(reposted from the thread on Whole New Thing)

I would suggest removing the "weighted rating" function on the database.

I was the one who initially suggested this feature and the formula that's used to calculate the ratings is based on the formula that IMDb uses. The reasoning behind it was that the top list should reflect popularity as well as ratings. What I thought it would achieve was that if two movies had an equal score (say 8.5), the one with 15 votes would appear first on the list, and the one with only 5 votes would appear second. I thought it would adjust the ratings only slightly. Like, the one with 15 votes would get a "weighted rating" of 8.51 and the one with only 5 votes would get a "weighted rating" of 8.49, or whatever.

The way it actually works is bizarre and incredibly annoying. This movie has 5 identical votes of 9 out of 10. And yet the "weighted rating" is 8.05. That makes absolutely no sense.

All the weighted ratings seem to do is consistently lower the ratings of well-liked movies. Maybe it only makes sense in a context where films get hundreds or thousands of votes like on IMDb.

It was an experiment, and I think it failed. So I would suggest removing it, if it's not too complicated.

PS: I think the "all by rating" listing function actually uses the raw ratings, not the weighted ratings. Which makes this function ever more unnecessary.

Yeah, I kind of understand the formula. Like I said, I was the one who originally researched this and suggested it. I just didn't realize to what extent it would make a difference. I think probably that formula is meant to be used with a very large sample base, like on IMDb. With our fairly dozen or two voters, it all seems a bit arbitrary.

However, I had forgotten that it was possible to switch from weighted to mean rating on the list, which is actually kind of a neat feature if you want to compare popularity vs. pure rating.

If I was better at math, I might try to adjust the formula so that the "weight" applied on the ratings isn't quite so heavy. But I really have no idea how to do that.

Either way... maybe it's worth keeping after all. I don't feel very strongly about it. I just remember Cal having some objections to the system when we were first trying it out, and since I was the one mostly pushing for it at the time, I thought I would let you guys know that I don't find it as useful as I thought it would be. But if no one really feels that strongly about it, we might as well keep it, I guess.

Here's the formula, by the way. It's the same one that IMDb uses for the top 250:

weighted rating (WR) = (v ÷ (v+m)) × R + (m ÷ (v+m)) × C

where:
  R = average for the movie (mean) = (Rating)
  v = number of votes for the movie = (votes)
  m = minimum votes required to be on the list (at IMDb it's currently 1300 / here, it's 5)
  C = the mean vote across the whole report (currently 6.7 on IMDb / don't know what it is on BA)

Yeah, that does seem like the best solution and application, JK.

Cool. Thanks. I hope no one objects. But I guess if they had an opinion, they would have posted it here.

concerning the minimum number of votes required for the average rating / top list inclusion...

It's currently set at 5. But considering how few people actually vote on films, I wonder if we shouldn't bring it down to 3.

What do you think?