The Song Fight Bulletin Board

Who is winning SongFight! overall?

Since voting records have been kept, Hostess Mostess has had 41 entries and received a total of 343 votes. That's a whopping 8.37 votes per song.

(according to DillFrog)

Since each fight has a different number of votes, the mean number of votes is kind of meaningless. A better measure is the mean percentage of each fight's votes. We get, on average, 10.9% of the votes, though that's really skewed by a couple big wins. Median is 7.6%.

Top that!

First band to pop into my head- The Hell Yeahs.

16 Entries. 216 votes. I don't feel like doing the math because this is my lunch break (and that's crossword puzzle time, my brain has got a completely different focus) but that should blow your scores out of the water.

Without consulting anything, how can it be anyone but Front?

Or Rone. Remember that time he got 42 votes? That probably pumped up his average pretty good. My guess is that mine would be around 2, but I should probably verify that.

Current Leaderboard!

Frontalot* - 177
Hell Yeahs - 13.5
Hostess Mostess - 8.37
Jolly Roger (including Our Dog and Andrew Reist) - 2.25

*some songs couldn't be included in the count, since they're so old that the votes are not shown in the archive. Basically, this is an average between two songs.

Of course, you have to think about Front. But that starts making the process very muddy. Front only has 4 songs during the record keeping era. 5 if you count Rapper Name Redacted (8 if you count Urine Luck but no one ever does). Can you really be winning with only 4 enties? That leads to a question of what do you consider "winning"...and then the conversation turns to chaos.

In Songfight there are no winners, just survivors.

TEH REAL WINNAR IS TEH LISTENER

Just kidding, we're all fucking losers.

I think Front is a fine winner. Obviously a success story, though not only in SF! I imagine if Fiona Apple ever entered, she would do pretty well. Probably win the whole pie with one song.

I am interested in tracking HM's performance in the votes over time and am having some difficulty thinking of the best single metric:

1) number of votes is important, but some weeks have more or fewer because of whatever, so that number depends on the luck of the draw to some extent

2) percentage of votes is good, but ignores chance: the number of entries. A 50-entry fight with 40 voters won't readily yield as many expected votes as a 10-entry fight with 70 voters.

3) some sort of chance-corrected measure [(#songvotes) - (#songvotes/#entries)]/(#totalvotes) would probably be better, but now we're getting kind of abstract. Though I like this because it can be negative if you could expect a few votes by chance alone but didn't get any.

4) adjusting vote totals or percentages to make winning part of the metric? The stat from #3 would be cool if (#winnervotes) was in the demominator. Then 1.0 means you won every fight you were in.

5) adjusting a band's score it by the number of fights. This would do a disservice to Front since he has relatively few entries. But longtimers like Octothorpe (and Hostess Mostess!) would get a little reward for persistance/consistancy. This is irrelevant to my personal idea of tracking one band's performance. But all this talk about winning made me want to gin up the measures.

And now back to work...

Holy crap, you think too damn much.

by the way, here is what it says on my Artist page. Statistics: 10 entries, 1 wins = 10 percent. But in all fairness, that 1 win was because I told my friends to go vote for me.

HM, we discussed a few other approaches in this thread

Mostess wrote:I think Front is a fine winner. Obviously a success story, though not only in SF! I imagine if Fiona Apple ever entered, she would do pretty well. Probably win the whole pie with one song.

I am interested in tracking HM's performance in the votes over time and am having some difficulty thinking of the best single metric:

1) number of votes is important, but some weeks have more or fewer because of whatever, so that number depends on the luck of the draw to some extent

2) percentage of votes is good, but ignores chance: the number of entries. A 50-entry fight with 40 voters won't readily yield as many expected votes as a 10-entry fight with 70 voters.

3) some sort of chance-corrected measure [(#songvotes) - (#songvotes/#entries)]/(#totalvotes) would probably be better, but now we're getting kind of abstract. Though I like this because it can be negative if you could expect a few votes by chance alone but didn't get any.

4) adjusting vote totals or percentages to make winning part of the metric? The stat from #3 would be cool if (#winnervotes) was in the demominator. Then 1.0 means you won every fight you were in.

5) adjusting a band's score it by the number of fights. This would do a disservice to Front since he has relatively few entries. But longtimers like Octothorpe (and Hostess Mostess!) would get a little reward for persistance/consistancy. This is irrelevant to my personal idea of tracking one band's performance. But all this talk about winning made me want to gin up the measures.

And now back to work...

what we need is a songfight simulator to test your hypothesis.

There used to be a Songfight Simulator. It was called the Sharing Machine. It didn't work to well...

(much better as a record label)

blue wrote:what we need is a songfight simulator to test your hypothesis.

...or a songfight stimulator, hypothetically for the testes.

Mostess wrote:Who is winning SongFight! overall?

Since voting records have been kept, Hostess Mostess has had 41 entries and received a total of 343 votes. That's a whopping 8.37 votes per song.

(according to DillFrog)

Since each fight has a different number of votes, the mean number of votes is kind of meaningless. A better measure is the mean percentage of each fight's votes. We get, on average, 10.9% of the votes, though that's really skewed by a couple big wins. Median is 7.6%.

Top that!

Dillfrog also says Wreckdom beats you more often than you beat them, so aren't they winning then?

rdurand wrote:Dillfrog also says Wreckdom beats you more often than you beat them, so aren't they winning then? :-)

They are the hardest working band on SongFight! And crazy ass fun. I love WreckdoM and applaud their relative victory.

Another thing I want my metric to account for: If you come in 2nd out of 10 songs, it should mean more if the 3rd place winner was far behind you. In some sense the %age of votes and the number of entries takes care of this, but not quite:

a) 10 entries, 100 votes, winner got 50 votes, I got 25, 3rd place got 15

vs.

b) 10 entries, 100 votes, winner got 50, I got 25, 3rd place got 4

I think my (b) song outperformed my (a) song by some amount. Maybe not. My gut feeling is that since each vote is a zero-sum game, every entry that systemmatically attracts votes away from mine is a loss for me.

Deshead's thread is very entertaining and a good start.

everyone already knows the terrorists are winning songfight.

Only if we let them. Music isn't free, you know.