People ask me this more than almost anything else: when’s the “for you” feed coming? A recommendation engine that learns what I like and surfaces more of it. It’s a reasonable question — it’s what every major platform does, and it’s genuinely what a lot of users expect a scrolling app to have by default in 2026. I still don’t have one, and it’s not because I haven’t gotten around to it. It’s a deliberate choice, and I want to actually explain it instead of just deflecting the question again.
What Redinfinite Does Instead
You pick a subreddit. You scroll it, chronologically or by whatever sort Reddit itself offers for that subreddit. That’s it. There’s no layer on top deciding what you’d probably want to see more of based on what you lingered on last time. What you get is what’s actually in the subreddit you chose, in the order it actually exists.
The Honest Case for Not Building One
I don’t want to guess what you want. A recommendation algorithm is, at its core, a bet — a model’s best guess at what will keep you engaged, based on signals that are proxies for interest, not interest itself. Sometimes that bet is right and feels like magic. Sometimes it’s wrong in a way that’s hard to notice, because a feed optimized for engagement doesn’t necessarily overlap with a feed optimized for what you’d actually say you wanted if you thought about it for five seconds. I don’t trust myself to build that model well, and I don’t trust the incentive structure it creates even if I did.
Chronological is legible in a way “personalized” isn’t. When you scroll a subreddit on Redinfinite, you know exactly why you’re seeing what you’re seeing: you picked that subreddit, and this is what’s in it. There’s no black box deciding that today you get more of one topic and less of another. That legibility is worth something on its own, independent of whether the personalized version would technically perform better by some engagement metric.
A recommendation engine is an incentive to optimize for time-on-app, not for you. The moment the product’s success metric becomes “did they stay longer,” the algorithm’s job quietly shifts from “show what’s relevant” to “show what’s sticky,” and those aren’t the same thing. I don’t want that pressure built into the product, because I don’t think I’d resist it as cleanly as I’d like to believe.
What It Actually Costs
I’m not going to pretend this is free. No algorithm means no serendipity layer — Redinfinite won’t ever show you a subreddit you didn’t know you wanted, the way a good recommendation system occasionally does. It also means more effort on your end: you have to know what you’re interested in and go find it, instead of the app inferring it for you. For some people, that’s a dealbreaker, and I understand why. A tool that requires you to already know what you want is a worse fit for someone who wants to be surprised.
Who This Is Actually For
Redinfinite is built for people who already know which two or three subreddits they actually care about and just want a clean way to scroll them without an interface arguing with them about what else they should be looking at. If you want discovery to be automatic, this isn’t going to feel like the right tool, and that’s a fair reaction. If you want the app to get out of the way of communities you’ve already chosen, this is exactly the tradeoff you’re looking for.
The Bottom Line
I could build a recommendation engine. It’s not a technical limitation. It’s that the moment Redinfinite decides what you should see instead of showing you what you asked for, it stops being the thing I built it to be. That’s a real cost, paid deliberately, not a missing feature I haven’t shipped yet.
