Report on thread 130329: Thread life and activity

Report on thread 130329: Thread life and activity

With the help of a good data sample we are now finally able to asses and make some basic conclusions about thread and board activity. For example, a common question (and one that fortuitously applies to the thread discussed here) asks what can explain a thread moving and/or 404ing quicker than normal? Is it increased posting? Increased board activity? Potentially, it could be a perfect storm of both. But in the case of thread 130329, we can conclude that it was the former, and not the latter. This report will explain, using statistical analysis of data collected over the past 4 months, how and why we can make that conclusion.

First let’s briefly talk about the bump limit, which is the threshold at which the thread can no longer be bumped, or pushed to the top. On /hr/, that limit is 300 posts. This means that when 300 replies have been posted in a thread, the thread no longer bumps no matter if it’s the most popular thread on the board; it slowly gets pushed to page 11 as newer threads make their way to the top. As we are the most active thread on the board, we reach the bump limit in a matter of hours, unlike other threads on the board that can take days or weeks. This is why if you’re looking at the catalog, you see the same threads up for months and months (the most dated thread recorded lasted for over 296 days!). Why is this important to know? As you might be able to guess, the length of time it takes a thread to hit the bump limit offers us 1) a good prediction of how long the thread will be around and 2) a good measure of thread activity.

In addition to the bump limit time measure, we’ll be looking at two other statistics: First, the total number of posts on /hr/ from the point our thread is created to the time it reaches 300 posts. This is important because it can tell us how active the board is as well as how many posts in /hr/ come from our thread. Second is the number of threads created on /hr/ during the 6 days when the thread was active. This number also gives us an idea of how active the board was, except in terms of threads made instead of all posts made. Analyzing these three statistics will tell us whether the thread was more active than usual, the board, or both.

Let’s take a look at the data first, and then we’ll explain it one by one: http://i.imgur.com/QCAMYmt.png

‘Hours to bump limit’ has already been explained in great detail. We see that while the average number of hours to hit 300 posts is 18, last thread it only took 8 hours. This means that the thread started dropping less than half a day after it began. In fact, 94% of the thread’s life was spent post-bump limit. So in summary, what we learn from this is that posting in the thread during the bump period was higher than usual, indicating more activity in the thread.

Next we look at the total number of posts on /hr/ from the point the thread was created to the time it reached 300 posts. We see that the average is 796 posts, while the number for this thread was 481. What this means is that in the period of time from when the thread was created to when it reached the bump limit, 62% of all posts in /hr/ were made just in our thread alone. In summary, what we learn from this is that board posting during the bump period for this thread was lower than usual, indicating less activity on /hr/.

Next we’ll look at the total number of threads posted in /hr/ from the point our thread was created to when it 404’d. We see that the average is 47, and that in the case of this thread, only 39 were made. Note that while this may not seem like a large difference, the two figures are still more than a standard deviation apart, which does in fact make the difference significant. In summary, what we learn from this is that board posting during the bump period for this thread was lower than usual, indicating less activity on /hr/.

And let’s look at one more statistic just for fun. The average number of posts in threads lasting 6 days is 1863. The number of posts in this thread? 2092. That equates to 12% more posts in the same amount of time. Obviously this tells us what we already know: the thread was more active than usual.

So what are we left with? How do we make this conclusion? We examined four statistics. Two were related to thread activity, both of which told us that the thread was more active than usual. The other two were related to board activity, both of which told us /hr/ was less active than usual. Put like this, it seems easy to conclude that the thread moved and 404’d quicker than normal because of high thread activity and not high board activity.

Here are some questions we’re left with: Can we predict how long a thread will last based on how quickly the bump limit is reached? If so, can we take measures to make the threads last longer? And finally, would we want to? I believe the answer to the first question is yes, we can. And in fact, I already have, to some extent. For example, when the thread in question hit the bump limit in 8 hours, I correctly made the prediction that the thread would last 6 days. This current thread hit the limit in a range similar to threads that lasted 7 days, so my prediction is 7 days, and I’m excited to be proven right or wrong [EDIT: Thread 130406 lasted 8 days, one day longer than my prediction]. The answer to the next question I believe is also yes – it is possible to prolong thread life by maintaining a delicate balance of posting within the first 300 posts. Too much activity will cause the thread to reach the limit quicker. But too little activity might also be a bad thing, in that no bumps will also cause the thread to drop to the point where, when we do hit the bump limit, the remainder of the thread’s life will be much shorter. However, I find it hard to imagine a scenario in which these threads suffer from inactivity. Finally is the question, do we want longer threads? I don’t have an answer to that, but there are some arguments on both sides. Some say that longer threads correlate with better quality, more enriching discussion, and less shitposting. Others contend that longer threads (like the 4000+ thread) are laggy and really do nothing to stop shitposting. Something else to consider is that longer threads tend to hit the image limit, which prevents any other images from being posted in the thread.

Tl;dr
Statistic: hours to bump limit.
Description: tells us how active the thread is.
Conclusion: the thread was more active than normal.

Statistic: average number of posts controlling for thread life.
Description: tells us how active the thread is.
Conclusion: the thread was more active than normal.

Statistic: total board posts from thread creation to bump limit.
Description: tells us how active the board is.
Conclusion: the board was less active than normal.

Statistic: total /hr/ threads posted from thread creation to 404.
Description: tells us how active the board is.
Conclusion: the board was less active than normal.

Summary: last thread’s increased thread activity, not board activity, caused it to move and 404 quicker than normal.