4 replies to this topic

[jgf]

First, being unbelievably lazy myself I certainly don't wish to create more work for someone else.  But perhaps there is a statistician among us who would care to tackle this:  the stats in the 'vote for tracks' area should be weighted by number of votes in order to present accurate data.

For example, sorting by "fun" the top track has a 92% rating with only 26 votes, is that really better than another track with an 87% rating from 227 votes?  And the number 12 track on the list is a beta with only one vote.

Looking at "must have" tracks, the number one track has a 95% rating ...from only three votes!  (You're going to accept the word of three people that this is the ultimate "must have" track for GPL?)

Sorting by "best graphics" we find that only three tracks in the top 12 have more than 10 votes.

[miklkit]

Hmm.  Weighting by number of votes might encourage more people to vote. I don't vote when downloading a track, and keep forgetting to got back and vote.

[sky]

i wanted to vote once and it didn't work. that was when i realized you need to register and login, at which point i couldn't be bothered again. sad, i know.

[ginetto]

The old Track Database had a 25 votes gate where tracks with less than 25 votes were not listed in the statistics.
Maybe it's easy enough doing something like this ?

[jgf]

Well, I started out looking for "new" tracks or updates, then decided to vote on my favorite tracks, then to vote on any tracks I'd had.  In the process I d/l'd about a dozen tracks, including new versions of several I already had, grabbed several dozen updates, reported numerous dead links, voted on over 300 tracks, and commented on many of them.  Took about an hour each evening for five days, but I got many new files and the feeling that this may have made a minor contribution to the site.  It was an interest in what were the most popular tracks in each category that divulged the inaccuracies in the system.