I was looking at statcounter when I came across that Linux once held 8% of the desktop marketshare worldwide. Since when did it get this far in popularity???

  • thingsiplay@lemmy.ml
    link
    fedilink
    arrow-up
    2
    ·
    2 小时前

    the charts are all useless, especially the steam chart

    I do not think they are useless.

    the methodology makes them highly variable month to month

    This is standard methodology by extrapolating some sample data. The variations based on this methodology is basically noise difference. The variations on Steam in example that happen from time to time is not because of this methodology, but because something happens usually. Those variations would probably still happen if very single user was tracked.

    It is scientifically accurate and correct to extrapolate sample data to represent the entire userbase and to do predictions. The only open question we have is how many sample data are used. Steam needs to show the number of “participants”. The random nature should make the results non biased.

    • Mactan@lemmy.ml
      link
      fedilink
      arrow-up
      3
      ·
      2 小时前

      what are people supposed to think when each item fluctuates +/- 5-15% month over month? doesn’t it need to be aggregated over a really long span of time to get a sense of some kind of an average?

      • thingsiplay@lemmy.ml
        link
        fedilink
        arrow-up
        3
        ·
        50 分钟前

        The data fluctuating does not mean its because of a flawed methodology. In example China has a huge impact. Let’s say they make changes to internet cafes, which can have a huge impact of the overall data in Steam. And if the methodology changed to account for every user, it still could have month over month changes.

        Collecting a sample data and extrapolating it to represetn the entire userbase is standard procedure. Steam doesn’t do something special different here. What do you suggest instead? Any suggestion to collect data as opt-out is a non starter discussion. Other than that, this is the best we have. At least in Steam.

        You first have to proof your argument that the methodology is the problem here (as this is your suggestion, not mine).