• Atemu@lemmy.ml
    link
    fedilink
    arrow-up
    92
    arrow-down
    16
    ·
    1 year ago

    What a terrible graph. That “huge” spike is a mere 0.5% increase. That might as well be noise.

    Don’t believe any graph whose y-axis starts at any value but 0 people.

    • jsdz@lemmy.ml
      link
      fedilink
      arrow-up
      34
      ·
      edit-2
      1 year ago

      Don’t believe any graph whose y-axis starts at any value but 0 people.

      This one is pretty bad but that is definitely not the right lesson to take from it. The one thing it does show us is that approximately 20k extra new users suddenly showed up compared to the trend, and that would be much more difficult to see if the relevant axis did start at zero. The bigger problem is that it shows too short a time span. It’s not clear how unusual this event was, or if it happens every week.

      The other weird thing is that bottom-right axis does start at zero for some reason. I’m guessing it might somehow be trying to indicate “toots” specifically made by those new users? But that’s not how it’s labelled and it seems unlikely they could have that data.

      • Atemu@lemmy.ml
        link
        fedilink
        arrow-up
        11
        arrow-down
        1
        ·
        1 year ago

        Indeed the right sides of the graph start at 0. The left side does not.

        Note that 2000/h (10^3) aren’t all that significant when there’s already 14000000 (10^7) users present.

        • Rob Bos@lemmy.ca
          link
          fedilink
          arrow-up
          4
          ·
          edit-2
          1 year ago

          Well. It is and it isn’t. It’s per hour. So per day (24^1) that’s 48000 (48*10^3) and per year it’s 17532000 (1.7*10^7). That adds up pretty fast, a 100% increase in the full year.

          Plus, hey, new friends!

    • Kichae@kbin.social
      link
      fedilink
      arrow-up
      13
      arrow-down
      1
      ·
      1 year ago

      There’s nothing wrong with graphs whose y axies don’t start at zero. They can be used to misdirect people, but if you’re capable of actually seeing the numbers in the axes and doing a little bit of thought, they tell you exactly what one that starts at zero does.

      Plus, the opaque spike is shown on the secondary y axis, which does start at 0. It’s the translucent layer that’s mapped to the primary axis.

      • Atemu@lemmy.ml
        link
        fedilink
        arrow-up
        1
        ·
        1 year ago

        I can’t remember the last time I saw a graph starting at a non-zero value where it showed anything other than noise whereas they almost always skew my initial impression of the data. If there’s no point in doing it but a major downside, I see no point in having them for any reason other than to mislead people.

      • Calavera@lemm.ee
        link
        fedilink
        arrow-up
        5
        ·
        1 year ago

        If you talk about a new wave of users, then the number of users is also important, really important

    • Dark Arc@social.packetloss.gg
      link
      fedilink
      English
      arrow-up
      6
      arrow-down
      1
      ·
      edit-2
      1 year ago

      This is a bad take in this case. This graph is of total population, not if signups. It effectively is zeroed.

      Sure it’s a very small increase relative to the total but relative to recent history this is very significant.

      Edit: the bigger issue from a data interpretation perspective is the date range sampled is small.

    • OpenStars@kbin.social
      link
      fedilink
      arrow-up
      2
      arrow-down
      2
      ·
      1 year ago

      Plus the second graph shows the average number of instances went down compared to yesterday, which was itself down further from the day before.

      This “wave” is looking mighty sus.

      • Rob Bos@lemmy.ca
        link
        fedilink
        arrow-up
        3
        ·
        edit-2
        1 year ago

        Yeah, a couple days of temporary spike does not a wave make.

        Mastodon (and the Fediverse) tends to see “scalloped” growth: big increases, followed by gradual declines. Every time Musk does something dumb, you see days or weeks of increased signups. Then the new users fall off, and they become inactive. Usually, it stabilizes a little higher than the last wave.

        The waves come in, and the tide rises. The weather passes over, but the climate stays stable (or increases).

        If Twitter collapses, then the tsunami arrives. :P