An automatically updated list utilizing an IMDb style Bayes estimator to calculate weighted ratings based on all IFDB ratings. Questions and comments can be placed here.
Formula used for weighted ratings:
(r × v + c × m) ÷ (v + m)
Where:
r = average of ratings for the work
v = number of ratings for the work
c = the mean rating on all of IFDB (3.92)
m = minimum amount of ratings required (13)
But some games don't appear on this list for an unknowable reason. I speculate that Pegbiter has updated values for c and m without updating the description, but it's impossible to know what the new values are, or why.
If we re-implemented the IFDB Top 100, we could guarantee that the data is fresh (Pegbiter currently scrapes IFDB's API). We could also investigate how it works and fix bugs in it.
(BTW, it's not clear how anyone would determine a "good" value for m. The description used to say that m=9, and now it says m=13. Why? Would we compute m as some arbitrary percentile? I speculate that m is currently 19, pretty close to the 7th percentile today. Is the 7th percentile a good choice? Why not the 6th percentile m=21 or the 8th percentile m=18?)
https://ifdb.org/viewlist?id=k7rrytlz3wihmx2o
But some games don't appear on this list for an unknowable reason. I speculate that Pegbiter has updated values for
candmwithout updating the description, but it's impossible to know what the new values are, or why.If we re-implemented the IFDB Top 100, we could guarantee that the data is fresh (Pegbiter currently scrapes IFDB's API). We could also investigate how it works and fix bugs in it.
(BTW, it's not clear how anyone would determine a "good" value for
m. The description used to say thatm=9, and now it saysm=13. Why? Would we computemas some arbitrary percentile? I speculate thatmis currently 19, pretty close to the 7th percentile today. Is the 7th percentile a good choice? Why not the 6th percentilem=21or the 8th percentilem=18?)