GASMs and the Wisdom of Crowds

[Triple Zero]

I just came across this insanely interesting article:

http://www.alistapart.com/articles/the-wisdom-of-community/

It's an overview of a book called "the Wisdom of Crowds" by James Surowiecki (2004). I should really check this book out, I think.

Check out the article and apply it to, for example, the Memebomb database voting tool.

[Telarus]

Good article:

Just because you're collecting votes doesn't mean you have to crown the item with the most votes the winner. You could just as easily look for items that are controversial (high percentage of both good and bad votes) or are undiscovered (low number of total votes).

This especially.

We should re-examine the user page, get rid of the Top100 or whatever, go for a RandomFromTop1-100 that spits 10-20 random ones from the top, and also some from the Controversial and Undiscovered categories. Then have RandomFromTop101-200, 201-300, or larger chunks like 1-500, 501-1000, etc.

Nice find, 000!

[Verbal Mike]

Fucking awesome article. Thanks! And that's a shiny-nice idea you gots there, Telarus!

[Triple Zero]

Thanks for your ideas, Telarus.

I had already been discussing randomization with Cram, I like the idea a lot.

To keep track of "controversial" and "undiscovered" memebombs, I'd need to edit the code a littlebit. Currently it just keeps score with one number and adds +1 or -1. But having read that article I now know this is a waste of wonderful crowd wisdom information Plus it just requires me to add one extra database column, no biggie.

Question, if I were to make that change, I would convert the current scores to positive votes (even if a memebomb might have been controversial?) or perhaps even reset all the scores?

maybe best would be to add two extra columns, one for positive votes and one for negative, both starting at zero. then I could keep the additive score, for a while to keep the ordering. but have its importance in ranking "ebb away" over the course of some weeks as it obtains a proper tally of both pos and neg votes.

[Triple Zero]

Turns out the memebomb tool has been collecting data on individual positive and negative votes, along with IP and timestamp all along!

So suggestions for different ratings are welcome.

I could technically rate votes made in a short time from the same IP differently, but it would be hard and personally I dont like punishing people for actually putting some time into rating stuff.

So I think for now, I just need a sort of ordering based on positive votes and negative votes.

[LMNO]

I still can't vote on my computer with the current layout.

[Cramulus]

Turns out the memebomb tool has been collecting data on individual positive and negative votes, along with IP and timestamp all along!

So suggestions for different ratings are welcome.

I could technically rate votes made in a short time from the same IP differently, but it would be hard and personally I dont like punishing people for actually putting some time into rating stuff.

So I think for now, I just need a sort of ordering based on positive votes and negative votes.

http://www.urbandictionary.com/define.php?term=spag

these guys rank by total votes. This seems to make the entries that are "just good" or "just bad" sink beneath more controversial entries. So entries are basically ranked by celebrity.

[Triple Zero]

I still can't vote on my computer with the current layout.

I know, sorry. But I have been working on that, too.

A more basic, crossbrowser layout will be available soon!

for now, you can use the really really minimal layout, by appending &template=minimal to the URL:

http://www.principiadiscordia.com/memebombs/?action=list&o=rating&m=100&template=minimal

that's at least usable. I will style it to something slightly more pleasing when I get time.

http://www.urbandictionary.com/define.php?term=spag

these guys rank by total votes. This seems to make the entries that are "just good" or "just bad" sink beneath more controversial entries. So entries are basically ranked by celebrity.

wait so you're saying on urbandictionary even if you're voting something down, you're increasing its ranking?

[Cramulus]

oops, had a closer look and no, that's not how it work. 

Here's the scores, for example, for the first two pages of "Bible"

4422 up, 1956 down

2258 up, 1452 down

1561 up, 1311 down

726 up, 540 down

434 up, 305 down

1319 up, 1192 down

271 up, 181 down

176 up, 106 down

272 up, 210 down

341 up, 280 down

416 up, 362 down

183 up, 130 down

280 up, 236 down

836 up, 793 down



I e-mailed them. I'll let you guys know if they reply.

[Cramulus]

their reply:

they're reordered by up votes minus down votes, at the beginning of every week!

isnt that exciting!