubkeys you’re not following, and then order them using several methods. I don’t have a column for precisely what you’ve suggested but it would be easy enough to add - perhaps I will do that.
Although if I may suggest: I think the spirit of the GrapeRank influence score is very closely aligned to what you’re asking for, and probably produces a very similar list (if not exactly the same).
How hard is it to publish on Vercel like that? I've been playing in sandbox. Haven't published yet.
Not too hard. Pretty easy in fact. It’s very simple to push changes to GitHub and have them show up automatically. I’m also using AWS EC2, and it took me a while to figure out how to set up the ci/cd pipeline. Now that I’ve done it, that’s easy. But vercel was easier.
Hmm. I'll have to stumble thru it.
How do you snag the Vercel sub domain?
Vercel takes care of it for you. If you want CappyNate-vercel.app, you got it! (Assuming no one else already has it)
Ok but do they have hosting too? Or no?
The problem is that they will price gouge you to hell
You mean using your domain name and not -vercel.app? I don’t know .. probably. I’ve been trying out different hosting providers as I try out different tools.
Vercel is made by the creators of next.js so if you want to use nextjs, vercel might be a good fit.
For me, neo4j is resource hungry, but I’m not sure precisely how resource hungry my project is gonna get which means I’m gonna need to scale resources up as I go along and find out what’s possible. All part of R&D. Which is why I moved my back end to AWS.
The screenshot looks interesting. On GraheneOS vanadium, the remote signer login didn't reach my remote signer. I used the extension on Firefox and logged in. But now it's not getting past "... checking the Brainstorm Calculation Engine API for user profile with pubkey"
This is a bug I haven’t figured out yet. But if I refresh the page it usually does the trick … 😬
This is really neat! And you're using follow lists to calculate the paths?
Yup. Right now the db has FOLLOW and MUTE relationships. Gonna experiment with more types of relationships, like reports, but slowly so as not to overwhelm the server.
Makes sense haha.
I've noticed that different clients store mutes I different places. Do you pull any mute lists that are commonly used or just specific kinds? For me, a mute carries more weight than a follow.
I use kind 3 for follows and kind 10000 for mutes. Would be simple to cast a wider net though.
I agree: mutes count for more. The default params give a follow a weight of 0.03, as compared to a mute which I give a weight of 0.50. You can go to settings, change those params, recalculate and see the changes.
I picked those defaults after just trying a lot of different params and seeing which seemed to be the most useful in terms of stratification of the results.
impressive!
