Wonder if it started before that to be honest. It's pretty common knowledge that game reviews have ridiculously inflated scores for example, with anything lower than 70% being seen as bad and people not bothering with anything under 80-90% when choosing what to buy.

So perhaps the average for a review score/rating has just steadily crept upwards over time, til the scale collapsed into a two choice system with 'bad' and 'great' being the only options.

Coincidentally, the inflated subjective scores from game reviews tends to match the percentage of players that rate a game as "recommended" on Steam.

The median rating on Steam is about 80% positive (meaning 80% of players who rated it would recommend it). A game with a 70% Steam rating tends to be mediocre.

It's interesting that even though a rating system where you choose between "recommended" and "not recommended" is much less subjective than trying to rate a game on a scale from 0-100, in aggregate, you get similar scores.

Yes, gaming scores were inflated indeed. This gaming magazine from the 90s called Hoog Spel [1] never gave ratings to games. Instead they gave a summary of plus and minus points. Such a summary takes slightly more time to read (but not a lot) and gives the reader a good idea on the positive and negative sides of the story.

[1] https://en.wikipedia.org/wiki/Hoog_Spel

It was eBay indeed which introduced me to this system. eBay is just a symptom though. The problem is fundamental, that a rating does not tell much. Because what you rate 3/5 someone else rates different. Its the same with movie reviews and game reviews. You'd need to read the review in order to understand it. I do that on e.g. Amazon (and my conclusion is often: this writer is clueless).