The Song Fight Bulletin Board

Lunkhead wrote:Ah, thanks for clearing that up. So fluffy, do you feel like a drop in votes could be the result of a lack of votes from folks like you and frankie, who it sounds like aren't voting (as much, anyway) because the fights are too big for you to listen to all of all the songs? That could be.

It's certainly affected my number of votes. Most weeks with a single big fight I don't even download them. Back when we had more I did, more often
This week I have downloaded and already listened once, but that's because I'm in it.
-bill

I do not believe that the voting has dropped. There has been no change in the system, there has been no change in voter eligibility, there has been no change in the actual number of voters that I can detect.

I understand that you guy are bored and want the next great thing. OK. Got that. We are working on it.

SPUD

i don't really see it being a big deal: the amount of participation (number of entries as well as number of votes) has always fluctuated greatly, and has never dropped to zero... maybe people could spend less time worrying about the number of people voting, and more time voting. just a thought.

when voting for federal, state, or local officials, there is no requirement for me to know who all -- or even any -- of the candidates actually are, or anything whatsoever about any of them. i am only required to select names from lists of names, and could do so completely randomly, or by any selection method i desired, rather than actually choosing the one i think is the "best" out of every choice offered to me. the winners of this process subsequently get a salary (usually) paid in part by my tax money, and have a job which (often) involves making decisions that affect my life, at least in some way. i see no reason why an internet competition where the winner gets their name highlighted on a website, and essentially nothing else, should have a more complex or rigorous selection process.



(edit) to fend off possible nit-picks about my argument before they start: yes i realize that my description of the democratic process shows that it is inherently flawed, but i do not believe those demonstrated flaws justify calls for "fixing" voting on song fight!. rather, the point i am driving at is, go and fix democracy first and THEN worry about your internet song contest. (perhaps the actual point at which i am driving is not nearly so grandiose as this, but that's about the extent of the eloquence you're likely to get prior to 5am, sorry.)

Eric Y. wrote: when voting for federal, state, or local officials, there is no requirement for me to know who all -- or even any -- of the candidates actually are, or anything whatsoever about any of them. i am only required to select names from lists of names, and could do so completely randomly, or by any selection method i desired, rather than actually choosing the one i think is the "best" out of every choice offered to me. the winners of this process subsequently get a salary (usually) paid in part by my tax money, and have a job which (often) involves making decisions that affect my life, at least in some way. i see no reason why an internet competition where the winner gets their name highlighted on a website, and essentially nothing else, should have a more complex or rigorous selection process.

In Australia - voting is compulsory, and anyone that doesn't vote gets fined.

....just a thought.

fluffy wrote:Erik: Yes but those are unbounded sets, whereas there is a set of 25 songs. Whenever I point friends to Song Fight they're all "oh, so I have to listen to all these songs before I can vote? Fuck that."

How is the menu at Threadgills an unbounded set?

The constantly changing specials board?

The infinite condiment bar?

The all-you-can-eat pasta bar?

Salmonella recall on chicken livers?

frankie big face wrote:The constantly changing specials board?

The infinite condiment bar?

The all-you-can-eat pasta bar?

Sure, some restaurants are like that. There are also some restaurants that have a set menu and no specials ever and none of the other stuff you mentioned.

My point is this: there are lots of examples in the real world where people use the word "favorite" to describe a situation where they have sampled some (but not all) of a bounded set and determined one thing that they like better than all others they have sampled. Songfight doesn't have to be any different.

Yes yes I understood your point. And in different contexts things can have different psychological effects. It is hard to select a favorite from a transitory set where the expectation is for you to know all of the items in the set. It is easy to select a favorite from a larger set where you may not have tried everything but you have tried a few things enough times to decide that you really like something.

I AM SORRY FOR LETTING MATH GET IN THE WAY OF SIMPLE EXPLANATIONS. Christ.

Anyway, I would still like to reiterate that, while the voting system on here is "broken," it's also completely pointless. It's not like you really win anything or whatever, and it has no effect on anything else. That is the position I've held for years. I only even make any mentions of how the voting system could potentialy be improved because THIS KEEPS COMING UP FOR SOME REASON so obviously it's VITALLY IMPORTANT to some of you people.

I'm done here.

SongFight! asks one simple question: Which of these songs is the best? Anything more than that complicates things, and we get into telling you, the voter, how to go about making that decision. That's not our job, it's yours.

fluffy wrote:Yes yes I understood your point.

My point is that there should be no expectation that anyone listen to all the songs. I don't think you misunderstand my point as much as you and I just disagree about this.

I'm not very handy with spreadsheets, so I couldn't figure out how to whip up nifty graphs about the votes over time. Perhaps some other person who is bored and good with spreadsheets could, though, using this CSV file with every fight so far, the number of songs, who won, the number of votes, and the start/end dates. (Don't forget to filter out fights with fewer than 2 votes recorded.)

I think the biggest significance of the votes to me is as a possible representation of the number of listeners. It seems like a fight that gets >100 votes (assuming they're not all for one participant) probably had more listeners of more songs than a fight that got 25 votes, though I'm just speculating about that, of course.

Could someone remind me what the problem is again?

The section I originally mentioned is from fight #1 to about fight #34, which represents a lull (though obviously not an unprecedented one) compared to the section before that to about fight #75. As I originally said, it looks to me like we're in a period with a lower average number of votes versus the average for all fights and particularly versus the average for a recent period of higher activity. I didn't say anything about where I thought this current situation was trending. It could just be part of a normal cycle of ebb and flow in the number of votes, as the period of fight #75-110 was also kind of low (avg. = 57 votes / fight). Regardless, I was just curious what people thought may be causing or at least related to the lull in voting. I didn't mean to imply any functional problems with the site, except for a recent seeming lack of participation from the anonymous voters, which is not "something wrong with SongFight!" but just something the site is experiencing at the moment.

EDIT: Also, just in case anybody else doesn't figure it out at first glance, like me, the most recent fights are on the left in that graph.

Uh, I thought we were over 500 fights? Plus, I just flipped through the archives back as far as Lunkhead's reference (to History of My Broken Heart) and there was certainly not a fight that had over 500 votes, not even one that had over 100 votes.

I'm sure there's a simple explanation and I'm misreading the graph, but...


EDIT: I just read Lunkhead's explanation. Way to put up a faulty and misleading graph, Spud.

EDIT#2: I still love it when fluffy gets mad. Every time. It's glorious.

Sorry about the misleading part. It is true, the graph is organized just like the archive, most recent fights first.

About it being faulty, there are over 500 fights, but I took out all the fights on which we have no voting data, which represents the entire narbotic period, plus a few fights for which we lost the vote counts due to either a human error or a code glitch (which, of course, is also human error).

SPUD

Spud wrote:Sorry about the misleading part. It is true, the graph is organized just like the archive, most recent fights first.

About it being faulty, there are over 500 fights, but I took out all the fights on which we have no voting data, which represents the entire narbotic period, plus a few fights for which we lost the vote counts due to either a human error or a code glitch (which, of course, is also human error).

SPUD

Actually, I was just joking, but I do have a question. When there were two fights in a week, does the graph represent them as separate fights or one fight? Because if there are two fights and one got 37 votes and the other 40 votes, I don't think you can rightly say that 77 people voted that week.

For example, "The Last Year" received only 25 votes, but your graph data doesn't appear to show any fights that received much less than, say, 40 votes.

Every fight is shown individually. The Last Year is in there.

Here is a graph arranged from the first fight on which we have data to the latest. Certain anomalous results have been removed, notably fights featuring MC Frontalot and a few others with votes in the stratosphere to try to get a better picture of "normal" turnout. Removing this data also allows us to see a little more detail in the range of interest.

Well, see, that actually does look like a decline. Look at spikes below the 50 mark from halfway to the present.

frankie big face wrote:When there were two fights in a week, does the graph represent them as separate fights or one fight? Because if there are two fights and one got 37 votes and the other 40 votes, I don't think you can rightly say that 77 people voted that week.

That's a good point, Frankie, and one that I hadn't considered. In the data that I gathered, multiple fights that happened at the same time are listed separately. Just glancing at it now, summing the votes for the "fight periods" may actually return the total number of votes per "fight period" to the range of the pre-"HoMBH" era, or higher. In which case, er, um, nevermind.

Of course, some people are likely voting in multiple fights per "fight period", so the total number of -voters- may be lower than the sum of the votes. Anyway, I did include the start/end dates for all the fights, so a better spreadsheet wizard than I could possibly make a graph with the X axis as the dates and the points plotted as the sum of the votes for all fights for the dates. Or I can provide another CSV which combines the multiple fights, but I won't be able to do that till tonight after work.

What happened around fight 370, where it kind of jumps up there?

erik wrote:What happened around fight 370, where it kind of jumps up there?

A switch from multiple to single fight format, which lasted until around 415, with multiples returning from 415 to 432 with a few singles thrown in.