TLDR: 2 rounds of rating - the 1st round would be the same it's now. The output wouldn't be the final rankings however, but tiers (so e.g. top 20, 21-40 etc).
The 2nd round would be similar, but the community would rate just the 1st tier of games (for each category).
This would result in tiers of games with the top 20 of each category (for compo and jam).

Why?

As most of us are aware, the rating system is far from perfect. Would the proposed solution solve all its flaws? Not really, but it would address one of the main sticking points - unrepresentative sample sizes of ratings of winning games.

Rating by its nature is very subjective. We all understand the categories differently and our ratings wildly differ. Your Fun 3* is probably different from my 3* and the same goes for all the other categories. There are people who are happy to give away all-around and there are others who rarely rate something a 5* in a single category. (Keep in mind I'm not pointing at anyone in particular, so avoid posting defensive comments, please). This just might lead to a winner with the minimal number of votes needed. The dev could have got lucky or they published an amazing game, but has gone mostly unnoticed. Either way, it's bound to raise some eyebrows.

How?

Add a second round of rating.
The first could be pretty much the same as it's now, except the results wouldn't be precise rankings, but rather tiers based on the the actual hidden rank. The number per rank is upto debate. I'd start with 10 or 20, but it doesn't need to be fixed necessarily (there could be 100 games in rank 10 let's say etc).
After said 1st round, a second round would take place, but only the top games (Tier 1) would be rated. The ratings could be pretty much the same, except the greatly reduced pool. The devs of those games shouldn't advertise their games at that point, but the landing page should lead all users into the 2nd round. The second round should probably last untill all the games get a given threshold of (round 2) ratings (200?).
After that, the ranking of Tier 1 would be published. (And possibly other tiers as well).

Benefits

• Everyone would actually have a chance to look at the creme de la creme and vote on the winner
• If someone gets into Tier 1 by having an unrepresentative sample, this should somewhat remedy the situation
• If someone places in the top 3, but they should have opted from the winning category, there should still be time to do right by the runners-up
• There would be further motivation to playtest more games (which I do for the winners anyway)
• Quite likely extra feedback for Tier 1

Drawbacks

• The voting would take more time (hey, more to do 'till the next LD... :laughing: )
• You tell me

What do you think, guys? Would this work? Am I completely nuts?
Please, do share your thoughts.