You’ll find the SAAB showing up as:
Airborne Variant – Engines: 2
Cargo Variant – Engines: 2
Passenger Variant – Engines: 2

Ah, in this case, “other” means the user had more than 12 cores—maybe they had a Mac Pro, with 8 to 12 physical cores (and thus 16 to 24 hyperthreaded cores reported in the data collection).

In the raw data, we get a small number of “0” core CPUs—which just indicates that for some reason we couldn’t get information about the CPU from the operating system. (Likewise, we have a few entries in the raw data of people with “0” GB of RAM.)

Sigh. This is why we can’t have nice things. I’m going to go ahead and make the data proportional, but for what it’s worth, it’s not quite that simple for a number of reasons. Off the top of my head:

1) The data from the JSON is sampled (in a randomized way) from the full data set. (This is a matter of practicality on our end.)
2) The data has been filtered a bit to remove items below a certain threshold. (Lest we reveal the habits of, say, the one person in the world who owns a particular joystick.)
3) We don’t know what percentage of users opt-in to send usage data. (Ballpark: somewhere between 5% and 50%…??)
4) The distribution of new users versus old users in the data is still likely to be skewed toward the new, since only new users are *forced* to make a decision on data collection. (People who had X-Plane installed prior to 10.40 are automatically opted-out until they open the Operations & Warnings window, at which point they are asked to make a decision.)
5) Software sales skew toward tons and tons of sales upon the release of a new major version, with a much smaller “long tail” following thereafter. Even if we knew *exactly* how many people had fired up X-Plane in the last 6 months, it would tell you relatively little about how many sales we made at the time of the X-Plane 10 release.

Two things confound this analysis:
– first, the data from the JSON is sampled (in a randomized way)
– second, it’s been filtered a bit to remove items below a certain threshold.

Re: flight controls, we’re reporting on any controls which could potentially be used to control X-Plane. (There’s nothing except practicality stopping you from using a racing wheel to fly!)

The fact that the SAAB A340 isn’t mentioned means it most likely either a) wasn’t flown by anyone sending usage data in the given time period, or b) was flown a very small number of times by people sending usage data. There’s a third possibility—that it was flown a lot, but was the victim of incredibly unlucky sampling—but that’s very unlikely.

Publishing more info is on my to-do list, but it won’t happen soon, unfortunately.

Re: what X-Plane sends, it actually just URL-encodes the information it’s sending, so it makes a request to a URL like this: http://collection-endpoint.com/?gpu=“Intel Integrated”&cpu=”Core i7″&os=”Windows”

– The amount of flight controls (ca. 80.000) are roughly the double of the amount of cpu’s (ca. 40.000), numbers are extracted from the JSON file. The graph suggest that a mouse is the most important flight control. I think this is overstating the importancy of the mouse as mouse control, because many users use two or more controls.

– There are different flight controls mentioned that will not be used in X-Plane but the are connected to the PC for other purposes. Like steering wheels for racing games. Are only the controls used in X-Plane measured or are all USB controls counted?

– I checked the data with the airplane add-ons I have in my virtual hangar. I discoverd that the SAAB A340 (LES) is not mentioned. I cannot believe that this popular add-on is not used after september.

Ah, forgot: It would be nice to see what exactly sends XP to you. I assume it’s a simple text file?

Looking to the json file it looks like it is possible to get an idea of the total number of active X-Plane users.