Steam Linux Usage Came Up Short In January

[asdfblah]

I am doing all my work on Linux. Causally I play CS and TF2 on Steam. But my Radeon 4850 ( which is more than enough for these games) sucks at these games with the open source drivers.
For this reason only I dual boot in to W7.

Try this: http://phoronix.com/forums/showthrea...-driver-report

[Temar]

HB numbers say that about such percentage amongst HB users. It's clearly higher than Linux desktop usage according to your own estimations (since you mentioned 1%). And, since HB is one of the few (and surely the biggest) digital distributors where you can buy DRM-free Linux games you can extrapolate these numbers to some degree. Less or more choice in general is irrelevant, since we are talking about cross platform games (i.e. in the context of those games there is equal choice).

So what do you want to prove with these stats? They certainly don't support your original assumption that Steam stats can not be used to extrapolate the percentage of Linux gamers. HB stats are only good for estimating linux sales for cross platform releases. They are useless for estimating the percentage of linux users/gamers.

So why are you clinging to these stats? It's like clinging to w3schools.com statistics to prove Linux usage above 2%.

[shmerl]

I'm not proving anything, I'm presenting them as useful stats, since you can select uniform statistics from there (cross platform releases). On Steam you can't select that, so Windows only games obscure the numbers rendering them quite useless in general (even in the context of Steam itself).

[grok]

Why do people keep talking about being "prompt for a survey". What the heck is that anyway? I've never seen such a survey pops up on Steam, neither on Linux nor on Windows. And why would they need a survey for the statistics when they already know what client you are signing in from? Also, who said their statistics is based on a survey rather than simply checking sign-in logs?

I think I was surveyed twice, on Windows long ago and on Linux not long after the launch.
Steam has been doing these surveys for very long. A survey would tend to mean you're taking a sample, the sample size may be considerably high but less than 100%.
There may be technical consideration : scanning the whole config at each login would slow login down? (esp. a decade ago), the scanning itself plus the upload, millions times every day.
Ethical consideration : people can object to being scanned every single fucking time with the result sent to the mothership. People would have found out and the software and whole distribution platform would have gained a reputation of being spyware.

[GreatEmerald]

shmerl, Temar, I think you two are debating potential vs realised userbase there. Steam statistics represent realised current userbase of those who are using Steam on Linux. HIB statistics represent the potential userbase, that is, the userbase one would have under more or less ideal conditions. Since Steam has DRM and is not likely to drop it in the future, while HIB always has no DRM, there is no way that Steam would reach the numbers of HIB (unless you count SteamOS, of course); but other services that are more Linux-friendly (or otherwise have conditions closer to ideal, like all games being strictly cross-platform) could.

[shmerl]

GreatEmerald: You are right, I'm talking about potential. I.e. for example from perspective of some studio which contemplates whether to make a Linux release or not based on estimating some potential sales. Steam numbers are useless for them.

[GreatEmerald]

There may be technical consideration : scanning the whole config at each login would slow login down? (esp. a decade ago), the scanning itself plus the upload, millions times every day.
Ethical consideration : people can object to being scanned every single burning time with the result sent to the mothership. People would have found out and the software and whole distribution platform would have gained a reputation of being spyware.

I don't think that technical consideration is valid for OS choices. The scanning and sending amounts to the data around equal to the `uname -a` line.
The ethical consideration is more likely, and I think that's why they chose this method.

[Temar]

I'm not proving anything, I'm presenting them as useful stats, since you can select uniform statistics from there (cross platform releases). On Steam you can't select that, so Windows only games obscure the numbers rendering them quite useless in general (even in the context of Steam itself).

Windows only games don't obscure anything. Steam stats are still more or less accurate when it comes to estimating Linux adoption. As they are in line with all the other stats we have about Linux adoption, they are also believable.

They might not be accurate for _sales_ estimates but that's not the point of these stats.

EDIT: You are right GreatEmerald, I think we are debating different aspects here.

[shmerl]

They might not be accurate for _sales_ estimates but that's not the point of these stats.

Yet, often Steam numbers are brought as a proof that Linux is not a very viable option for developing. That was my whole point above, such kind of "proof" is wrong.

[mmstick]

I am doing all my work on Linux. Causally I play CS and TF2 on Steam. But my Radeon 4850 ( which is more than enough for these games) sucks at these games with the open source drivers.
For this reason only I dual boot in to W7.

How? I even have a laptop with HD 4650M that plays TF2 fine in Linux.