making PRs

I have actually made a few PRs when I first came to lemmy. I fixed a few bugs that bothered me, implemented a feature I wanted, and took a couple extra bugs from the issue tracker that bothered others. I thought about making my own client, but decided to try patching the existing ones first, and that ended up being easier.

I've stopped being active though since I'm satisfied with the platform as it is. I've considered hosting my own, which might get me to optimize the BE a bit, so I guess we'll see. But I really do think the project is solving the wrong problems, so I prefer to spend my hobby time experimenting with P2P apps since I think that's what we'll ultimately want, but I'll absolutely help if there's any project that's remotely close to what I want.

I don't have anything to show yet since it's rough and I don't want to publish anything without a good moderation story, but hopefully I'll have something later this year.

new search function

I almost did, but the current one (new since I was originally interested) is good enough. Maybe I'll add it to Jerboa or something since it completely lacks post/comment search AFAIK (should be easy now), but searching on my mobile browser works well enough.

But this gets back to the core design decisions. It can't search stuff it doesn't have cached locally, and abusing ActivityPub to broadcast search would have a risk of enabling amplification attacks.

The proper solution to search, imo, is a separate service that indexes as much of the fediverse as possible. That's a massive project, and about on the scale of building a replacement, not to mention hosting costs. I could probably build it as a P2P app though, but at that point I might as well continue with my project since it has other benefits as well (e.g. single namespace, almost no hosting costs outside a few relays, etc).

Voting, as in you vote for something to happen. Democracy. We should federate it so everyone can do it, probably cryptographically or some other way.

Right, but how does that actually work? Every proposal I've seen for distributed voting systems has issues, and federating it won't solve them. Here are a few off the top of my head:

• barrier to entry - how are you going to get Grandma to use it?
• malware voting on your behalf
• privacy - how can you prove dead people aren't voting while also preventing people from knowing if you voted?
• are normal people going to trust it? We have enough issues with people not trusting voting machines, despite no evidence that voting machines have been exploited to any real degree

I'm satisfied with the current system in my area, which is mail ballots with a barcode so voters can see whether their vote was counted. That's good enough, to the point where I'm going to put my efforts toward getting better voting systems (i.e. ranked, approval, or STAR voting) instead of more cryptography.

You‘re making an alternative to the fediverse because of issues you cant be bothered to solve

I'm making an alternative because the issues I want to solve are fundamental to federation, namely:

• confusing namespaces - you need "community@instance", and most people would prefer just "community"
• power hungry mods - the ones I've seen are okay, but they were also okay on Reddit until they weren't; we could vote for mods, but then interested parties could just bot spam their way in
• hosting costs are high - you need to store everything for every community your users are interested in; that's not going to scale well, especially with so much duplication

I can't submit a PR to fix those, because if I try, it'll just be a hack that's going to have repercussions. Those are design decisions we'll just have to live with for now.

So I'm addressing it with a personal research project, and here's briefly how I'm solving them:

• no namespaces, just "topics"
• no permanent moderators, moderation is based on people you explicitly or implicitly trust (everyone would start with some default set)
• hosting costs are $0, unless you run a relay on a $5 VPS; all storage is on user devices (aside from caching nodes to help with availability)

The hardest part is moderation, which is also the biggest selling point, at least until lemmy instance admins can no longer afford to keep hosting.

I think I can make an ActivityPub bridge as well, and I may end up having it act like a lemmy instance to help seed with data. But that's not in the initial goals, I just want to see if client-side moderation based on votes and whatnot can actually work well.