📊 I Started Collecting Torn Forum Data — Here's What the First ~22.5 Hours Showed
A while ago I started a small experiment because I was curious about something:
Can you actually measure what makes a Torn forum thread gain attention?
Instead of manually checking the forums, I set up a small data collection system that automatically records forum activity every 10 minutes.
Since the collector was started, I've been continuously storing snapshots of the forum and comparing them against previous snapshots.
For each thread, I'm currently tracking things such as:
-
Thread ID
-
Thread title
-
Forum section
-
Number of views
-
Number of posts
-
Author
-
Last poster
-
Thread creation time
-
Last activity time
-
Poll / locked / sticky status
-
Exact time the snapshot was collected
The data is stored locally and each new snapshot is compared with the previous ones to calculate changes rather than just looking at the current numbers.
And this is where it started getting interesting.
🔥 The biggest 10-minute spike
One of the largest spikes recorded so far was:
Banning of third party casinos
+1,731 views
+89 posts
in a single 10-minute interval.
Across the entire observation period, the same thread gained approximately:
+3,991 views
+361 posts
That's a pretty significant difference compared with most other threads in the dataset.
💬 Some surprisingly discussion-heavy bursts
A few examples from the 10-minute comparisons:
BUM FIGHTS - E Tu Whanau Production
+40 views / +23 posts
Rl casino
+56 views / +11 posts
These are interesting because the amount of discussion is disproportionately high compared with the number of new views.
🧠 What I'm actually trying to find
I'm not really interested in simply finding the "most popular thread."
I'm trying to identify patterns in forum behaviour.
For example:
Reader behaviour
High view growth + little/no posting.
Discussion behaviour
Moderate view growth + unusually high posting.
Explosive behaviour
Large increases in both views and posts within a short period.
I'm also recording the timestamps, so eventually I should be able to compare activity by hour and see whether certain periods consistently produce higher activity.
🔬 The experiment isn't finished
This is only the first ~22.5 hours of data.
The collector is still running, so the dataset will continue growing.
The idea is to eventually have enough snapshots to answer questions like:
-
What actually causes a thread to suddenly spike?
-
Which topics generate discussion rather than just views?
-
How long does a thread usually stay active?
-
Do giveaways attract readers or actual participants?
-
Does controversial content behave differently from normal discussion?
-
Are there specific hours where forum activity consistently increases?
-
Can early activity be used to identify a thread that's about to become popular?
Right now, these are observations, not conclusions.
I'm going to keep collecting the data and see whether these patterns survive as the dataset gets much larger.
22 hours is interesting. A few days should be much more interesting. 👀📈
So far I have:
78 clean snapshots
~22.5 hours of observation
100 threads per snapshot
13,500+ positive view changes
1,900+ positive post changes
working on some improvements and open for any suggestion or tips and will do a 1 week version of this data collecting thing
views based on TCT (torn city time)
highest overall activity