by Thomas Diendorf
Make no mistake, I'm not proposing this simply because Pandemic Legacy is at the #1 spot and I personally think it should be lower than that. That's not the reason. The reason is because I don't believe a game should reach the #1 spot on the same year it has been released. It needs to slowly climb to that spot and earn its way there overtime. It needs to be able to stand the test of time long enough to work it's way there. Because being #1 shouldn't be that easy of a feat.
I'm not sure exactly how BGG initially did it's Geek Rating, what mathematical/statistical formula it has been using up 'till now, but it does seem as if it hasn't taken into account that the number base would grow for board gaming. In other words, it was a formula designed to work with a smaller number, like the numbers the site used to have 1-2 years ago. But the gamer base has grown, and such a formula only taking into account a smaller base doesn't work as well anymore. This is particularly evident given the number of board games released in 2015 and 2016 (as of this posting 15 games from 2015-2016 are in the top 100 according to the BGG Rank system).
The reason why this is bad is because it pushes out too many games that worked their way so high too quickly. It's too fast. Games that are ranked that high shouldn't be moving around so quickly as the result of a new release. It's detrimental to their popularity, especially since the purpose of the BGG Rank is for the purpose of establishing the greatest games of all time. And greatest games of all time that have been around for years shouldn't be dethroned by a new release so fast. A new release may end up being a game that is superior to those already in the top 100, or even the top 500, but that is something that should be made more clear overtime, over the course of a year or two, not over the course of a few months.
Thus a new system should be put in place, something that can adapt to the growing numbers. In addition, since I believe the ranking system should be time based (ie impossible to shoot up to the #1 spot within the first several months of release, minimum; within first year or two of release, maximum).
I'm not exactly a math expert, but I'm going to throw out a suggestion anyway. For Pandemic Legacy (currently at the #1 spot), I've done some excel input to get these average results for the first 6 months of release (April, May, and June 2016):
April average (440 voters): 8.7047727273 (8.7 rounded)
May average (786 voters): 8.6960559796 (8.7 rounded)
June average (660 voters): 8.6218181818 (8.6 rounded)
July average (715 voters): 8.4714685315 (8.5 rounded)
August average (859 voters): 8.5349243306 (8.5 rounded)
September average (827 voters): 8.5804111245 (8.6 rounded)
The first 6 months of a game this popular are important. Initially, I believe BGG was putting in a bunch of "5 rating" dummy votes to offset the starting popularity of a game, the idea being that it would start average, and from there would eventually grow to something great, mediocre, or terrible overtime due to opinions from new/analytical players over the years. So I suggest that, starting from the first 500 votes or the first 2 months, whichever comes first, put in a number of "5 rating" votes dependent on the number of total votes for that first month or 2.
Going by a monthly basis, I suggest the first month get a number of "5 rating" votes that's about 80% of the total votes. In the case of April, which has 440 votes, that's 352 "5 rating dummy votes. For each consecutive month, make it 60% dummy votes, then 40%, then 20%, then finally no dummy votes.
April average (with 352 dummy votes, total of 792 votes): 5.01049511 (5.0 rounded)
May average (with 471 dummy votes, total of 1257 votes): 5.00783063 (5.0 rounded)
June average (with 264 dummy votes, total of 924 votes): 5.01366724 (5.0 rounded)
July average (with 143 dummy votes, total of 858 votes): 5.02410742 (5.0 rounded)
August and September unchanged.
Now take the average of the the above with the dummy votes, plus the results of August and September.
(5.01049511 + 5.00783063 + 5.01366724 + 5.02410742 + 8.5349243306 + 8.5804111245) / 6 =
6.19523931 (6.2 rounded)
Assuming it maintains the current average it holds (8.447), here's what it would be at the end of November 2016 under this system:
(6.19523931 + 8.447 + 8.447) / 3 = 7.6964131 (
7.7 rounded)
If one considers this too slow of a growth (which I don't), you could start with 60% dummy votes in April, then 40% in May, 20% in June, and then nothing for the rest; just for comparison, here's the result of what would happen if we did the 60-40-20% instead:
(5.01398027 + 5.02769445 + 5.02723172 + 8.4714685315 + 8.5349243306 + 8.5804111245) / 6 =
6.77595174 (6.8 rounded)
Assuming it maintains the current average it holds (8.447), here's what it would be at the end of November under this system:
(6.77595174 + 8.447 + 8.447) / 3 = 7.88998391 (7.9 rounded)
You could change it up to 10% increments instead, and start at 30, 40, or 50% instead of 80% and go from there, increase or decrease how many months get a certain number of dummy votes, etc, but this is the general idea. In any case, I consider this an improvement, as now the game is following the same growth pattern of all other BGG games in the past with this new formula for the BGG Rank system. An incremental dummy system that takes into account the numbers and the length of time for ranking a game would seem to fit much better than what is currently in place.
What do you think?