be nice to do a 'paired comparison' ranking system and see how they compare. In that method you are presented with two games, and you pick which is "better". Obviously you need to have some level of knowledge of a game to make a decision, but that is true of the current system. This would at least for an individual get a clear ranking of the games they are qualified to rank. That can be combined with those of others to get a ranking that eliminates some of the variation in what I think a '6' (or 10) is and what someone else thinks it is. Ultimately this also gives an individuals ranked list greater accuracy then just 'rank your top 10'.
It could be set up that you select X of the top 100 games you are qualified to rank. With some explanation for what qualified means (e.g. you have played it more then X times, you have at minimum, a basic understanding of the rules/objectives of the game, etc.). The system randomly picks two games to present to you to pick which is better. It does this until all combinations are completed and a rank is generated. There are probably only a handful of people who are "qualified" to rank all of the top 100 games.
You would not need to complete your rankings in a single session, just like the current system. You could just pick a game you have played, and have the system then make you rank it against each other game you have previously ranked. Betting we would find for lots of folks, that between two games the game they ranked higher, is not the same they ranked higher when doing it on the rating scale.