I agree, the ranking algorithms should be based off genuine engagement in a discussion, not opaque votes or likes. I'm working on a new discussion site https://sqwok.im, with the intent to foster open, frictionless, and accessible conversations with anyone, where the relevance is determined by other signals including activity. There are different issues to work around with that model but it more closely resembles how we interact irl (real world "vote" is usually just walking away).

I'm interested. When you mention activity do you mean something like "total active engagement on a post" or "how active is one particular user relative to others". Also is it your intention to maintain fully anonymous participation with or without moderation? Also one thing I think could be a cool feature would be a reddit style "subreddits" but created by users who then collectively decide what it's going to be about by their participation. Good work.

Hey, thank you!

By activity I mean chat activity in a given post. Right now the ranking algo for "hot" is based on how many people are chatting + decay time function. It's somewhat simplistic for now (still mvp), but I have ideas for other signals that could be used to determine how relevant a particular conversation is. It's certainly vulnerable to being gamed, but I think there are ways we could try to limit that, and filter out noise etc.

One of my goals is to focus as much as possible on the conversation and build features around that. For instance instead of a "like" button, it could require a user to _write_ something relevant (just idea). I think the key is that we try new things, hopefully learn from ideas that have been tried elsewhere, and build a place for high-quality realtime conversations.

The site does follow the twitter/IG/twitch model where it's a flat user(nest) hierarchy, but I have considered later enhancing it to allow multiple users to manage a single nest which would allow both community-led or single-entity led. Right now the goal is just to get mvp features and start getting feedback! So thank you :)

Interesting. I wonder if you could factor in “views” for measuring activity, like how many people have visited a thread and for how long. It’s my understanding that most big websites are extremely interested in minutia like where someone spends most of their time on the page, where their cursor is etc. that may be a useful metric for determining engagement.

yes! I think there are a few potentially interesting data points along this thread, such as time spent viewing, conversion from viewer -> active participant, and more. I like measuring the view bounce rate because I envision someone listening to others speak, and then either walk away or step closer to engage further. Cursor position could be interesting, maybe to see if they scrolled and read the content description?