About

Why it exists

A store page looks much the same whether a game shipped a patch yesterday or was quietly left alone two years ago. The information that would tell the two apart is public — every update a studio posts to Steam is there, with a date on it — but nobody reads four years of announcements before buying something.

So this reads them instead, for every game on Steam, and says what it found in a sentence. It is not a score out of ten and not an opinion: the method is one measurement, applied the same way to everything.

Who made it

I am Yevhenii Shyshkin, a backend developer living in Stavanger, Norway. I started this because I kept asking the question myself, and the answer turned out to be sitting in public data that nobody had bothered to read.

What it runs on

A Python pipeline reads Steam every day — announcements, prices, review counts, store details — and loads it into PostgreSQL. There are 192 000 games and 650 000 announcements in there now, going back to 2008, about 11 GB. It works asynchronously, in batches, and it backs off when Steam tells it to.

The reading itself is SQL: the gaps between a game’s own updates, the middle gap taken as its usual pace, and the silence since the last update measured against it. It runs in the database rather than in application code.

A FastAPI service hands the results out, and this site reads that service and nothing else — every number on every page comes from one place.

Getting in touch

[email protected] reaches me, and I read it. If a game’s reading looks wrong to you, I would genuinely like to hear about it — the measurement is only as good as what it is told, and the cases it gets wrong are the ones I learn from.