Visualizing the endorsement graph

So, um, hey again. I guess most of you'll be surprised to see me here again, apart from Eluvatar and a few who've seen me around on IRC. Also, sorry for that episode last year. RL issues closed in, but I shouldn't have disappeared like that.

During a recent discussion on endorsements and influence, I thought that it would be interesting to see the endorsement network not merely as a ranking of the nations with the most inbound/outbound endorsements, but also as a graphic. In particular, a force-directed layout would show clusters and allow visualizing things like invasions.

Now, graph drawing (and particular force direction) is a pretty complicated thing to program, but there are some open-source tools available already. So I looked around, and put some data from my scanner into one of these tools, and ended up with a few neat pictures.



To try out the force direction thing, I used a region that is currently occupied (the_proletariat_coalition), and it was great:

someone has a lot of time on their hands.....

Actually I just spent an hour or so on this.

nice

Cool.

The end graph is incredible. I love it. I had an idea about doing this sort of graphical clustering a while back to show embassy interconnectivity in NationStates. But I never thought about doing it to show regional endorsement clustering.

I like it. Is there a way to include the NS influence ratings in something like this? That wiuld help explain or illustrate what many of us think as to our theory of regional security.

I admit I'm clueless with programming, and I do comphehend graphs. So that might be an interesting experiment to see if it can be done, and if it able to show us something useful.

It's not too difficult to display an additional dimension using size or color. The difficult part would be accurately measuring influence, since the ordinary ranking - and even the ejection cost estimate - isn't fine enough. It might be possible to combine ejection-cost surveys from lots of different nations.

Eg:


While neither nation A nor nation B can rank all three nations (CD look identical to A, and DE to B), we can rank C>D>E by combining them.

I'm pretty sure Eluvatar has a lot more experience and data on that topic, though.

I have a plan to eventually make rankings by interweaving N+1 rankings, after N nations do a survey for me of ejection costs.

The best data we have is for the higher ranks of Influence, at http://safalra.com/other/nationstates-influence/

Using this data, and a little bit of guesswork, I see that the nations with Handshaker and above in TNP appear to have between 31% and 44% of all the influence, between them. Looking at just those of them who have been active on the forum, we get a less massive figure of between 20% and 28%. Presumably that figure would be significantly higher were I to include our forumer Trucklers and Vassals.

The problem with including our Trucklers and Vassals in that is that we have much less data on what those lower labels mean. From an experiment I ran some time ago, I believe I can tell you that 1% of all influence in the region falls into the Vassal category. So, for Duckspeakers Handshakers and Trucklers we know they have more than 1.0% and less than 3.2%.

With the assumption that banjection consumes exactly half of the target's influence, and knowing that Zemnaya Svoboda (a Duckspeaker) would nearly all of its influence to banject Great Bights Mum, we can assert that some point in the Duckspeaker range is greater than half of the exclusive lower bound on Negotiator.

That means at least some part of the Duckspeaker range carries more than (1/22)/2 = 2.3%. I don't know if 2.3% is in the Duckspeaker, Handshaker, Truckler, or even Vassal range, however. I can only be sure that 2.3% is not in the Envoy range.

nice graphs

Seriously cool stuff. Nice job.