As with my leaderboard analysis, this is for PC only. If anyone does the same for a console, I'll link it.
The Squadrons feature is one of the few features in the game which exposes raw numbers, so I thought I'd do some statistics on it. Counting every single squadron is impractical, so I took the following sampling approach:
1) Each squadron has a unique 4-character code consisting only of letters and numbers.
2) Generate 100 random 2-character codes from the same valid character set
3) Sample the squadrons whose 4-character codes start with those 2-character codes (which can be done in the in-game search [1])
4) Read the sizes off the search screen and put them in a spreadsheet.
Using the codes to search should give the most random sample as the code of a squadron shouldn't be correlated much if at all with its size.
This gave 535 squadrons, for 100 out of 1,296 possible 2-character prefixes, implying a total number of squadrons around 7,000. Actually, we know the number must be at least 7,451, because there were 746 pages on last season's exploration board ... but it's not bad for a sample, and I'm not bored enough to check the rest. Still, bear in mind that these numbers are estimates. It's better than running a forum poll on "how big is your squadron" would be, though
As you'd expect, the squadron sizes have a big number of small squadrons, and a small number of big squadrons.
This makes a sampling approach to questions like "what's the mean size of a squadron" tricky - for example, say there were only ten squadrons, nine with one member, and one with five hundred members. We try to determine the mean by sampling a random three squadrons ... on no occasion will we be particularly close to the true value of "51" (we'll either get "1" or "167", depending on whether we sampled the big one). In the sample, I got 8.7 players, and it's probably somewhere between 5 and 10 players, but it's very hard to say for certain.
Here's a graph of the squadron sizes versus count in the sample (with logarithmic size axis, to make it easier to read) - as you can see, the most common size is "2" and over 90% of squadrons have a single-digit player count.
This is not the same as saying that most players are in small squadrons, however - only about 30% of players in squadrons are in squadrons with single-digit player counts, and almost half of players in squadrons belong to a squadron with over 100 players - the top 4 largest squadrons in the sample, for instance, each had more players than all the three-player squadrons in the sample together.
So how many players in total are in squadrons? This is hard to tell, because the mean is very tough to estimate, but in the sample taken the mean players per squadron was 8.7, which multiplied by the estimated number of squadrons gives around 60,000 total PC players in squadrons - though a real figure anywhere between 40,000 and 100,000 would easily be compatible with this sample. Still, that suggests the feature has been pretty well-used so far, and hopefully will attract more development in future - it has a lot of potential for further work.
[1] Quick note if you want to try to replicate this: the search will return squadrons with either a tag or a name starting with the 2-character code. I ignored the ones where the name started with the code but the tag didn't, to ensure I didn't count the same squadron twice.
The Squadrons feature is one of the few features in the game which exposes raw numbers, so I thought I'd do some statistics on it. Counting every single squadron is impractical, so I took the following sampling approach:
1) Each squadron has a unique 4-character code consisting only of letters and numbers.
2) Generate 100 random 2-character codes from the same valid character set
3) Sample the squadrons whose 4-character codes start with those 2-character codes (which can be done in the in-game search [1])
4) Read the sizes off the search screen and put them in a spreadsheet.
Using the codes to search should give the most random sample as the code of a squadron shouldn't be correlated much if at all with its size.
This gave 535 squadrons, for 100 out of 1,296 possible 2-character prefixes, implying a total number of squadrons around 7,000. Actually, we know the number must be at least 7,451, because there were 746 pages on last season's exploration board ... but it's not bad for a sample, and I'm not bored enough to check the rest. Still, bear in mind that these numbers are estimates. It's better than running a forum poll on "how big is your squadron" would be, though
As you'd expect, the squadron sizes have a big number of small squadrons, and a small number of big squadrons.
This makes a sampling approach to questions like "what's the mean size of a squadron" tricky - for example, say there were only ten squadrons, nine with one member, and one with five hundred members. We try to determine the mean by sampling a random three squadrons ... on no occasion will we be particularly close to the true value of "51" (we'll either get "1" or "167", depending on whether we sampled the big one). In the sample, I got 8.7 players, and it's probably somewhere between 5 and 10 players, but it's very hard to say for certain.
Here's a graph of the squadron sizes versus count in the sample (with logarithmic size axis, to make it easier to read) - as you can see, the most common size is "2" and over 90% of squadrons have a single-digit player count.
This is not the same as saying that most players are in small squadrons, however - only about 30% of players in squadrons are in squadrons with single-digit player counts, and almost half of players in squadrons belong to a squadron with over 100 players - the top 4 largest squadrons in the sample, for instance, each had more players than all the three-player squadrons in the sample together.
So how many players in total are in squadrons? This is hard to tell, because the mean is very tough to estimate, but in the sample taken the mean players per squadron was 8.7, which multiplied by the estimated number of squadrons gives around 60,000 total PC players in squadrons - though a real figure anywhere between 40,000 and 100,000 would easily be compatible with this sample. Still, that suggests the feature has been pretty well-used so far, and hopefully will attract more development in future - it has a lot of potential for further work.
[1] Quick note if you want to try to replicate this: the search will return squadrons with either a tag or a name starting with the 2-character code. I ignored the ones where the name started with the code but the tag didn't, to ensure I didn't count the same squadron twice.