Steam Hardware & Software Survey For August

These always surprise me and inform me as to how much of a bubble I live in. Frequenting sites like [H], I assume that everyone has a powerful system; quad-core, DX12 capable graphics, etc, but in reality, typically around half in most scenarios are using much older hardware.

Makes me realize how lucky I am to afford even the modest system I have.
 
I didn't realize how many people were running dual processor setups... 46.87%...

SEEMS LEGIT!
 
I hate to say it but I want windows 10 now. Not because of anything really important , I am on 7 atm and love it, but I am keeping a close eye on the ps3 emulator out there running dx12. It looks amazing and I can't get it on win 7. :(
 
The STEAM surveys are bullshit, because they don't include everyone. I have maybe been surveyed ONCE this year, and I may not actually have been. How exactly can that result in reliable OS statistics? I'm sorry, but this is no way to do surveys. This has been a problem for a long time, and I don't see VALVe improving this.

It's not like I even declined to be surveyed, I just haven't been presented with it, and there are lots and lots of people that report the same.
 
I always assumed it did the surveying in the background once agreed to it from then on.
 
The STEAM surveys are bullshit, because they don't include everyone. I have maybe been surveyed ONCE this year, and I may not actually have been. How exactly can that result in reliable OS statistics? I'm sorry, but this is no way to do surveys. This has been a problem for a long time, and I don't see VALVe improving this.

It's not like I even declined to be surveyed, I just haven't been presented with it, and there are lots and lots of people that report the same.

lol don't be obtuse, the Survey is monthly and is automatic, you only are asked to submit to the steam survey when they update or change the agreements and etc.. nothing else.. every steam patch or steam update register your machine and count to the steam survey but they are only published monthly..
 
  • Like
Reactions: AK0tA
like this
lol don't be obtuse, the Survey is monthly and is automatic, you only are asked to submit to the steam survey when they update or change the agreements and etc.. nothing else.. every steam patch or steam update register your machine and count to the steam survey but they are only published monthly..
That can't be right. They asked me if I wanted to participate a while back and I said NO. They haven't asked me in a while, but I haven't given them permission to run the survey of my machine since the very first one one which I participated in, so maybe they got the hint finally.
 
I might be wrong, but much of that data can be gained without a formal survey, can't it? Isn't most of the hardware (and listed software, like the OS) visible to them just by running games through Steam? For example, websites see your screen resolution, OS, installed languages, etc just by your visiting them (with Java on); no survey is taken. Doesn't Steam work on the same principle?
 
That can't be right. They asked me if I wanted to participate a while back and I said NO. They haven't asked me in a while, but I haven't given them permission to run the survey of my machine since the very first one one which I participated in, so maybe they got the hint finally.

they ask you for the big steam hardware survey that's annually and that ask for things like internet connection speed, installed applications, and large misc.. the typical hardware survey is monthly and automatic, they don't need a survey to collect that information as Prisoner849 pointed above.
 
That can't be right. They asked me if I wanted to participate a while back and I said NO. They haven't asked me in a while, but I haven't given them permission to run the survey of my machine since the very first one one which I participated in, so maybe they got the hint finally.

In other news, public votes are only decided by people that attend the polls. Which is clearly bullshit by this logic, because not everyone went.
 
Interesting that they don't itemize SteamOS in the Linux group. I suspect if it were doing well, they would be happy to provide the number.
 
I didn't realize how many people were running dual processor setups... 46.87%...

SEEMS LEGIT!

Keep in mind a lot of laptops, even the newer skylake ones with i7s in them, are still dualcore.
 
The people still using 2 cpu's are mostly likely using cpu graphics also.
 
It certainly gives perspective. I was sitting here starting to fret about how badly out of date my machine is. I actually discovered my old C2D sitting in its box with the fan and was laughing with my wife about when that was top of the line and how I didn't even know if I could find a motherboard for it right now if I wanted to do something with it.
 
The STEAM surveys are bullshit, because they don't include everyone. I have maybe been surveyed ONCE this year, and I may not actually have been. How exactly can that result in reliable OS statistics? I'm sorry, but this is no way to do surveys. This has been a problem for a long time, and I don't see VALVe improving this.

It's not like I even declined to be surveyed, I just haven't been presented with it, and there are lots and lots of people that report the same.


Statistics are not your strong suit, that was the only thing I got from your post.
 
Keep in mind a lot of laptops, even the newer skylake ones with i7s in them, are still dualcore.

I think he thought it meant dual processors, as in 2 CPU's on one motherboard. Do i3's with HT show up as dual cores or quad cores? I would assume dual cores.
 
Every time these charts come out, it still surprises me just how many people are on dual core CPU's.

I mean, I know the general population is different than the people who hang out in hardware enthusiast forums, but this is Steam! These are gamers! These are supposed to be OUR people! :p
 
The STEAM surveys are bullshit, because they don't include everyone. I have maybe been surveyed ONCE this year, and I may not actually have been. How exactly can that result in reliable OS statistics? I'm sorry, but this is no way to do surveys. This has been a problem for a long time, and I don't see VALVe improving this.

It's not like I even declined to be surveyed, I just haven't been presented with it, and there are lots and lots of people that report the same.

Statistics are not your strong suit, that was the only thing I got from your post.

Exactly.

Inference statistics uses a random sample. If that sample is large enough and sufficiently random, you can use it to infer things on the whole population.

This is how statistical sampling always work, and depending on your sample size, you can predict 95%, 99%, or even higher accuracy. They don't NEED to sample even anywhere near every user every month to get extremely accurate results. Do you really think the pre-election polls sample the whole population? Do you think companies that make products requiring destructive testing, test 100% of what they produce?

The only issue I find with the Steam survey is that people can opt out. It could wind up being inaccurate if there is any link what so ever between the type of people most likely to opt out, and the type of hardware they have. I think one could make an argument for that those who opt in are generally prouder of their hardware, and thus more likely to love the change to be a part of the survey, whereas those who opt out aren't. This could skew results towards the higher end.
 
I think he thought it meant dual processors, as in 2 CPU's on one motherboard. Do i3's with HT show up as dual cores or quad cores? I would assume dual cores.

Steam survey counts physical cores, my Xeon doesn't show up as an 8 core processor. So an i3 would be see as dual.

Despite pointing out the laptops with i5s and i7s still being dualcore, the fact that it's still such a high percentage does surprise me. That means, potentially, a lot of people are still on Core2Duos as well.
 
What I meant is what Strelok caught onto, Physical CPU's and Cores are two different things. In the IT world, 2 Physical CPU's would mean a dual Xeon processor build and almost 50 percent using dual Xeon processors just seems funny to me. They really need to change their terminology to "Physical Cores." A Core i3/i5/i7 is a single processor with multiple cores.
 
What I meant is what Strelok caught onto, Physical CPU's and Cores are two different things. In the IT world, 2 Physical CPU's would mean a dual Xeon processor build and almost 50 percent using dual Xeon processors just seems funny to me. They really need to change their terminology to "Physical Cores." A Core i3/i5/i7 is a single processor with multiple cores.

You're confusing 'physical' with 'discrete'

Two cores on a single die is still two physical processors. It wouldn't show up any differently to the OS than a dual socketed machine running a single core CPU in each.
 
Or, because we have the means, we could survey EVERYONE. Oh no! Haven forbid we actually get total or near total stats!



Exactly.

Inference statistics uses a random sample. If that sample is large enough and sufficiently random, you can use it to infer things on the whole population.

This is how statistical sampling always work, and depending on your sample size, you can predict 95%, 99%, or even higher accuracy. They don't NEED to sample even anywhere near every user every month to get extremely accurate results. Do you really think the pre-election polls sample the whole population? Do you think companies that make products requiring destructive testing, test 100% of what they produce?

The only issue I find with the Steam survey is that people can opt out. It could wind up being inaccurate if there is any link what so ever between the type of people most likely to opt out, and the type of hardware they have. I think one could make an argument for that those who opt in are generally prouder of their hardware, and thus more likely to love the change to be a part of the survey, whereas those who opt out aren't. This could skew results towards the higher end.
 
Or, because we have the means, we could survey EVERYONE. Oh no! Haven forbid we actually get total or near total stats!

Well, you have to balance the benefits vs the drawbacks.

In most cases, the drawbacks is that it is difficult to collect and process all that data, with Steam, this obviously isn't the case.

They might worry that if they are constantly pestering everyone to include them in the survey, that more people will opt out, and there may be a pattern to their opting out, upsetting the data. Since there are 65 million regular users on Steam, even if you sample 0.01% of the Steam population, you get a HUGE sample size and are - using inferential statistics going to get accuracy in the upper 99.99% area somewhere.

The risk to reward just isn't worth constant sampling of all users.
 
The typical reason why small bodies of the populous are polled is due to logistically being unable to poll everyone.

While your argument is valid, of exhausting the interest of the population, they could do it quarterly, or once half a year. Even quarterly this would be way less frequent to bother anyone. I do agree once a month would get in the way of people though.

One thing I am confident of though, is that these stats really can only be trusted so much. I understand there are "traditional" ways of doing stats, but I honestly don't believe they're applicable here, because they have the logistically ability to poll _LITERALLY EVERYONE_, unlike any other in-person polling mechanism.



Well, you have to balance the benefits vs the drawbacks.

In most cases, the drawbacks is that it is difficult to collect and process all that data, with Steam, this obviously isn't the case.

They might worry that if they are constantly pestering everyone to include them in the survey, that more people will opt out, and there may be a pattern to their opting out, upsetting the data. Since there are 65 million regular users on Steam, even if you sample 0.01% of the Steam population, you get a HUGE sample size and are - using inferential statistics going to get accuracy in the upper 99.99% area somewhere.

The risk to reward just isn't worth constant sampling of all users.
 
The typical reason why small bodies of the populous are polled is due to logistically being unable to poll everyone.

While your argument is valid, of exhausting the interest of the population, they could do it quarterly, or once half a year. Even quarterly this would be way less frequent to bother anyone. I do agree once a month would get in the way of people though.

One thing I am confident of though, is that these stats really can only be trusted so much. I understand there are "traditional" ways of doing stats, but I honestly don't believe they're applicable here, because they have the logistically ability to poll _LITERALLY EVERYONE_, unlike any other in-person polling mechanism.


It would be interesting to find out how they actually do their polling.

Someone previously in this thread suggested that it has only come when they have reinstalled windows or done major hardware upgrades.

Maybe the client looks for hardware changes, and only asks you to submit if there have been changes? This could make sense. I mean, if there have been no hardware changes, there is no need to resubmit data. They already know what you have.
 
  • Like
Reactions: Void
like this
Part of the problem is the lack of communication from VALVe. Often you will get prompted for the survey if you install a fresh OS. It's not always about hardware changes.

I'm particularly bothered by how they do stats because I'm confident they under represent Linux usage, by enough to matter. Their numbers have gone down, yet demand is visibly higher by a lot. I've had more people ask me about gaming on Linux than ever before, and I see online increasing volume of people asking how to get into Linux gaming

The stats are bogus.


It would be interesting to find out how they actually do their polling.

Someone previously in this thread suggested that it has only come when they have reinstalled windows or done major hardware upgrades.

Maybe the client looks for hardware changes, and only asks you to submit if there have been changes? This could make sense. I mean, if there have been no hardware changes, there is no need to resubmit data. They already know what you have.
 
Keep in mind a lot of laptops, even the newer skylake ones with i7s in them, are still dualcore.
Does steam count physical cores, or thread count?
Not that it means much, for example my AMD laptop has a 4 core CPU, but any 2-core celeron would destroy it.
 
Back
Top