Just sort by top of the month, all our top posts range from 1 week to 1 month old, with very few being from the past few days of record "online" numbers.

Anyone who knows how Reddit communities work in the same sense know that when community member numbers go up along with concurrent online users, the post vote counts go up proportionally.

It doesn't seem to make sense that with nearly 1900 users online at once the top post of the day has 214 points while posts from many weeks ago were getting more.

Did we change how "online user" is determined or is there a glitch?

edit:

Tested what people were saying about how "online users" is really just "open tabs"

https://hexbear.net/post/38707

  • asaharyev [he/him]
    ·
    4 years ago

    I imagine a few things are happening:

    • "online users" algorithm changed
    • more people online with no account
    • enthusiasm for the site has dampened somewhat, so people aren't just upvoting everything in the same way

    but the most likely is simply:

    • more users = more libs = worse posts :)
    • 90u9y8gb9t86vytv97g [they/them]
      hexagon
      ·
      edit-2
      4 years ago

      Yeah maybe an extension of how long ago a person had to have visited the site to count as "online".

      For Reddit I think it's around 20–30 minutes, for here they may have extended it to hours or the day.

      Do we know if this site even counts non-account visits as online users? That wouldn't make sense to me.

      It would be nice to have a public traffic page like Reddit has for their mods, I'd love to see the site growing over time. Would be encouraging.

    • buh [she/her]
      ·
      4 years ago

      Actively opening and closing tabs to maintain an online count of 1917

    • 90u9y8gb9t86vytv97g [they/them]
      hexagon
      ·
      4 years ago

      Do you know if this is how "online users" has been calculated over the life of the site?

      So when we had ~800 users online for weeks, it was really like 250 people?

        • 90u9y8gb9t86vytv97g [they/them]
          hexagon
          ·
          4 years ago

          Yeah I didn't think so, only that something fucky was going on with the system.

          No site with nearly 2000 people online should be so slow with posts.

          Having a full transparency post about the traffic we get would be appreciated, and make it a lot better to track how outreach on Reddit effects activity and traffic here.

        • sysgen [none/use name,they/them]
          ·
          edit-2
          4 years ago

          The code was this way since before we even started chapo.chat, so not really no. Lemmy counts websockets and the front-end opens one websocket per tab as it comes, and I even think someone noticed this on the first time we fired up the dev server.

            • sysgen [none/use name,they/them]
              ·
              4 years ago

              I'd have to do some analysis, and I don't have access to the prod server, but if I had to guess I'd say a combination of new users as well as the debates causing everyone to open a tab for the thread. Or maybe something else, but the issue for the online count being by tab was there from the get-go.

            • garbology [he/him]
              ·
              4 years ago

              My theory has always been that we got a lot of traffic from the low-engagement lurkers we lost with the sub ban who re-discovered the new chapo via the watermarked memes a comrade is spreading around on reddit or where ever. People who don't post and might not even make an account, but just browse/consume.

    • sysgen [none/use name,they/them]
      ·
      4 years ago

      Hmm I've got an idea how it could be fixed, been a while since I've touched the code but I could attack it this weekend when I'm done with my math exams lol

  • Awoo [she/her]
    arrow-down
    1
    ·
    4 years ago

    Comments are going up drastically. People are probably visiting fewer posts than before because their session is spent in comment sections longer than visibly looking at and voting on posts.

    If you think of a person's session as having an average length, let's say 1 hour to be convenient. If someone was spending 30 minutes viewing posts and 30 minutes in comment sections previously but is now spending 20 minutes viewing posts but 40 minutes viewing comment sections -- the result you will get is fewer votes on posts and significantly more comments.

    Here's the comment data per day.

    The total posts are trending upwards too, but not as strongly as the comments. My guess is people are considerably enjoying the comment sections here and getting sucked into them more than over on reddit. User behaviour is still adapting and changing to adapt to seeing ChapoChat as its own separate thing to reddit so there are currently changing habits in the userbase. It will probably take a long time to settle especially with an actively changing site due to the fast dev work.

    I'm speculating of course but if you keep everyone's session length the same and change how much of that session is spent within each area of the site you would end up changing comment totals, post total and vote totals in a variety of ways and since post totals aren't going up as quickly as comments this is what I think. People just love how friendly and pleasant ChapoChat is compared to reddit, getting a message notification is a pleasing thing instead of something the users dread like on reddit.

    • 90u9y8gb9t86vytv97g [they/them]
      hexagon
      arrow-down
      3
      ·
      edit-2
      4 years ago

      All of this might be true, but the weird inflated number of online users comes because we're counting every tab as someone online, as established elsewhere in this thread.

      • Awoo [she/her]
        ·
        4 years ago

        Oh yeah absolutely. I had noted it and noticed that the jump in online users had not also corresponded with a jump in comments or posts. It's clearly just a change in the backend caused by ongoing development that occurred on the 7th.

        Things look very healthy besides that bug though. Everything continues to trend upwards nicely at a steady and comfortable pace. New user gain is accelerating slowly too. Ignore everything before the spike, data in that period was clearly warped by mass account creation for the purpose of wrecking and you can clearly see the exact date wrecker nonsense ceased. Average new users per day is trending upwards at a steady pace. The site will double in size in a year if no change in user gain occurs but I would bet on a larger gain in size due to acceleration. The more users we have, the faster we will grow. Probably 3 times the size we are now or more in 12 months time. By 24 months the site could be back at subreddit numbers. By 36 months who knows what.

  • emizeko [they/them]
    ·
    edit-2
    4 years ago

    i agree, it sure feels like the online user count isn't right.

    • 90u9y8gb9t86vytv97g [they/them]
      hexagon
      arrow-down
      1
      ·
      4 years ago

      It feels like when subreddits used to edit their CSS to put a "100," in front of their user counts to give the appearance of a much larger community.

      So quiet, especially considering only a few weeks ago we had ~800 users online on average and posting was high.

  • ThePeoplesGuillotine [he/him]
    ·
    4 years ago

    You're right that the user count is inflated via tabs but we have more upvotes on comments and posts in the last 2 weeks than we did the previous 2 weeks from that. Posts: 250k vs 219k Comments: 755k vs 645k

    14 days into Oct compared to 14 days into Sept Posts: 256k vs 155k Comments: 760k vs 423k

    Similarly for pure post/comment count: Comments: 72k vs 42k Posts: 6.2k vs 5.2k

    • 90u9y8gb9t86vytv97g [they/them]
      hexagon
      arrow-down
      3
      ·
      edit-2
      4 years ago

      That and the site is counting users at least 5x over by counting "tabs open" not individual IP addresses.

      Just tested it by opening like 150 tabs: https://hexbear.net/post/38707

      This means we probably only have 300 people online, not 1700.

        • qublics [they/them,she/her]
          ·
          edit-2
          4 years ago

          Although if you closed the browser then continued session those tabs would not be active until reloading.

  • diode [none/use name]
    ·
    4 years ago

    I'm still waiting for a person that runs 100s of tabs to confess to their bit.

    • 90u9y8gb9t86vytv97g [they/them]
      hexagon
      arrow-down
      1
      ·
      4 years ago

      Counting separate tabs as users would be hilariously dumb, I'd count as like 14 people right now.

      spoiler

      Yes I have a tab problem, no I will not seek help.

      • diode [none/use name]
        ·
        4 years ago

        It counts websocket connections, so that's exactly how it's counted.

          • 90u9y8gb9t86vytv97g [they/them]
            hexagon
            ·
            4 years ago

            Which means we really have "1867 tabs open" not "online users".

            And for an average of maybe 6 tabs open, that's ~310 people online.

            Oof

        • 90u9y8gb9t86vytv97g [they/them]
          hexagon
          ·
          edit-2
          4 years ago

          Wait actually? Has it done that the entire existence of the site?

          That means we likely have at most 30% the users reported as online, if not far less.

          Even at that overestimate it puts us at ~622 online...

          • diode [none/use name]
            ·
            4 years ago

            From what I have seen in these user surge threads, that's the way it has always been.

            • 90u9y8gb9t86vytv97g [they/them]
              hexagon
              ·
              edit-2
              4 years ago

              Yikes

              So when we were consistently celebrating 800 people online we might have barely had 200 or so.

              Why why why would you not just count IP addresses as individuals.

              • emizeko [they/them]
                ·
                edit-2
                4 years ago

                just count IP addresses as individuals

                that would undercount, but probably better than overcounting

                • 90u9y8gb9t86vytv97g [they/them]
                  hexagon
                  ·
                  4 years ago

                  Yes, but it would undercount by a far smaller amount than we're currently overcounting apparently.

                  I regularly visit and open like 10 tabs of interesting threads I don't want to lose when the site auto-updates, but an IP address might have what, four people on it max, disregarding universities (which are currently closed anyway).

              • Ryaina [she/her]
                ·
                4 years ago

                until a recent merge, I had a patch in place that did exactly this, for the vast majority of the site's existence the online count has been IP based.

  • Fakename_Bill [he/him]
    arrow-down
    1
    ·
    4 years ago

    On Reddit, you can upvote or downvote posts without joining the subreddit. Unless I'm mistaken, you can't upvote or downvote here unless you make an account. So an increase in active users alone can't influence upvote numbers.

    • cilantrofellow [any]
      ·
      edit-2
      4 years ago

      When you fix it, make sure you change the metric to “unique users online” or some shit so people don’t think we had a huge drop off for some reason

  • D61 [any]
    ·
    4 years ago

    From the comments talking about how upvoting is for reddit libs, I'm worried about being gulaged.