Idea for coup de coeur algorithm
category: offtopic [glöplog]
okay, coup de cour is like three or four times a year? whats the point of having a coup de cour, it happens so rarely.
What I miss:
1. Have coup de cour each or twice a month. Make the time lapse static not random.
2. Have algorithm that - if it uses a random number generator - make possibilities higher for "not so known production", shown in cdc-bar. Might have a chance of getting more coup de cours or more views (not that I care too much about that).
Roughly estimated facts:
100 coup de cours has allready been taken (the past 14 years).
There's about 1000 productions that has atleast one coup de cour (ruling out number of coup de cours per prod).
This roughly estimates: it will take 100/1000*14 = 140 years until an untaken cdc-production will show up in dialog-bar on the upper left -> if the random numbergenerator used don't take same production more than one time.
There are two unknown's here, because of ruling out number of cdc's per prod and number of cdc's per year.
100/(1000+x)*(14+y) = 140+? years will not be right formula but an idea.
An exact formula can be calculated.
You get the idea. 140 years+ is a long time.
What I miss:
1. Have coup de cour each or twice a month. Make the time lapse static not random.
2. Have algorithm that - if it uses a random number generator - make possibilities higher for "not so known production", shown in cdc-bar. Might have a chance of getting more coup de cours or more views (not that I care too much about that).
Roughly estimated facts:
100 coup de cours has allready been taken (the past 14 years).
There's about 1000 productions that has atleast one coup de cour (ruling out number of coup de cours per prod).
This roughly estimates: it will take 100/1000*14 = 140 years until an untaken cdc-production will show up in dialog-bar on the upper left -> if the random numbergenerator used don't take same production more than one time.
There are two unknown's here, because of ruling out number of cdc's per prod and number of cdc's per year.
100/(1000+x)*(14+y) = 140+? years will not be right formula but an idea.
An exact formula can be calculated.
You get the idea. 140 years+ is a long time.
Coup the coeurs are sponsored by our dicatator admins, there's no algorithm behind it. And I think it's nice as it is, compared to some random fomula spitting out prods.
it's so "nice" that I've actually adjusted my settings not to show coup de coeur at all, after being irritated with the same product for a months/years at the top position on the site.
sorry, but I'm with rudi on this one. whatever logic/algorithm is to be applied to select cdc, it can't be worst than the current method.
sorry, but I'm with rudi on this one. whatever logic/algorithm is to be applied to select cdc, it can't be worst than the current method.
bonefish: thanks for the tip! i'll turn it off.
Maybe after a new CDC has been there for a while the panel could start cycling through older CDC's as well? Keeps it fresher and allows those who weren't around during earlier CDC's to still notice those prods
There's no point in featuring just random things. The point of the CDC is to feature a relatively unknown production that has some kind of a special quality or significance to it. If you want to just see random productions, just open up a prod page, type a random number and see what comes up.
I tried it and got this. Not exactly a great, significant or interesting producton in any way. I by far prefer the dictators' taste.
I tried it and got this. Not exactly a great, significant or interesting producton in any way. I by far prefer the dictators' taste.
Every 2-3 months or so someone pokes me and says "hey why don't you change the CDC to [prod]" and I do.
i don't have a problem if it is changed by an admin or moderator, i' ve seen demos that didn't know they existed at all... but just to introduce a new idea: why not make it that people vote on which prod they want to see every month there, say every first of the month, with the limitation that a prod that has appeared on the cdc list can not be voted to appear again? and if time and work doesn't permit for sth like this to be coded why not change it every month?
Will you make a demo about it?
Now officially pouetized!
defiance: yes, let's spend time on a feature no one really cares about and that will never be used. Good plan.
If only pouet were open source... imagine the possibilities.
MySQL error. i'm on it ^_^
Yeah, I think each month or so doing a:
SELECT * FROM demos where CDCs>0 AND nrComments<30 AND HasBeenPublicCDC = 0 LIMIT 1
Would be a nice idea...
The only thing is that most people doesn't use their CDCs to highlight those hidden gems, instead most people use them as a super-thumb-up, meaning that it is the popular newly released stuff that almost exclusively has a large number of CDCs.
SELECT * FROM demos where CDCs>0 AND nrComments<30 AND HasBeenPublicCDC = 0 LIMIT 1
Would be a nice idea...
The only thing is that most people doesn't use their CDCs to highlight those hidden gems, instead most people use them as a super-thumb-up, meaning that it is the popular newly released stuff that almost exclusively has a large number of CDCs.
Did mog include CDC's from thumb up buddies in his marvelous http://metapouet.net/?
so basically you want to see obscure shit that only one person really really likes?
Yes, please.
I reckon there are 2 options here:
1. Spend lots of time trying to get a good algorithm to select "undiscovered gems" automatically, diverting gargaj's precious time from more interesting stuff and demo making
2. Suggest a suitable CDC to gargaj a bit more often, and spend all that saved time making demos
1. Spend lots of time trying to get a good algorithm to select "undiscovered gems" automatically, diverting gargaj's precious time from more interesting stuff and demo making
2. Suggest a suitable CDC to gargaj a bit more often, and spend all that saved time making demos
baah I parse them, but I don't use them yet - thought about showing them upfront - but, hmm..
@maali: demos are obscure shits that very very few people like!
but still more than one!
One person is enough.
@maali: sometimes there are prods that i feel are really underrated... I want to show them to the world, but the world might not agree, damn fucking world!
I keep wondering. Why do people still care about these thumbs, cdcs, tops and other rating crap? On any resource on teh internets, even such an underground one, besides normal (thinking) users (and hackers), there's always a bunch of organized zombies that will direct any stats as they want anyway. So why care?
I've always liked the CDC section. It is great to discover hidden gems. I would like it too to be updated more often, but as it is is good enough for me :)