I hear you. I love Linux as a general program but in terms of software for media production Windows / Apple still are a must. Especially when you are actually in the industry, not just an amateur, given that you need to be able to reliably cooperate with others and be able to share data across computers and to standards easily.

On the other hand things are moving. On the film, not photography side at least, DaVinciResolve is cross platform PC / Linux / Apple since the start. It's certainly established itself as one of THE industry standard grading softwares now and is making strides in actual video and audio editing / cutting, too. I've edited short clips on Linux entirely by now. It's not quite there in editing features to rival the industry standards for long-form film / documentary - but it goes a long way.

My hope is that someone can really convince Serif to add Linux versions of their Designer / Photo software - I can't see Adobe ever doing that. Darktable is a solid alternative to Lightroom - minus some of the special editing features.

That depends. There are professional commercial 3D and film production tools released specifically for Linux. Ask pro film makers about it.

There was even some video about making of the Hobbit film, where you can see KDE desktop in the background with open tools and editor. So, Linux has no problem addressing creative production needs. Some needs have less coverage, for instance book publishing which is dominated by Adobe, and Adobe are pretty bad at Linux support.

"Optimal sample size" will lead to loss of precision, so I don't see any difference between smaller cake piece and larger cake piece - its still a piece, if its a piece.

Claim that "it is skewed towards people interested in an OS poll" makes no impact on the vote distribution, because it affects everyone equally.

Claim of "other thousands" is taken out of thin air as its active users that are present on the forum and not random people or inactive users.

Claim "public poll on the KSP is unrelated to KSP" is likewise speculative, because KSP puts heavy emphasis on community, and its official community is its official forum.

There is only one thing to criticize - its sample size. However that topic was been there in public for quite some time, so its not like taking a sample of 1% of the data ocean, its 1% collected over sufficient period.

This not really the time or place to talk about statistics, but... it depends. If you assume the total population is more or less evenly distributed / homogeneous, then sampling 1% might get you enough to extrapolate up to 100% without too much of a statistical error.

A lot of the statistics you see used in polls is based on assumptions regarding the global statistical population it analyses, such as having a normal distribution, or at least a "speculated" known distribution. Most of the times this works well enough if you pick a sufficiently represented/dispersed subset from a much greater whole.

Even if you don't make assumptions and look at the distribution as you're getting new data, that should be fine. If you're far off the mark you should be able to figure out something is wrong by increasing the subset you're analyzing and keeping a close eye on trends.

What I'm trying to say is that the math works, the studies of probabilities and statistics have been laid down for hundreds of years. Are the people answering the poll good enough to provide an extrapolation base? You'll have to track the poll results as they're being answered and see the overall trend of the deviation you're getting.

An e.g.: if you notice the number of Linux users growing disproportionately to mac and Windows users over a set period of time, then your initial subset was biased. If you keep getting evenly distributed proportional growth over a long period of time, then your initial population was already a good representation of the general population.

Grab MATLAB and plot some graphs if you're into that sort of thing..

Oh cmmon, I show you the real thing and you soft-talk it into distributed homogenization.

Here, 23%, public poll, in 10/10 rated game.

I have already purchased few DRM-free titles on HB, exactly because GOG does not offer Linux version of them since forever. Speaking of which, HB has no Galaxy. Them, priorities and resulting purchase loses. Don't warp me into MATHLAB, pretty please? :)

Nope OS is a Java Interpreter/Compiler/VM AFAIK
iOS however... I think there is Unix under it
Either way you can't "market share" Linux or count his frontend users. You can however count the server machines on the internet, and there at least they mainstreamed the shit out of microsoft.

no, Android uses Linux under the hood.

Linux also pretty much rules the embedded market. From routers, smart-TVs to cars and washing machines, and even satellites.

I think it is actually understandable that Galaxy for Linux is no a priority for GOG. The increase in market share by introducing Galaxy for Linux would be negligible. The reason is two-fold: One, Linux still has a smaller userbase than Windows. Two, typically Linux-users care about things like privacy and having complete control over their computers. That's one of the main reasons to use Linux. So I guess the overlap between Linux users and people who don't like Galaxy (or any clients) is quite large. Which reduces the possible Linux-Galaxy users again. (that's of course a personally biased opinion since I am one of those Linux-users who will definitely not use Galaxy, ever)

So probably if GOG introduced Galaxy for Linux, a handfull of users would cry "Yay!" - while a bigger number of users would cry "Booo! Stay away with Galaxy!" ... so why dedicate resources to that?

It looks different in the Mac world. Sure, Mac users are also a smaller number than Windows users, but typically they are comfortable with being spied upon and using clients and so on. Otherwise they wouldn't be using anything made by Apple. So making a client for Mac is sensible.

Sure. Take a look at the comments in the article linked by the OP, and look at how uncomfortable those Linux users are with clients. </s>

The comparison between Windows' and Linux's userbases is pointless. Under that view anything other than Windows support won't ever make sense financially. The real question is: why does GOG offers Galaxy for macOS? That cost should obtain a way better ROI if it were used to improve the quality of the Windows client, shouldn't it?

gog had mac os downloader aka galaxy precursor from early days in 2012 or something like that.

dropping support for existing os you supported for years and not adding support for another new os are two different things.

Fair point, hadn't considered that.

But then again, this merely moves my question back in time: why did GOG decide to develop a macOS (OSX back then) Downloader for such a tiny userbase? All the excuses/explanations used today for (the lack of) Linux Galaxy should apply back then for OSX Downloader, shouldn't they?

None of your issues had anything to do with maths, except valid claim of insufficient sample size - which was still best-possible. WinterSnowfall merely said that he needs more samples, but this is not the ground to discard results of an already existing sample.

The issues I mentioned that can alter the results of a poll have been studied to determine how they make the data deviate from the real population. That's math. But also common sense.

You are giving a disproportionate importance to a specific piece of evidence without questioning it because it fits your expected results, and disregarding anything else that is said, even neutral explanations of how a professional statistical study is done. That's confirmation bias.

I don't know how to keep debating without repeating what I have already said. I have never enjoyed neverending quote wars. I think at this point we can assume no one is going to change the other's views.

What do you mean with "giving disproportionate importance"? There is one poll, I shared its results, thats it. You covered it with validity bias, like "it must be Linux users sturming over it, because I think so (aka common sense)". Thats same for someone to claim Web statistic is sturmed by Windows users on purpose.

There is no professional study ever done, I have not claimed that I have done any study. If you want to do it - go ahead, I just have no idea how you'd do it beside using the same poll.

I was not trying to change your view, I posted valid sample.
It says in AAA spaceflight simulator that provides binaries to all OSes, the 23% user base are Linux gamers. Its been open to public vote for about month so far.