This post is inspired by the recent Caverna rising from Rank 6 to Rank 4 recently, climbing the ranks and knocking Puerto Rico out of the top 5.
When you hover over the browse tab and hit Games on any BGG page, you'll see a listing of the top 100 Games and then several hundred more pages (754 pages for about 75400 games so far in the database as of Feb 2015). The games are ranked by the Geek Rating, which is an unknown algorithm using the average ranking and some other numbers. The average ranking is, I believe, all the people who have rated it on a scale of 1-10 and averaged by the number of raters.
The general theory behind these numbers is that they differ because the Geek rating seeds its data with 100 votes of 5.5. So when a game is made, its initial geek rating is close to 5.5. This is to prevent a designer voting a 10 on his game, and having that game shot up to #1 immediately.
This is the general idea behind this game rating system.
The BGG System goes:
10 - Outstanding. Always want to play, expect this will never change.
9 - Excellent. Always want to play.
8 - Very good. Like to play, will probably suggest it, will never turn it down.
7 - Good. Usually willing to play.
6 - Fair. Some fun or challenge at least, will play occasionally if in the right mood.
5 - Average. No significant appeal, take it or leave it.
4 - Below average. Slightly boring, could be talked into it on occasion.
3 - Poor. Likely won't play this again although could be convinced.
2 - Very poor. Annoying, I plan to never play this again.
1 - Defies description of a game. You won't catch me dead playing this. Clearly broken.
It doesn't allow scores below 1, like 0.5 or 0.8. Thus, the average score is 5.5, which means its slightly above "average".
Some "flaws" some may see with the system is that it rewards niches, 2nd editions/3rd editions/anniversary editions/etc, and it generally skews more positively toward heavy euro games.
For the niches, I refer to bands of gamers who love a specific game in a small group. These guys will come rate the game very highly, and because its a niche, not as many negative people rate it. In general, this will boost a game up even if another game may have larger appeal to a wider audience.
For extra editions beyond the first, the majority of these have much, much higher ratings than the first edition of the game. The first edition is usually improved or streamlined in some way, so the second probably should be ranked higher. But what about the people who disliked the first edition? I'll tell you: They don't come back for the second edition. So all those negative votes from the first don't carry over, while the second edition does get the first's good votes. This skews it heavily positive. This can be seen in games like War of the Ring 2nd Edition or BattleCon: Devastation of Indines, both of which are sequel games with very high ratings. (BTW if you wanted an example of a niche game, Battlecon can be seen as one since its a fighting game simulator and of course, video game fighter players will likely flock to it more)
Lastly, because of how the Geek algorithm works, it gives higher geek scores to games where the people rating it have rated other games a certain way. I'm not sure on the details, but many have commented how
Game: Avg Rating Geek Rating Num Voters
Agricola 8.14 8.063 37212
Puerto Rico 8.15 8.060 37685
situations like this occur. PR has a higher Average rating and higher number of voters, so it should be above Agricola right? But no, its Geek Rating is below Agricola's. Why? From what I've read, its because the people who rated Agricola have rated other games well in a certain pattern and Agricola's voters get more weight.
The last theory behind this disparity is shill votes. These are fake accounts made by users who vote 1's or 10's on their hated/favorited games to adjust the position. The BGG system automatically removes these. But, by removing these, they also leave room for error as they may remove votes by legitimate users.
Anyway, these are some of the flaws of the system. There may be some others, so feel free to post your thoughts below!
Some advantages of this system is that it holds down games that are cult of the new better than some other rating systems might, and its a generally good indicator of "good" games to try out, at least initially.
For example, Kickstarter games typically get lots of initial high votes as people back it and start rating right away based on its look/hype, sometimes even when they haven't played it. This can lead to some extremely highly rated games. But they aren't highly ranked - their Geek score is low so they don't appear high on the rankings.
Other games like TKG Arena (which randomly got 100 users from Asia logging in and rating it a 10.0 on a certain day) have false ratings and still don't appear so high. I think this is due to shill votes being removed (extreme votes by fake accounts).
General Thoughts and Alternatives
I definitely think this system isn't perfect, but I think most/all would agree that there's no such perfect system. But, it does do a generally good job of ranking and rating games. As long as you don't read too much into the ranks and ratings and use them as general guidelines/starting points for researching certain games, you'll be likely to get a use out of the rankings and ratings.
In general, ratings of 6 or more tend to be worth a look, though looking at games with ratings of 7 or more are probably worth a look first if you haven't played many at first. (BTW quick tip for searching games, do Advanced Search, select 100 or 1000 User Votes or more, search by your standards and time desired, and then you can sort of rank or rating to see different lists geared toward different things)
Reddit's R/Boardgame ranking is decent as well, and polls with lots of voters are good too for a relatively fair representation of the popularity and goodness of games, such as the Dice Tower's People's Choice (obviously, top 100's from reviewers like Tom Vasel are not representative of a good ranking). I'll post the Reddit links below to these two and the BGG one for comparison:
Both of these top 10s, Reddit's and Dice Tower's, have a fair number of voters (on the order of thousands) and are worth a look as an alternative to the BGG Ranking system.
Anyway, final takeaway: Don't read too much into the ratings/rankings - they're accurate to a degree but lots of good games are pushed down more than they may deserve and some not-as-good games are pushed up more than they may deserve. Consider Reddit's Top 100 and the Dice Tower's recent 2014 Top 100 People's Choice as alternatives to look at for popular games. But don't buy on ranking alone! Do your own research and read lots of reviews and comments from the posters about the games. Lastly, do use the Advanced Search features to narrow down games based on your tastes more precisely.
Feel free to post your thoughts on the ranking/rating systems below too.