I made a font that is really good and you have to try it out, the font name is Brynda1231.

Eth, the idea is novel, involves a lot of work and time, and has been thought through rather well. I think that first-wave FSors score high due to exposition, a fan base, being known in FS and elsewhere, years of practice and personal/skill development. Looking at them I noticed that some early members must have stopped the DLs quite a while after publishing, clocking high popularity in the past they would now continue to show in the top of your league table; I also thought of the large number of members creating and discussing which at the moment is quite reduced so that scoring points would be very difficult for many newcomers. 

If your idea were in any way published as a general rather than a font-specific, then I think this could easily create an insurmountable gap for students working on a project, for talented font design newbes, for occasionalista FSors who might require some tips or want to refine a good design before feeling happy to share their font. They might work for days/weeks/months and improve but they would hardly see their fonts score high enough for a mention anytime soon. Waiting to arrive even in the top 100 published FSs might take a year; it is possible that such public summary of (lacking) points could turn people off FS and this fabulous expressive and www-accessible type of creativity. However, as an overview of a font this system ight increase individual desire to become more involved in FS and the forum.

I think that your idea involves more than a simple labour of love. I'd admire anybody sorting though thousands of fonts shown in the two 'live' presentations on the site even if access to a large variety of font stats or similar info is provided about all published fonts. Maybe doing something like that in a public access forum rather than for separate fonts every trimester or semester might give new FSers or the latest fonts an equal shot at this sort of peer appraisal and encouragement; might it make finding additional information faster? The scale of what you explained is gigantic , the idea has great possibilities but might require a large analytics team to work as you foresaw even if it is not site-wide.

I had a similar idea for my fonts but ignored creation date. I kept it simpler (??!): DLs -- constructive comments and general observations related to the design (ignoring the general comment count as something negative is not constructive) gave points -- favourites count -- TP count -- glyph count -- I used the date of publishing as a factor of change, but unfortunately I never chart when I reduce DL time to allow myself font modernising through improved skills or for extensive increase of the glyph count so I had no idea if 3 favs were in 1 or in 20 months of visibility. I did substract points for every trimester since font publishing which I noted because I thought it indicated the true popularity of a design based on any interest it had generated recently. I did this analysis every 4 months as I am not all that prolific; in my case I know that DLs, comments etc don't change much in a trimester. I have lost my list, but I remember that my deco-fun designs were more popular than deco-headliner designs and that both those display types combined seemed as popular as text fonts. I still make most of my fonts to share with FS even though the more important and comp fonts, the time consuming designs, and fonts I make for my personal circles or my other hobbies are not cloneable for at least 21 months.

Interesting concept.

Just to point out that the more you get a vote the higher possibility you will get higher balanced rating, as balanced rating system counts the number of voters, CMIIW.

Imagine a font with two voters giving 10, it will get a balanced rating of 8.40. I think it suffers the font a lot since it needs many voters giving superb vote to higher its balanced rating.

My idea is to use average rating instead so that fonts with smaller recognition (in this case fewer voters) don't get "double downgrade" point for having few voters and little balanced rating (because of few voters), which give to fairer results (in my humble opinion)...

It's actually quite easy to run your original algorithm as a database query (on all public fontstructions). The current results are:

FS Ariapenciroman    1411.5246427511452
Hommage à Escher    1308.9987531707059
Badger Spine    1170.6263419220536
Bat Country    1152.438769113141
Dioptical    1129.081753385882
SwiftedStrokes    1026.5949947399904
RMWL Uncialic    868.016958976845
Foldstruct    863.1606605523766
Structurosa    782.4649076678752
en origami - pencil shaded    726.1597334388159

– so indeed a big bias in favour of the old “classics” there.