Skip to content

What should the packs front page contain?

add your opinion

If you're experiencing problems with the website (packs pages/forum/wiki) or you would like to suggest something, you can post here.

Moderator: Staff

Post by vampirefrog »

Kaminari wrote:The revised rating system works nice for me, thanks.
Great!
Kaminari wrote:The stats are an interesting addition. I didn't realize I had rated that many vgms already!
Keep on rating. We'll be using the ratings for more complex operations, such as recommended packs, related packs and so on.

Post by Kaminari »

It looks like Dracula X68 doesn't appear in the top rated listing. I don't know the cause, maybe other packs are affected as well.

Post by vampirefrog »

It looks like it's not a clean 5.00, it has a bunch of 5 ratings, then a 4 rating. The problem is that it's so close to 5.00 (it's 4.86) that the stars are displayed as 5 stars instead of 4.5.

I've modified the star display formula to display the floor value instead of the rounded value.

I might change the algorithm to sort by 'number of 5 star ratings' instead of by 'rating and number of ratings'.

Post by vampirefrog »

Fixed latest forum thread links to go to last post instead of first post.

Post by vampirefrog »

I just wanted to write down these ideas before I forget them:

on the front page, the latest 10, the top 10 and the latest top rated packs can be made into an image carousel

we can have collections of packs called "80's arcade music" or "90's arcade music"

looking at google trends, even though video game music and similar terms are trending down, sometimes it spikes up when someone writes a "top 10" or "best of" blog article, so I figured we'd have similar playlists, such as "top 10 80's arcade music" or "top 10 pc engine music" or whatever.

have a blog with regular postings about video game and vgm history, and also research news.

Post by Kaminari »

I'm a bit concerned that a game with seven 5-star ratings suddenly disappears from the top list whenever someone adds a 4-star rating, especially when that list is mostly composed of a bunch of games with only one or two 5-star ratings. I realize it's not a trivial issue — what does make a game with five 5-stars plus one 1-star better than another with just five 5-stars? But surely there must be a more efficient way of evaluating the popularity of a music pack.
  • ctr Offline
  • Posts: 492
  • Joined: 2013-07-17, 23:32:39

Post by ctr »

The problem is that we don't have enough votes on the packs to establish a "stable" rating (ie a single vote cannot influence the rank). The only real solution to this would of course be to have more voters, but since that might not be possible, there's several alternatives that would work better with a smaller voter base.

An alternative could be to replace the average rating with a median rating instead. Just like with the current average rating, sorting by amount of votes secondary would be required. Since most ratings are 5 stars, it would probably make most packs appear as 5 stars and bias the rating towards packs with many ratings.

Another alternative could be to use a different ranking algorithm where positive votes (4-5 stars) increase the rank, negative (1-2 stars) reduce it and neutral does not affect it. Just like a median rating, this would be biased towards packs with many positive votes. I would personally favor this one. Just like the median rating, this would favor packs with many ratings but it would still give votes some effect on the rank.
It would make it a little difficult to display the rank using the stars graphic though. I would suggest just showing the (rounded) average rating there.

Another alternative is to replace the stars with thumbs up/thumbs down option. Youtube did this a few years ago.

A final alternative could be to not only weigh the score, but also amount of views and downloads. This would bias the rank towards older packs and especially packs that has spent some time on the front page (eg Out Run, Dragon Saber, etc)
Post Reply