Every time I use Steam’s discovery queue or any “what to play next” site, I get bombarded with stuff from the last 6 months. I get it - that’s what generates clicks and sales - but it’s genuinely unhelpful for how most of us here actually want to play.

I’ve been quietly working on a tool to change that. The core idea - your taste doesn’t have an expiration date, so recommendations shouldn’t either. Something from 2011 that fits exactly what you’re looking for should surface just as easily as a 2024 release.

It’s early and rough around the edges, but I’m at the point where I want to validate whether this is even a problem worth solving for other people or just a me.

If a recommendation algorithm for games like this existed - smarter discovery that actually respects older games - would you use it?

What features would make it genuinely useful vs just another thing you try once and forget about? I want it to be the tool someone actually recommends to a friend, not just upvotes and forgets.

  • Nytefyre@piefed.socialEnglish
    3·
    6 days ago

    I feel that all and any recommendations or even the wheelhaus thing doesn’t work for me. It doesn’t work for me because I’m going to end up disagreeing with what the choice made for me is going to be. I don’t know, I’d like for there to actually be a more engaging method in how games are decided for me than just giving me this feeling on my shoulders that I HAVE to play something just because some random decider thing told me to.

    It’s a furious cycle of its own.

    • YUART@lemmy.zipOPEnglish
      1·
      6 days ago

      Hi, this is good feedback. I have a similar problem with algorithms like those on YouTube or Netflix because:

      1. I don’t know what the algorithm thinks I like.
      2. I don’t know how to influence it beyond liking or disliking things.

      The same applies to Steam recommendations. I’ve noticed it tries to recommend games based on my wishlist, but the issue is that adding a game to my wishlist doesn’t necessarily mean I like it or want to play it, or want similar recommendations.

      Would you be satisfied with a game recommendation algorithm if you could see what it thinks you like (e.g., your top 5 preferred genres or dislikes) and be able to fine-tune it to produce more personalized recommendations? For example, you could tell the algorithm that you like sci-fi and that it should prioritize sci-fi games.