Election for World Government

Threedee said:
Lincoln's probably the safest pre-20th Century bet on that list. I'm surprised that Watermark allowed the list to be so US-centric, given that he's a leftist, and they generally imagine a world government not presided/ruled-over by an American.
Click to expand...

It's an American based board, most of the figures I could think of that were likely to be universally admired were Americans. I did put a couple of Brits in there for Tom to jerk off to though.

This is for data collection to test my system, not to confirm my political beliefs.
 
A turnip could do a better job than any of them. It would leave us alone, for one thing. I'm perfectly capable of governoring myself. Most responsible adults are.
 
The Wokest said:
It's an American based board, most of the figures I could think of that were likely to be universally admired were Americans. I did put a couple of Brits in there for Tom to jerk off to though.

This is for data collection to test my system, not to confirm my political beliefs.
Click to expand...

Well, good, because your political beliefs are wrong.
 
Code: 
def mult_ballots(l, weights):
    return np.concatenate([np.repeat(i[0], i[1]) for i in zip(l,weights)])

wb1 = mult_ballots(data.iloc[:, 2].values, data["Weight"].values)
wb2 = mult_ballots(data.iloc[:, 3].values, data["Weight"].values)
fs = list(zip(wb1, wb2))
for index in range(4, len(data.columns)):
    wb = mult_ballots(data.iloc[:, index].values, data["Weight"].values)
    fs = [x + (y,) for x, y in zip(fs, wb)]
ballots = np.array(fs, dtype=np.int8)

It's not too difficult to make a known amount of lists into an N dimensional array. It's very difficult and annoying to generalize the problem so that you can turn any number of lists into an array of any dimension.
 
Last edited:
You know, actually I think k-means is probably ideal for this sort of data. I'll run the beta version of the scikit implementation when I get home this afternoon on the current ballots, I'll just pick two winners because there are so few data points though. I still need more data as it is, I wouldn't feel like the results of a five seat election would be meaningful without at least 10 ballots.
 
The Ugly Truth said:
A turnip could do a better job than any of them. It would leave us alone, for one thing. I'm perfectly capable of governoring myself. Most responsible adults are.
Click to expand...

THIS.

With regards to the Founders on the Wokest list, the premise, that a select few should rule the world, is exactly what they fought against.

Why do liberals support central rule?
 
Dark Soul said:
THIS.

With regards to the Founders on the Wokest list, the premise, that a select few should rule the world, is exactly what they fought against.

Why do liberals support central rule?
Click to expand...

LOL. Then they should vote to disband the council. Hilarious that such a silly premise triggers you so much. I literally don't care about anything here besides the data, the ridiculous world council concept is just an excuse to get people's scores on a variety of figures to feed into my system.
 
Last edited:
You've seen through me. I started this project on a direct order from the council of the Jews, it's purpose was to help in the Muslim takeover of the west, socialism, and propagation of globalism. And I would have gotten away with it too, if it had been for you meddlesome kids! Trump should send you a letter pretty soon awarding you medals for standing guard and protecting freedom and individual liberty.

maxresdefault.jpg
 
Winners:

George Washington
Score: 10.0
Group members: Billy, Kacper

Martin Luther King
Score: 10.0
Group members: Watermark, Tsuke

Affinities:

Watermark - Billy : 19.60
Watermark - Kacper : 21.31
Watermark - Tsuke : 34.60

Billy - Kacper : 15.43
Billy - Watermark : 19.60
Billy - Tsuke : 25.36

Kacper - Billy : 15.43
Kacper - Watermark : 21.31
Kacper - Tsuke : 25.28

Tsuke- Kacper : 25.28
Tsuke - Billy : 25.36
Tsuke - Watermark : 34.60

The metric used to judge affinity is the same one used in the k-means clustering algorithm - euclidean distance. Technically, this algorithm is considering each persons ballot as a point space, with each candidate being its own dimension. So, in this case, the distance between two points in 22-dimensional space. Tsuke is way out on the edge of the space due to how his ballot was cast, while Billy and Kacper are the closest. Billy is the closest to me, but those two still got grouped together because they're closer to each other than I am to either. So I get thrown in with Tsuke, and the average of our scores gives the highest to MLK (I don't think we agreed on anything else).

This run didn't produce anything way out of left field. But I think I'm still going to work on the grouping a bit. I think it might be better to weight ballots according to their distance from the centroids, and take the score of all ballots given that, than simply group ballots and not consider anyone's score besides the groups. But what I've thrown together is good for a demonstration of concept.

Still I want more data points, this code should be good for any number of ballots and I should be able to easily add more in. Come on cast more ballots people!
 
Last edited:
If you increase the grouping to three you do get some crazy results:

Winner 1 is George Washington
Score: 10.0
Group members: Kacper


Winner 2 is Alexander Hamilton
Score: 8.5
Group members: Watermark, Billy


Winner 3 is Adolf Hitler
Score: 10.0
Group members: Tsuke

LOL, Adolf did it.

What really happened here is that Tsuke was thrown into his own group, and since all of his scores were ties the argmin function in Python literally just takes the first element in the list that matches, which in this case happened to be Adolf. Which is why I want more ballots, to reduce the likelihood of ties like this which don't give meaningful results.
 
You know what, there's definitely something wrong with the code I put in to enforce equal cluster sizes. If you take it out you get this:

Winner 1 is Abraham Lincoln
Score: 10.0
Group members: Watermark |
Winner 2 is Adolf Hitler
Score: 10.0
Group members: Tsuke |
Winner 3 is George Washington
Score: 10.0
Group members: Billy | Kacper |

Which is objectively better, as Billy and Kacper are closer. There's no reason I should have been swapped out for Kacper in this instance. I've got to fix that.
 
You must log in or register to reply here.
Back
Top