| |
Raistlin
Registered: Mar 2007
Posts: 671 |
Event id #3003 : Only Sprites Compo
Competition Rules:
* You can ONLY display sprites - no chars or bitmaps. Magic byte(s) CAN be used.
* Change any VIC properties that you like
* The Effect must be a one-screener in one PRG (preferably compressed) (no disk streaming)
* Variations of the effect are allowed
* Interaction is allowed - make it a game if you like ;-p
* Music is allowed and encouraged
* Borders can of course be removed
Winner will be chosen based on CSDB's rankings. |
|
... 99 posts hidden. Click here to view all posts....
|
| |
Raistlin
Registered: Mar 2007
Posts: 671 |
Not till Tuesday, Hedders, not till Tuesday. Any entries sent before then will automatically go to Out of Compo :-p |
| |
Raistlin
Registered: Mar 2007
Posts: 671 |
Compo starts TOMORROW :-) |
| |
Raistlin
Registered: Mar 2007
Posts: 671 |
"Winner will be chosen based on CSDB's rankings."
.. I'll take a look at scores at midnight (GMT) on 22nd January. Unless everyone disagrees, I'd like to just take a simple average of the score rather than the ranking - so I would total up the score and divide by the number of votes... no weighted algorithm. |
| |
Oswald
Registered: Apr 2002
Posts: 5086 |
Quote: "Winner will be chosen based on CSDB's rankings."
.. I'll take a look at scores at midnight (GMT) on 22nd January. Unless everyone disagrees, I'd like to just take a simple average of the score rather than the ranking - so I would total up the score and divide by the number of votes... no weighted algorithm.
you'll make the downvoters happy with that. |
| |
Raistlin
Registered: Mar 2007
Posts: 671 |
Quote: you'll make the downvoters happy with that.
I haven't seen evidence that CSDb's weighted voting works all that well. Somebody pointed out to me that if there's a prize involved, and the scoring system is "behind closed doors" (a "secret" algorithm), then the result could be contested.
The prize is substantial on this one (1000Euro value)... so the "calculation" of who's won needs to be plain IMO. |
| |
Jammer
Registered: Nov 2002
Posts: 1335 |
I recommend counting squared number of votes :) Really reliable, even if final average seems a little too sweet. No classic weights here. |
| |
Raistlin
Registered: Mar 2007
Posts: 671 |
Quote: I recommend counting squared number of votes :) Really reliable, even if final average seems a little too sweet. No classic weights here.
This sounds like a good idea. So something like:-
(((#10votes)^2 * 10) + ((#9votes)^2 * 9) + ... + ((#1votes)^2 * 1)) / ((#10votes)^2 + ... + (#1votes)^2)
That does seem to work well.
Check:-
Demo 1.. 20 people vote 10, 1 votes 1
Demo 2.. 20 people vote 1, 1 votes 10
Demo 1
Reg: (20 * 10 + 1 * 1) / 21 = 201 / 21 = 9.571
Squ: (20^2 * 10 + 1^2 * 1)/(20^2 + 1^2) = 4001/401 = 9.9775..
Demo 2
Reg: (1 * 10 + 20 * 1) / 21 = 30 / 21 = 1.428
Squ: (1^2 * 10 + 20^2 * 1)/(1^2 + 20^2) = 410/401 = 1.022
Ok, let's go with the squared calculation. Is everyone ok with this..? |
| |
Smasher
Registered: Feb 2003
Posts: 519 |
compo is yours. decision too :) |
| |
Raistlin
Registered: Mar 2007
Posts: 671 |
Yep, I'll use the squared calculation. Thanks Jammer for the idea. |
| |
Jammer
Registered: Nov 2002
Posts: 1335 |
I've used it once already in my own compo but I was yet a bit afraid of utilizing it alone. Hence I mixed it with simple mean - now I regret a bit :D |
| Previous - 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 - Next |
