Re: [Xconq-general] Xconq Ranking at Sourceforge

[Eric McDonald <mcdonald at phy dot cmich dot edu>]

Elijah Meeks wrote:

> Though I'm extremely proud of this, I wonder if it's a
> glitch. 

It could be. Sourceforge has said that they are experiencing some 
problems with the rankings system. However, take Bochs for example; 
although they outclass us in page views and downloads, their tracker 
activity is relatively low. Tracker activity is factored into the rankings.

> I'd like to see the system sf.net uses to
> determine percentile.

To quote Sourceforge docs:

"The current project rankings formula is as follows:

log (3 * # of forum posts for that week) + log (4 * # of tasks ftw) + 
log (3 * # bugs ftw) + log (10 * patches ftw) + log (5 * tracker items 
ftw) + log (# commits to CVS ftw) + log (5 * # file releases ftw) + log 
(.3 * # downloads ftw)"

As you can see, downloads are dropped to 3/10 of their value, but 
tracker items are multiplied by 5 and patches by 10. Since logs 
(presumably base-10) are being taken, an order of magnitude will only 
vary by 1. If Bochs' downloads are 100 times more than ours, their score 
only gains 2 over ours. By contrast, we had 3 patches in the past week, 
and 3*10 = 30, so our score gained 1.x for that. Plus, we had 3 new 
tasks, so our score gained 1.y for that. So, with those two terms alone 
we could hypothetically close the gap.

That said, and as I have mentioned to you in private email, I do not 
suspect that this is sustainable. And, I am more interested in 
developing a good game than in getting good rankings on SF. Things will 
continue to fluctuate, probably wildly. Next week, we could be back to 
hanging out around 400 or 500 again. Big deal. Development continues.

I only mentioned the ranking yesterday because I think it reflects well 
on our project; tracker activity indicates vitality (but not necessarily 
popularity).

Eric