Map popularity experiment results

[SKB]

I like scaffold more than shine :>

Perhaps shine is a bit overplayed?

It could also be "bad" pairing (complex vs shine, i'd pick complex any day), seeing how some of the maps were presented more frequently.

I'm intrigued by desert, 2f2f, shine and complex having over 300 views each, while the others average 145, with σ~12. What made these four maps come up so much more often?

There's nothing in computing that's random. So, some ranges/values get picked more frequently, as a side effect of an algorithm. For the next experiment one of the maps presented is not going to be randomised, or maybe not fully randomised. I still have no clue how I'll set it up, thus suggestions are welcome.

[Ronald_Reagan]

Also, random won't get you a uniform pattern. Its random.

Best text I could find about this, only skimmed it. The Illusion of Randomness


But first thing that comes to mind to fix your problem is to create an array with all of the possible permutations, randomize the array. Each user gets a different (random) index to start at, and they work their way through the array. Once at the end of the array, it goes to the beginning.

Might not fix it persay, but even it out a bit.

More typical solution would be make it so maps with less views come up more often.

[Orynge]

I think one of the problems with this experiment, is that one desert fanboy and shine hater might've answered this thing 300 times, which would really change the answers.

[failure]

As it stands now, the information that you want and the information that you are gathering aren't exactly represented by the results.

The information you want: Which are the most well liked maps.
The information you are gathering: Each poll taker's map preferences.

These are, by far, not the same thing. If a person sits down for 20 rounds of the poll, you will know which map they like best, which they like least, and the rankings compared to the other map presented in the head to head. This is good information to have, and could help in figuring out the desired information.

However, there are two poll styles that I think would be better and give you the information you want pretty much directly.

First and simplest, put up a single map with radio buttons for "like" and "don't like." Continue through all of the maps. You can repeat to make sure the answers are consistent, but that's whatever.

Second and more complicated. List all of the maps, check boxes, choose 5(or whatever number you want) that you like best. You could also have the reverse of this where you pick 5 that you dislike most. This also has the possibility of expansion to include weighted selections, but that makes it more complicated and might give a level of exactness that isn't necessary.

[SKB]

The information you want: Which are the most well liked maps.
The information you are gathering: Each poll taker's map preferences.

Isn't the end result exactly what I want? A person gets to compare two maps and more picks - more "popularity". The whole poll in essence is like bubble sort, i.e. mass aggregate data (unconsciously?) sorted by people. Anyway, this is just an experiment, so results should be taken with a grain of salt.

---

~3000 comparisons later: rattrap got almost 60 additional votes, moving it to 43rd place, power is 1st. heh

[failure]

Notice the portion where I specifically mention "compared to the other map in the head to head." The head to head view is a skewing factor that doesn't necessarily end up giving you the specific information that is wanted.

If your poll somehow put map A against map B a greater number of times than any other map, and map B sucks, then map A will have an artificially inflated score because of the map to which it is being compared.

In the two poll styles that I suggested, you are getting directly to the exact data you want on the first pass, every time, with no confounding factors. Each map is weighed on its own merits and is either liked or not liked, in the first. In the second, people pick their top liked maps, with an option to put them in weighted order.

These give only the information desired and nothing more, with the least possibility for unintentional bias.

[DogDancing]

I clicked on it when you linked it earlier, my first vote was either Scaffold or Shine..both of my maps. So I voted for the less played one(scaf) and quit since was pinning my maps against eachother :( *cries* without even accurate information. :(

[sniperz]

go go ac_africa :D

[Oracle]

Don't hate the poll, hate the poll maker \:D.

[Woolly]

Results are skewed and shopped , Douze is not #1...



*leaves bomb, walks away*

[Oracle]

Results are skewed

Positively or negatively skewed? \:D
* Oracle is just loving the math jokes.

[Woolly]

Results are skewed

Positively or negatively skewed? \:D
* ShadowFlameZ is just loving the math jokes.

aladeen skewed

[failure]

Hey, I stole that first! And you know it. Dammit.

[Woolly]

Hey, I stole that first! And you know it. Dammit.

are you aladeen ?

[Oracle]

are you aladeen ?

are you the dictator?

[Woolly]

are you aladeen ?

are you the dictator?

I'm aladeen sure I am not Aladeen.

[failure]

are you aladeen ?

Very aladeen.

[#M|A#Wolf]

I'm quite tempted to close this thread. Stop ruining SKB's thread, it was quite productive until the aladeen jokes.

[YourSister]

i used the program for around 15 minutes, voted shortway over all the maps, still it is one of the last :(

[Oracle]

i used the program for around 15 minutes, voted shortway over all the maps, still it is one of the last :(

Well considering the fact that (and I'm just spitballing here) you were probably one of the few, if not only one, who voted for shortway, that make sense...
Sorry :s