Queue/Batch Discussion Thread

[Exias]

Re: Queue/Batch Discussion Thread

Awesome, once my song copyrights get back I'll submit my simfiles for them, but I don't really have give myself permission to use my own work now do I?

[TC_Halogen]

Re: Queue/Batch Discussion Thread

Well, unless if your music is on a record label, no need to give yourself permission.

[flashpantss]

Re: Queue/Batch Discussion Thread

If you guys want to use any of my songs, you have permission. I will constantly have new songs up.

http://flashpantss.newgrounds.com/

[bmah]

Re: Queue/Batch Discussion Thread

You sent the link to all judges via PM for sure.
and that link was still alive, I forwarded it again.

BTW I was thinking of 13 points or more as borderline but that's a bit harsh.
So, new idea is like these, this time.

[Accept]
+ + + (15)
+ + +. (14)
+ +. +. (13)
+ + +? (13)
+ + ? (12)
+ +. +? (12)
+. +. +. (12)
+ + - (11)

[Borderline]
+ +. ? (11)
+. +. +? (11)
+ +? +? (11)
+ +. - (10)

[Reject]
+. +. ? (10)
+. +? +? (10)

So the boundary between borderline and reject is within the point division of 10? How about setting the boundary between points, such as 9 is a reject and 10 is borderline? Also, consider the other combinations as suggested by Silvuh.

If you guys want to use any of my songs, you have permission. I will constantly have new songs up.

http://flashpantss.newgrounds.com/

Promote your songs on this thread: http://www.flashflashrevolution.com/...d.php?t=112805

[jimerax]

Re: Queue/Batch Discussion Thread

+ + - and + +. - are as good as 1 point higher reqs, since - in these case is considered to be "way too underrated".
so we can avoid rejecting files just because of one judge's rating.

also, + +? ? is a reject.

[dore]

Re: Queue/Batch Discussion Thread

makes sense to me because 10 is pretty weak if the judges are consistent with each other, but strong if two judges like it and the other hates it for some reason

[bmah]

Re: Queue/Batch Discussion Thread

That's pretty much the logic I'm hoping for there. That should fix the odds and ends that result from quantitative determination.

edit: dore, you were planning on cutting White Walls in half, right? If you can recall, can you specify where you want it cut?

[Patashu]

Re: Queue/Batch Discussion Thread

maybe strictly points based isn't the right way to formalize it, then?

i'm picturing it something like this: a file has an imaginary true rating, X, and the judges are three samples of this rating, varying according to some distribution (e.g. a normal distribution). using these three samples we can assign a distribution to the imaginary true rating, and if it is likely enough that it falls in accept we should accept it.

this means that a 'meh meh meh' 10 is weak because that assigns a low variance to the predicted true value, while a 'yes yes horrible' is stronger since it has more variance and a substantial chance it's in the 'accepted' range for real

[Silvuh]

Re: Queue/Batch Discussion Thread

Standard sum rating:
(disregarding standard deviation)
15 + + +
14 + + +.
13 + + +?, + +.+.
12 + + ? , + +.+?, +.+.+.
11 + + - , + +.? , + +?+?, +.+.+?
10 + +.- , + +?? , +.+.? , +.+?+?
09 + +?- , + ? ? , +.+.- , +.+?? , +?+?+?
08 + ? - , +.+?- , +.? ? , +?+??
07 + - - , +.? - , +?+?- , +?? ?
06 +.- - , +?? - , ? ? ?
05 +?- - , ? ? -
04 ? - -
03 - - -

Highest reasonable rating:
(adding the standard deviation)
15 + + +
14 + + +., + + +?
13 + + ? , + +.+., + + - , + +.+?
12 + +.? , + +?+?, + +.- , +.+.+.
11 +.+.+?, + +?? , +.+.? , + +?-
10 + ? ? = +.+.- , +.+?+?, + ? - , +.+??
09 +.+?- , + - - , +.? ? , +?+?+?
08 +?+?? , +.? - , +?+?-
07 +.- - , +?? ? , +?? -
06 +?- - , ? ? ?
05 ? ? -
04 ? - -
03 - - -

Lowest reasonable rating:
(subtracting the standard deviation)
15 + + +
14
13 + + +.
12 + +.+., +.+.+.
11 + + +?, + +.+?
10 +.+.+?, + + ?
09 + +?+?, + +.? , +.+?+?, +?+?+?
08 +.+.? , + + - , + +?? , +.+??
07 + +.- , +?+?? , + ? ? = +.+.- , + +?-
06 +.? ? , +.+?- , +?? ? , ? ? ?
05 + ? - , +?+?- , +.? - , +?? -
04 + - - , ? ? - , +.- -
03 +?- - , ? - - , - - -

Curved "balanced" rating:
(Adding or subtracting a curved standard deviation based on the direction of the variance)
15 + + +
14 + + +., + + +?
13 + + ?
12 + +.+., + +.?, + + -, + +.+?= +.+.+.
11 +.+.+?, + +.-, +.+.?
10 +.+.-
09 + +?+?, +.+?+?, +.+?- , + +?- = +.+?? = +?+?+?
08 + +?? , +.+.? , +?+?-
07 + ? ?
06 +.? ? , + ? - , +?? ? , +?? - = ? ? ?
05 + - - , +.? - , ? ? -
04 +.- -
03 +?- - , ? - - , - - -

Standard deviations:
(non-sequitur parenthetical for zen)
00 + + + , +.+.+., +?+?+?, ? ? ? , - - -
05 + + +., + +.+., +.+.+?, +.+?+?, +?+?? , +?? ? , ? ? - , ? - -
10 + +.+?, +.+?? , +?? -
11 + + +?, + +?+?, +.+.? , +.? ? , +?+?- , +?- -
15 + +.? , + +?? , +.+?- , +.? -
17 + + ? , + ? ? , +.+.- , +.- -
20 + +?-
21 + +.- , + ? -
23 + + - , + - -


(And they are ordered from highest to lowest in each row.)

... I hope this is useful somehow. It doesn't really help with the "likely" because I forgot how to do, like, confidence intervals and all that statistics-y stuff... Since after taking that AP test ... bwoop, aaaall gone...

[Patashu]

Re: Queue/Batch Discussion Thread

<3

[bmah]

Re: Queue/Batch Discussion Thread

Oh, that works really nicely.
p.s. the colors help a lot for visual representation O.O

[dore]

Re: Queue/Batch Discussion Thread

edit: dore, you were planning on cutting White Walls in half, right? If you can recall, can you specify where you want it cut?

Yeah if you have to chop it up, I'd like the (12) in m262 to be the last note of part 1 and the first note of part 2 (with like a measure of two lead-in on part two, so that there's a little overlap and continuity).

[jimerax]

Re: Queue/Batch Discussion Thread

statistical analysis there

[Patashu]

Re: Queue/Batch Discussion Thread

so? even seemingly simple ideas like 'when you look at a book on amazon, you should be recommended books like it' requires data analytical techniques

[jimerax]

Re: Queue/Batch Discussion Thread

more like listing mean values of these combinations with error bars in a graph