Announcement

Collapse
No announcement yet.

Steam's Hardware Survey Shows Not Much For Linux

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #11
    Originally posted by liamdawe View Post
    I disagree, i believe all surveys like this are heavily flawed (since if you only survey such a small amount you could easily miss out most of the Linux users just because it's such a small group), especially Steam's
    Then you should take remedial statistics at your local community college.

    If a group is 5% of a population, then a meaningful sampling of that population will show that group as 5% of the polled people. Simple as that. Even where other demographical information comes into play, you'll see those lines show up in the sampling. Saying, if Linux was 50% of one region and 1% the others, the sampling would (accurately) show the 50% of users sampled in that region, 1% elsewhere, and the X% average over the entire population. That's the first core piece to understanding statistics and even political science. You not agreeing or disbelieving it does not change how mathematical sampling works.

    This kind of sampling accurately finds far far more minute demographics in far smaller population pools than you have with Linux users among Steam gamers, so it's pretty silly to think that Linux users are so magical and special that math doesn't apply to them.

    It's accurate to question _why_ the numbers are the way they are and question certain conclusions, maybe claim that there are more Linux gamers who simply don't fall into the demographics sampled by Steam (which exclude Linux gamers that reject Steam or many Linux users who game on Windows due to the lackluster Linux game library size), but to claim that the sampling approach is invalid is just wrong.

    Comment


    • #12
      Originally posted by mendieta View Post
      Yeah, it seems crazy, is it really a Survey? Meaning do they ask people? Why not collect statistics? For people playing online, they could aggregate the number of hours each player stays online, per operating system. Another clear cut statistic would be number of downloads per platform. Oh well.
      They ask people if they want their statistic submitted. As in they won't ask a Wine user if he wants to be represented as a Linux user (unfortunate as it may be).

      Originally posted by elanthis View Post
      Then you should take remedial statistics at your local community college.

      If a group is 5% of a population, then a meaningful sampling of that population will show that group as 5% of the polled people. Simple as that. Even where other demographical information comes into play, you'll see those lines show up in the sampling. Saying, if Linux was 50% of one region and 1% the others, the sampling would (accurately) show the 50% of users sampled in that region, 1% elsewhere, and the X% average over the entire population. That's the first core piece to understanding statistics and even political science. You not agreeing or disbelieving it does not change how mathematical sampling works.

      This kind of sampling accurately finds far far more minute demographics in far smaller population pools than you have with Linux users among Steam gamers, so it's pretty silly to think that Linux users are so magical and special that math doesn't apply to them.

      It's accurate to question _why_ the numbers are the way they are and question certain conclusions, maybe claim that there are more Linux gamers who simply don't fall into the demographics sampled by Steam (which exclude Linux gamers that reject Steam or many Linux users who game on Windows due to the lackluster Linux game library size), but to claim that the sampling approach is invalid is just wrong.
      Do you know the sample size? Do you know the population size? Without at least the first one, it's impossible to tell how accurate the percentages are. They'll be in the ballpark, yes, but the same statistics course says that you cannot tell the proportions exactly, unless you ask absolutely everyone. The proportions can be anywhere from around 0.01% to 99.9%, it's just not very likely to be so. Now when you do know the sample size, you can start calculating around where it is likely that the real proportion is (it's very unlikely that the numbers given here are the actual true numbers, they're just somewhere close).

      Comment


      • #13
        Originally posted by ua=42 View Post
        For gaming I'm just about to build a new computer that I plan to set up as a duel boot.
        I'm hoping that I'll be able to do most of my gaming and stuff in linux and not have to go to windows except for the occasional game.


        My build:
        fx 8350
        HD 7870

        Hey AMD, because of your open source driver efforts (especially power-save) made me decide to buy. :-D
        If you want to play, go nvidia. I'm sorry, their open efforts are great, but you'll regret going AMD (for gpu, for cpu it's a great pick). I know I did.

        Comment


        • #14
          Funny to see that Arch Linux (Linux 64-Bit) is the most popular non-Ubuntu distro and even has more Steam users than Mint. Not to bad for a distro that doesn't aim to be userfriendly and popular. We already know that Other is 100% Linux. When Steam showed every distro other was left with 0.02%. http://www.gamingonlinux.com/article...y-distros.1845

          Comment


          • #15
            I guess this time there won't be the excuse that hordes of Linux users are changing their user agent strings.

            Comment


            • #16
              It would be actual nice to see some raw numbers or at least a more detailed statistic, like:
              % of the primary test group, relative to all users (accepted + rejected surveys) {Win|Mac|Lin}
              % of rejected surveys, relative to the primary test group {Win|Mac|Lin}

              And add wine to the list, so we know that the wine users are not actually counted as Windows.

              Comment


              • #17
                Originally posted by [Knuckles] View Post
                If you want to play, go nvidia. I'm sorry, their open efforts are great, but you'll regret going AMD (for gpu, for cpu it's a great pick). I know I did.
                Well, i have a fx-8320 and a radeon 7950 gpu and I have not regretted it a bit. I am gaming L4D2 and TF2 and Dota2 almost daily...Everything works perfect!

                Comment


                • #18
                  if the survey size was too small, then this would manifest as noise in the results. the percentages would fluctuate wildly from month to month. In fact the month to month variation is fairly smooth, so i'd guess they are good to better than 0.1%.

                  if there was a systematic error, it would be hard to detect. for example if there selection method has a bias (perhaps a bug that makes it less likely to record on some platforms). I can't imagine why steam would want to make linux looks small.

                  whether the steam survey is a good measure of linux desktop share is a very different question.

                  It would be also interesting to know what they are trying to actually count. Count of users, hours of game play, hours of the client running? For example it could be that the first time you connect in each month, there is a 1 in 10000 chance you will be surveyed. Or it could be for every hour of game play there is a 1 in 100000 chance you will be surveyed. Or maybe people who purchase more are more likely to be surveyed.

                  Comment


                  • #19
                    Originally posted by FateousMaximous View Post
                    I think this is a catch 22 sort of situation. Without enough games there will be a lot of holdouts that will stick to windows because there are more games there, but without enough gamers there isn't much of an incentive to make games for linux. As more Valve games get ported over for the steam box there will probably be more people switching over, and if the steam box does well there will be more devs making Linux games for it.
                    Most people wont switch over.
                    Most people have Windows, and they are okay with that.
                    Most people don't know about Linux, and don't care about Linux and see no reason to switch to Linux.

                    Windows works for them.
                    They have no compelling reason to switch to Linux.
                    Linux have less game and in most cases worse gaming performance.
                    Switching is uncomfortable and requires them to learn new stuff.

                    Even if Linux had all the games Windows had, most people would still not switch.
                    Even if Linux was faster than Windows, people still wouldn't switch.

                    Comment


                    • #20
                      Originally posted by Dukenukemx View Post
                      Directx9 state tracker would fix that. As great as Valve's games are, they aren't enough to get people to switch alone.
                      You could port the entire Steam library to Linux and it wouldn't change much. A small handful might switch, but an overwhelming majority would not. What is the incentive to leave Windows? What does Linux do better such that it would tempt somebody to come over?

                      Comment

                      Working...
                      X