BREAKING: NATIONSTATES ALGORITHM EXPOSED-FAVORS OSIRIS, TSP!

[Ever-Wandering Souls]

-Site lagged into an extra post, sorry!-

[Rhunnland]

Correct me if I'm wrong, but in order to update without a reload wouldn't it need to make a request from your browser to the server to get that specific information?

There are Strict Rules for interacting directly with the HTML NS Site with scripts - on a serious note, make sure you follow them if you ever attempt such a thing.

Definitely; I wasn't familiar with those specific rules but I was under the assumption that they were forbidden (or restricted in this case) for bots to use. Thanks for linking, though

[Lord Dominator]

So what you're saying is, the more ideologically consistent a GCR is on the R/D/I spectrum, the more nations they get

[Gorundu]

So what you're saying is, the more ideologically consistent a GCR is on the R/D/I spectrum, the more nations they get

Ideologically consistent as in recent actions, or throughout history? In any case I'd think Balder is the most ideologically consistent of the sinkers, yet...

[Ecclestia]

Wishing I had the time to be able to work that out. Very interesting indeed and thanks for sharing!

[Neo-Kahiki]

Not sure whether to find this amazingly interesting, or mildly disconcerting on a different level, or both...

[Lord Dominator]

So what you're saying is, the more ideologically consistent a GCR is on the R/D/I spectrum, the more nations they get

Ideologically consistent as in recent actions, or throughout history? In any case I'd think Balder is the most ideologically consistent of the sinkers, yet...

When thinking of the Independent alignment as being something in which a region both raids and defends, TNP has generally been the most consistent in doing both.

[Francois Isidore]

Typical algorithms smh

[Klaus Devestatorie]

The game admins should actually bump one of the regions to get about 2% more foundings than the others without telling anyone, and then sit and wait to see how long it takes anyone to catch on with hard numbers.

[Aurum Raider]

The game admins should actually bump one of the regions to get about 2% more foundings than the others without telling anyone, and then sit and wait to see how long it takes anyone to catch on with hard numbers.

What they should really do is bump one of the regions to get 2% more CTEs than the others. Nobody will catch on to that one.

[The Stalker]

Nationstates' spawn algorithm has been messed since day one! Hell has never spawned a nation! I demand 10% of all evil nations should spawn in Hell!

[Wabbitslayah]

The real travesty here is that the GCRS arent even properly colour coded. Like a not green TEP? Not light blue TSP? What the heck.

[Numero Capitan]

The game admins should actually bump one of the regions to get about 2% more foundings than the others without telling anyone, and then sit and wait to see how long it takes anyone to catch on with hard numbers.

What they should really do is bump one of the regions to get 2% more CTEs than the others. Nobody will catch on to that one.

Warzone Sinking Sand? Your nation CTE's after 24 hours of inactivity in that region

[Aurum Raider]

Warzone Sinking Sand? Your nation CTE's after 24 hours of inactivity in that region

That is unironically a $100 april fools idea.

[LollerLand]

What they should really do is bump one of the regions to get 2% more CTEs than the others. Nobody will catch on to that one.

Warzone Sinking Sand? Your nation CTE's after 24 hours of inactivity in that region

That would actually be really cool!

[Numero Capitan]

Would allow for some extra flouncy retirements from the game too

[Fauxia]

Back in my day the rants said that the bias favored TNP *shakes fist*

TNP is only 4 new nations behind TSP, so they can be accused of equal advantage!

The significance of this should not be dismissed. The North Pacific receives more nations than it should despite already being the largest region in the world. Clearly, the algorithm is designed to make the rich richer and the poor poorer.

[The Atlae Isles]

As a high school student stuck inside studying AP Statistics, this is funny and kinda hurts at the same time.

Where's the chi-square test!? Where's the p-test!? Why are you using a pie chart?!?!? (pie charts suck fite me)

Ignoring the class nations because that would probably refute my point and bc they're outliers (this might not be the most sound logic )

We'll use the chi-square homogeneity test, because we're trying to prove/disprove a uniform distribution. Our null hypothesis is that there is a uniform distribution, and our alternative hypothesis is that there is not a uniform distribution, and that there is a skew. Level of significance is about 0.05, like it is for most things.

For this test, these assumption conditions must be met:
1) Expected Counts are above 5 (which is why I removed the class nations )
2) The sample is representative (It only applies to this population during this specific time span)
3) The sample is independent (you can't be founded in more than 1 region).

	The West Pacific	The Pacific	The East Pacific	The North Pacific	The South Pacific	Total
Observed Nations	3205	3232	3273	3354	3358	16422
Expected Nations	3284.4	3284.4	3284.4	3284.4	3284.4	16422

I got the expected counts by taking the total number of founding (minus class nations) and dividing by the number of feeders (5). It's not a whole number because I removed the class nations.

X^2 (chi squared) = sigma((observed-expected)^2/expected) = (3205-3284.4)^2/3284.4+(3232-3284.4)^2/3284.4+(3273-3284.4)^2/3284.4+(3358-3284.4)^2/3284.4+(3358-3284.4)^2/3284.4 => 6.0936

Degrees of Freedom: 4 (number of regions(5) - 1)

p(X^2 > 6.0936, df = 4) = 0.1922
0.1922 > 0.05, our significance level
Fail to reject null hypothesis

Thus, there is an insufficient statistically significant level of evidence that suggests that there is a skew in NationStates' founding algorithm. There is a 19.22% chance that the differences could have been explained via random chance and the associated variation, above our level of significance.

I don't even know how much time I spent on this post. I guess I'm just jaded by all my stats homework and should stop procrastinating. I might've made a few mistakes here or there, so don't flame me for that.

Congratulations, though, to TEP and Lazarus, for being decidedly average.

Luv ya.

EDIT: I realized after making this post that I didn't do the sinkers either. Oh well. I'm too tired for this. The AP test is only 45 minutes long now anyway.

[Ever-Wandering Souls]

Clearly...and by clearly I mean that I did not consider statistical significance at all...there is favoritism in the algorithm towards certain regions!!

Thank you, Altae, for doing what I could not!!

xD

No, really.

If you are so inclined - I figured that the Feeder difference would not be significant, but was curious if the Sinker one was, given larger comparative gaps, or not.

PS: If not a pie chart, what *would* you use for "share out of a total?" Would you just compared the numbers with bars?

[Aurum Raider]

PS: If not a pie chart, what *would* you use for "share out of a total?" Would you just compared the numbers with bars?

Bar charts are better for displaying numerical information of this sort.

In groups of similar sizes, it is difficult to see the difference between two slices of a pie chart - which kind of defeats the purpose.

[The Atlae Isles]

Thank you, Atlae, for doing what I could not!!

FTFY

If you are so inclined - I figured that the Feeder difference would not be significant, but was curious if the Sinker one was, given larger comparative gaps, or not.

challenge accepted

No class nations this time! yay

We'll use the chi-square homogeneity test again, because we're trying to prove/disprove a uniform distribution. Our null hypothesis is that there is a uniform distribution, and our alternative hypothesis is that there is not a uniform distribution, and that there is a skew. Level of significance is about 0.05, like it is for most things. (I copy-pasted a lot of this)

For this test, these assumption conditions must be met:
1) Expected Counts are above 5
2) The sample is representative (It only applies to this population during this specific time span)
3) The sample is independent (you can't be refounded in more than 1 region).

	Balder	Lazarus	Osiris	Total
Observed Nations	250	259	280	789
Expected Nations	263	263	263	789

I got the expected counts by taking the total number of refoundings and dividing by the number of feeders (3). It is a whole number this time.

X^2 (chi squared) = sigma((observed-expected)^2/expected) = (250-263)^2/263+(259-263)^2/263+(280-263)^2/263 => 1.8022

Degrees of Freedom: 2 (number of regions(3) - 1)

p(X^2 > 1.8022, df = 2) = 0.4061
0.4061 > 0.05, our significance level
Fail to reject null hypothesis

Thus, there is an insufficient statistically significant level of evidence that suggests that there is a skew in NationStates' refounding algorithm. There is a 40.61% chance that the differences could have been explained via random chance and the associated variation, above our level of significance. As noted, much of the difference comes from Osiris having more rebounds than expected.

Interestingly, the significance of this is actually lower than with the feeder one. Personally, I think it's due to the fact that there's a larger sample size for foundings (16422) compared to refoundings (789). Looking at those numbers side by side also show that a lot of people came to NationStates from Drew (like a lot) but not many decided to actually come back once they got bored.

PS: If not a pie chart, what *would* you use for "share out of a total?" Would you just compared the numbers with bars?

Hm, maybe. I mainly objected because many statisticians (including my teacher) think that they're not that intuitive and they're highly overrated. Also, I do most of my homework on paper so a pie chart would mean using a protractor.

If we must do pie charts, there's one developed by Florence NIghtingale that would be pretty fun.

I could go the extra mile and go make a chart myself, but I'm lazy and I'm not that good at Excel.

[Ever-Wandering Souls]

[Kanglia]

If you are so inclined - I figured that the Feeder difference would not be significant, but was curious if the Sinker one was, given larger comparative gaps, or not.

challenge accepted
(snipping the good parts for length)

Brilliant work on both of them mate. Would love to see this be drawn out over a longer period of time, but I feel like that'd probably be a pain in the ass to get the numbers on

[Grenartia]

Gr8 clickb8 m8 I r8 8/8

[Ever-Wandering Souls]

challenge accepted
(snipping the good parts for length)

Brilliant work on both of them mate. Would love to see this be drawn out over a longer period of time, but I feel like that'd probably be a pain in the ass to get the numbers on

If anyone was assed, you could do it over a much greater period, or perpetually, with the happenings shard. But that's coding work. The large sample size of founding over one day was a great chance to do things the simple way.