I recently snapped all my BGG game ratings to whole numbers. I can still look at any 2 identically-rated games and say which I like better than the other, but I now believe that grouping them serves a better purpose. Rather than look at my 10's, then my 9.75's, then my 9.5's, etc., I can now look at my 10's, then my 9's, etc. It's easier to examine a single larger group for a game to play or to bring to game nights.
This snapping process vaulted a number of games to the honorable 10 spot. In alphabetical order, they are now:
Age of Steam: Not perfect, but really really good. I love the interplay of actions, the competition for goods, cities, and connections, and the share system. I'm not a fan of having to issue shares only at the start of each turn, of the incongruous complexities of town connections, nor of the random selection and random usefulness of the goods production mechanisms. And yes, even with all that, it's still a 10. I'm also looking forward to the Wallace/Mayfair game--whatever it ends up being called. Getting rid of the score/production boards, and being able to issue shares any time should be nice.
El Grande: Raw area influence with the exceptionally fun twists of the "bidding" cards and the Castillo. Some say it's too chaotic with 5, and not competitive enough with 3. I don't mind either number, but I do think 4 is the sweet spot for this game.
Goa: A great game with 2, 3, and 4. The auction system (players choose what to auction spatially, and pay each other during the auction) is inspired.
Liberté: Subtle and complex game of indirect area influence. You don't know which faction is going to win during any given turn. As the situation become more and more clear, players jockey to be on the winning team or to change the victor. Resolving ties is half the fun. The constant threat of the two alternate victory conditions if the other half.
Maharaja: Palace Building in India: A game of controlled chaos and constantly adjusting plans. Players must choose their two actions secretly and simultaneously, but they don't have to choose exactly how to implement those action until it's their turn. Turn order can change during the round. The order of city visits can change during the game.
Princes of Florence: Every time I play this game, it keeps getting better. The player interaction seems almost non-existent the first few plays, then subtle, then overwhelming. It has a basic auction/action system like Goa, but is otherwise completely different. Every time I think I am doing well, the better players blow me away in the final round.
Ra: One of the best twists on the auction mechanic ever designed. Players choose when to invoke an auction, but can only bid using a small and fixed set of Sun tiles. The Sun tile that wins the bid is put into the mix for the next auction. Won Sun tiles are kept for use in the next epoch, and also factor into the scoring at the end. This is simply the best closed-system auction game.
Taj Mahal: Unlike Maharaja, the sequence of "visits" is fixed. What is "unknown" is the cards, but even those can be mostly tracked if you can remember who drew what. Points are mostly gained through the collection of goods and by spatially connecting palaces on the board. Each round is effectively a simultaneous auction for 6 different rewards.
Vinci: A civilization-style game where players "buy" their developments using victory points. Over the course of the game, your civilizations will decline and new ones will appear. Choosing when to decline, what new developments to "buy", and where to start on the board are all important. Declined civilizations keep scoring points for you until they are overrun.