DevClad[<a href="https://devclad.com" rel="nofollow">https://devclad.com</a>] is a social-workspace platform for developers to team up on projects and hackathons seamlessly.<p>It functions by matching you 1-on-1 with a compatible developer every week :)
OP: this is not a direct criticism of your idea or implementation (in fact, by the end, you might notice I'm not sure I even have a valid criticism here), just a general observation on a trend I feel is common on this sort of initiative.<p>I don't know about y'all, but I hate this idea of random 1-on-1s. Nothing sends my anxiety levels to the freaking roof as the idea of having to talk alone to a complete stranger.<p>Make it 3+ people, though, and I'm absolutely game for it. This is not about being reclusive/introverted/autistic/antisocial/whatever: I actually like socializing, and meeting new people (well, to a certain extent, but I digress). It's just that, at least for me, 1-on-1 interactions are the most intense ones, even with people you know, let alone with strangers. Group interactions put much less pressure in each person, and it increases the chances of having at least one more extroverted person to lead the conversation.<p>And yet, almost every single instance of this trope of "helping people connect through random pairings" is 1-on-1.<p>Maybe I'm the exception here, or maybe it's easier to implement 1-on-1 pairings, I don't know. But I'd love to see more things like this, but that didn't promote one of my worst nightmares as a feature.
> Every week, you are matched with another dev using an ML algorithm for a 1 on 1 call.<p>I must admit I rolled my eyes when I read "ML algorithm" since it's usually marketing bullshit. Seemingly it's unsupervised knn w/ Scikit Learn, mixing variables like "open to video calls", "location", "openness to ideas" and "timezone".<p>Fairly neat.
I'll just be honest. One on ones suck. We all mostly have to do it year after year with managers and it's a huge uncomfortable waste of time.<p>Having to talk to some random person is even more awkward and a waste of time.<p>Make it useful and group up 3-4 randos at a time. The best conversations happen in small groups.
We all know how this goes from dating websites... just a bunch of non-technical co-founders looking for easy matches or pretending to be developers to keep tabs on their former engineers. Bet there are tons of fake developer accounts too. We are not pieces of meat!
It can be a good prescreening recruitment platform where developers try to find other developers to join their team.<p>With this kind of platform, you need to keep the quality and commitment high all the time, which is quite challenging. For example, I tried LunchClub for a few weeks, it was brilliant initially, but then you get those who canceled last minute or did not show up to the call and then stopped because I saw it as a waste of time.
> you are matched with another dev using an ML algorithm for a 1 on 1 call<p>I was wondering why so complicated and how the model was learned...<p>> <a href="https://github.com/DevClad-Inc/devclad/blob/main/apps/server/social/ml/match.py">https://github.com/DevClad-Inc/devclad/blob/main/apps/server...</a><p>Aha, better indeed to rely on your own human learning ;-)
Seems nice! I would prefer a twist of this more towards finding other's with an expertise you don't have yourself and having the option to center the conversation around a particular topic, and some expectation of sharing advice and expertise (not just chatting).<p>There could be a paid model, or you could imagine a credit system where you also offer your own expertise and time to others in equal measure. That could still lead to finding things to work together on, but could also lead to learning, consulting gigs or just networking - with the added benefit of being immediately useful at work. But that maybe already exists somewhere?
When I try logging in with GitHub:<p><pre><code> This application will be able to read and write all user data. This includes the following:
Private email addresses
Private profile information
Followers
</code></pre>
Seems odd?
As a suggestion you could also try and give users some levers on your matching algo.<p>For example some people might be interested in matching up with someone similar. Some (like me) would like to talk to people doing different things, on different stacks, with different problem domains and find synergies/learning opportunities that way.
I was wondering how you pair devs with similar interests and noticed that it says "using ML" exactly once. Is that the scope of it or are there settings?<p>Seems I misunderstood completely. The site uses ML to do the matching and devs feed the ML how?