The poll should read: Do you think these two new stats WOULD increase your enjoyment or interest in the site?

Ok all you people who have been talking about the problems in voting in supak's thread, make yourself heard here. This is an opportunity to support some exceptionally fair and strong stats that can handle the pressures of change over time in the way voting has occured on this site. I've put a ton of time, thought, and development into this.

Ok, I've got two major stats ideas.

Anyone who knows a little about stats will tell you that a ranking percentile would be better than just a simple rank per post because it accounts for change in the amount of entries over time. I would call this Rank Percentile Per Post (RPP) Here's an equation

rank = r
total number of entries in contest = t
percentile = p

p = 100 (1-(r/p))
or if we sub in 2nd place among 80 entries we get:
p = 100 (1-(2/80))
p = 97.5% or 97.5th percentile

This allows the ranking average to adjust for changes in site activity by being based off of the number of entries in a contest.

You could also apply the same premise to voting and it would help to show a more level playing field between the users of the past and present. However, you can't simply do a percentage of votes cast in a contest, you have to adjust it for how many votes were available to be cast at that time. I like to call this stat voting magnetism.
Quite simply, instead of dividing the number of votes you recived in a contest by the total votes cast in a contest, you further normalize the votes cast by dividing the votes cast in the contest by the votes available to be cast by a use at that time.

votes you recived for an entry = v
total votes cast in that contest = t
votes available to be cast by a member at the time of the contest = a
voting magnetism = vm

vm = 100 (v/(t/a))

or if we sub in 30 votes in an 850 vote contest when 25 votes could be used by a member
vm = 100 (30/(850/25))
vm = For that post, you rated a voting magnetism score of 88.24

Using this, we can see a more level measure of how good users were back a year or two ago when contests only had 200 or 300 votes and the winner had 20 or 30 some votes. This also adjusts for when people only had 15 votes to cast or less. Maybe this would help spark some interest in past users who may feel their stats have become obsolete.

Here are some examples of numbers that show how fair this actually is:

These are the vm's for some past and present winners:
anfa's winner in the passenger's view contest scored a vm of 143 (it had 67 votes)
on the flip side, graphix's winning popeye spinach patch entry in the front moving in contest had a vm of 153 (with only 40 votes), which is actually higher, but he had a more dominant entry in that contest so it should be higher. His entry more thoroughly dominated the contest and captured the attention of more voters who voted, thus that is why the stat is called voting magnetism. What this means is, the voting throughout all or at least most of psc history (excluding some of the very very early contests that would cause errors) can finally be placed onto a level playing field with these two stats.

I'm not saying change any way the site is currently run or the way things are voted for now and I'm not saying to get rid of vpp, I'm just suggesting two new, relatively simple, statistical measures (RPP and VM) that will allow for the changes in site activity and voting limits that have occurred over time.

I will be willing to write some semi-pseudo code as best I can for implementing this. It would be great just to test it out and see what happens. I know of some issues that will come up with it regarding very early contests on the site, but we can work on that. Any takers?

I think this would be a better way to rank the all-time leaders as well.

now thats a good idea!

emphatic bump

you got my blessing !! ... but still it would only work when people get less votes to give out ...

I appreciate the support, but this actually works no matter how many votes people have.

I`d say it`s worth a try...gotta be better than listening to all the in-fighting...

Sounds good to me too, but how about if somebody puts in 2 entries in one contest, how would that effect your new system BTVS? Oh...by the way...remind me to never sit in a card game with you! Ya Brainyack!

stephen, you're a genius.
i also agreed with the idea to get rid of "Votes Cast, Past 7 Days"

This issue won't affect the rpp too bad, you just count each post individually, but exclude one post from the total posts in the formula for each one to offset the other post of your own.
In the case of the vm, that is up to the powers to be, you could exclude your own OTHER post's votes from the total votes in the contest in the formula and that would keep you from competing against yourself in that regard. I thought of this issue before in the past, and had it slop my mind, so thanks for the remind. All these things would need to fo into the final formula for calculating the official stat.

And yes, I play a mean hand of poker

I dunno... whenever confronted by a new idea like this, I always ask myself "what would Scionshade do?"... and I think he'd say "screw all you liberal foreigners!". I think that's worth considering before we take this any further.

would i still #3 in the all time leaders

Counts fingers...counts fingers again....says the alphabet out loud....counts fingers....

*scratches head*

.



Billtv,

Shhhhhhhhhhh...

You had me at "vm = 100 (30/(850/25))"

HA!