Thank you for the kind feedback! I’ve also been thinking about adding a “Favorites” feature, as well as introducing a new game status like “In Queue” or something similar to indicate that I want to play a game sooner than others.
As for normalization—this is actually a tricky topic, which is why I implemented several ratings with an average score calculation. I often find myself wanting to give a game a perfect score at first glance, but when I start evaluating aspects like "gameplay" or "story," a perfect score never quite fits because some part of the game always falls short. Still, it’s hard to describe the full impression of a game using just dry scores like "gameplay/story/graphics." That’s why I added the "fappability" metric—a game might score poorly in all other areas but still have maximum "fappability" (essentially, a game worth playing once even if it’s not stellar overall).
Currently, I’m working on a server-catalog that will include all games from platforms like f95, Steam, DLsite, Itch.io, and others. To make this possible, I’ll need to rent something capable of handling user load and cover expenses for OpenAI, as I want more precise parsing. If anyone has suggestions for affordable hosting options or is willing to support this project, I’d be very grateful. My goal is to let users of my app easily search for games, add them to collections, visit their respective sites, and so on, without unnecessary hassle or extensions.
The idea of recommendations based on other players’ collections is great, but it will need to be designed carefully to ensure transparency and avoid making anyone uncomfortable. (For example, I personally wouldn’t want my app to send out information about my games.) One possible solution could be to allow access to recommendations only for users who’ve shared their own catalog. Of course, the shared data would be anonymized and limited to just a list of games and ratings. In short, it’s a cool idea, but it’ll require careful thought to make it genuinely useful and comfortable for everyone involved.