1. 21

  2. 6

    Is this entering a competition for the most misleading graph? It shows the total number increasing by around 2%, yet makes it appear to be an increase of several orders of magnitude.

    1. 8

      I thought that having the X axis very clearly labeled with those seven figure numbers would avoid any chance of people misreading it. I mostly outsourced that decision to Observable Plot, which I think makes good choices most of the time.

      I added another chart to the notebook just now which plots from zero - unsurprisingly it’s a straight horizontal line, which I don’t think is nearly as useful.


      1. 1

        Probably needs more data before the chart itself is more interesting than the technique.

        1. 2

          Yeah that’s definitely true - I’m much more excited about how this works than the actual data it’s collected so far.

          I’m hoping there will be some interesting shapes starting to emerge in a few days time.

    2. 1

      Just a hint, I know you’re not scraping instances themselves, but if you did plan to, please note that a lot of them have that explicitly forbidden in their rules pages (in addition to robots entry). I know it may not mean much but people are sensitive about it.

      1. 2

        Yeah that’s one of the reasons I’m only working with the instances.json data from https://instances.social/ - which works on an opt-in basis from the instances themselves.

      2. 1

        Aaaaaand it broke 😅 (the plots are empty and the number of users jumped to over 70 million)

        1. 3

          Yeah there are some new fake spam instances! I’ll fix that in the morning, got an open issue about it here: https://github.com/simonw/scrape-instances-social/issues/4