Support NeoGAF

[Grildon Tundy]

Edit: The Gaming Discussion version can be found here: https://www.neogaf.com/threads/neogaf-a-data-driven-view-pt-2-gaming-discussion-section.1640785/
And the Communities one here: https://www.neogaf.com/threads/neogaf-a-data-driven-view-pt-3-communities-section.1640829/

Original Post Below:

My friend and I are working on a business venture that requires learning how to parse website data, and I thought: why not start with an interesting subject that doesn't have much existing research? NeoGAF! Note that this is only a view of Off-Topic for now. Hopefully Gaming will come in the future.

Stay tuned for a trip down memory lane where we'll look at the most popular threads by month since 2018.

But for now, let's start with some highlights:

1. The average lifespan of an Off-Topic thread is 28 days. However, the median is drastically lower at 2 days.
   1. This means the majority of threads are short-lived compared to the ones that stick around and pull the average up
   2. If the thread makes it to one month, it's in the top 10% in terms of longevity
2. An average thread gets 61 comments; the median gets 29. Most threads die before seeing Page 2
3. There are anywhere from ~35-50 views for every comment: Who are all these lurkers?!? Come on out--we want to talk to you!

More findings:

1. There are around 215 new threads posted in Off-Topic each month, or about 2,500 per year
2. 380 Users created an Off-Topic thread so far in 2022, at an average of 3.7 threads apiece

GAF is a forum, so let's check out who's generating the most discussion.

Congratulations to ForAcademicPurposes : Out of 380 people, you're 2022's most "engaging" OP-to-date with 582 comments per thread on average! Special shoutout to the Top 10, too: OmegaSupreme , Amiga , Bartski , Represent. , S Star-Lord , ArtHands , near , K kikkis , & Draugoth ! (NOTE: OP's had to have submitted at least 2 threads in 2022 to be included in order to remove outliers).

The longest-running Off-Topic thread is: Stock-Age: Stocks, Options, and Dividends, oh my! Below you can see a list of other long-termers:

Finally, let's take a look at the most popular threads by month since 2018. Many of these are were umbrella topics for discussions that are still going on today ("What, are you listening too?", "Anyone seen any good movies lately?"). But others are snapshots in time:

April 2019 - Game of Thrones Season 8 Discussion
October 2019 - My Fucking Nutjob Neighbor
January 2020 - Mask Efficacy OT Wuhan!! Got You All In Check

That's all for now. Hoping to put something together for the Gaming section in the future.

EviLore , FYI, thought you might find some of this interesting.

 

[dr_octagon]

Is there a gif breakdown metric?

 

[LegendOfKage]

That's great! I know that for a while GAF was experiencing a lot of growth, but I haven't seen data regarding that lately. I was wondering recently if we've kind of hit a plateau when it comes to active users. I could be wrong, but it seems a bit slower to me lately.

This is all interesting data, though. Nice job!

 

[Amiga]

Movie/TV threads regularly get the most attention. Sports doesn't seem to get much traction for how popular it is in the mainstream.
politics is a minefield that cost many casualties so most are avoiding it or it would easily top the discussions, also tends to get ugly fast. the extreme passion it generates is why the biggest networks are based on politics.

 

[AJUMP23]

Posting a lot doesn’t make you the most engaging it just means you are always posting. I am super engaging with my dumb jokes and minimal post.

 

[bender]

 

[Chittagong]

Butts, tits, blondes, brunettes and redheads putting in work

 

[Grildon Tundy]

That's great! I know that for a while GAF was experiencing a lot of growth, but I haven't seen data regarding that lately. I was wondering recently if we've kind of hit a plateau when it comes to active users. I could be wrong, but it seems a bit slower to me lately.

This is all interesting data, though. Nice job!

I don't have user numbers, though you could infer them from the number of new posts per month. There was a huge drop when Politics got nuked. Seems pretty steady since.

Is there a gif breakdown metric?

I am working on getting down to comment-level detail, but unfortunately I don't think I can see which posts have gifs or not. I wanted to do something like "reactions received per gif posted" but probably can't

Posting a lot doesn’t make you the most engaging it just means you are always posting. I am super engaging with my dumb jokes and minimal post

I hear you. Thats why I measured comments per post and not numbers of posts.

 

[Maiden Voyage]

You should jazz up the data to make it more exciting.

 

[mortal]

This insight is pretty neat Grildon Tundy

 

[OmegaSupreme]

I have too much free time at work. That's my excuse.

 

[Grildon Tundy]

You should jazz up the data to make it more exciting.

Oh, I kept back the primo stuff. For $20, I'll tell you the best time to submit a thread for those sweet, sweet reaction scores.

Just kidding. I wanted to look at best timing, but the threads don't have submission timestamps

 

[jason10mm]

Open the politics area back up for a 24 hour then deleted free for all and watch those metrics EXPLODE

Like most forums, there are the plank owners and lots of casuals. How can you monetize the forum if its just a few hundred regulars?

Evilore: "I've found a wayyyyy"

The solution is obvious on youtube....outrage sells!

 

[Grildon Tundy]

Here's another way of looking at engagement--

Edit: I don't think it's the case the threads end because of these last posts. The intention was to estimate posting frequency with this as a proxy since I don't have comment-level detail data yet.

 

[BadBurger]

About to be 18 threads ended in 2022 bay-beee

 

[Little Mac]

What about a top 20 list ranking members based on their message to reaction score ratio?

 

[Kilau]

 

[Sub_Level]

AJUMP the kiss of death. You know your thread is fucked if he posts in it.

 

[Grildon Tundy]

What about a top 20 list ranking members based on their message to reaction score ratio?

I want to, and the data is grabbable, but the problem is I don't know how to automate going through each user like you can for thread info.

Edit: Actually, think I just figured it out. More to come.

 

[ForAcademicPurposes]

 

[Jennings]

 

[Con_Z_ǝdʇ]

Well done man! Very insightful. I hope you dig even deeper into the matter.

 

[Chittagong]

I want to, and the data is grabbable, but the problem is I don't know how to automate going through each user like you can for thread info.

Edit: Actually, think I just figured it out. More to come.

The challenge here is that reactions are a fairly recent thing, so if you include someone who has been a member for a decade before reactions were available, their reaction ratio will be much worse than for someone who joined after reactions were introduced.

 

[Doczu]

The challenge here is that reactions are a fairly recent thing, so if you include someone who has been a member for a decade before reactions were available, their reaction ratio will be much worse than for someone who joined after reactions were introduced.

So you're telling me to step up my game? Shit fam, say no more!

YOU GET A RECATION, YOU GET A REACTIO, YOU GET A REACTION. YOU WOULD GET TWO IF I COULD. REEEAAAAAACTIIIOOOONSSSSS!

 

[SJRB]

Look mom, I did it!

 

[jufonuk]

 

[Represent.]

My threads do be hitting hard I cant lie

 

[Star-Lord]

I don't know whether I should be proud or embarrassed.

 

[Maiden Voyage]

The challenge here is that reactions are a fairly recent thing, so if you include someone who has been a member for a decade before reactions were available, their reaction ratio will be much worse than for someone who joined after reactions were introduced.

Tell me more 2004.

 

[Grildon Tundy]

The challenge here is that reactions are a fairly recent thing, so if you include someone who has been a member for a decade before reactions were available, their reaction ratio will be much worse than for someone who joined after reactions were introduced.

Way ahead of you fam. Unfortunately I think the way to control for it is to cut out members who signed up prior to reactions being a thing. Unless evilore wants to give me access to the raw data lol

 

[jason10mm]

Here's another way of looking at engagement--

Edit: I don't think it's the case the threads end because of these last posts. It's probably a good way of estimating posting frequency since I don't have comment-level detail data yet.

In before the lock!!

 

[dr_octagon]

AJUMP23

 

[ahtlas7]

And the metric showing the collective IQ descent ascent over time?

 

[dr_octagon]

And the metric showing the collective IQ descent over time?

 

[InverseAgonist]

That's all for now. Hoping to put something together for the Gaming section in the future.

Just wanted to say that's a wonderful analysis and dataset to start! Very interesting.

Best of luck in future ventures!

 

[dr_octagon]

Here's another way of looking at engagement--

Edit: I don't think it's the case the threads end because of these last posts. It's probably a good way of estimating posting frequency since I don't have comment-level detail data yet.

ÆMNE22A!C
It says 22 brah...

 

[ahtlas7]

That was a typo autocorrect. I fixed it

 

[ÆMNE22A!C]

ÆMNE22A!C
It says 22 brah...

Holy shit LOL

 

[dr_octagon]

That was a typo autocorrect. I fixed it

Thanks. Thank.

 

[Chittagong]

Way ahead of you fam. Unfortunately I think the way to control for it is to cut out members who signed up prior to reactions being a thing. Unless evilore wants to give me access to the raw data lol

How about determining post counts per member with a COUNTIF post date later than reaction date

 

[Biff]

Any way to track overall forum engagement over time?

I don't know about you guys but the forum feels as alive as it ever has since, erm, "The Event".

I'm sure this is easily trackable via SimilarWeb or whatever. But just the feeling of the community, how quickly people reply to threads, the return of hilarious insta first-posts, great GIFs... I dunno - Neogaf feels like it's now finally "back".

 

[West Texas CEO]

wow, this thread was more insightful than I expected. Good job, Grildon Tundy .

 

[Grildon Tundy]

How about determining post counts per member with a COUNTIF post date later than reaction date

I don't have comment-level detail, unfortunately, just aggregate comment counts, reaction scores, account creation and most recent activity dates. So can't do that.

Doing what I can working with the data that's scrapable. I'd love to have full access to the server-side data. Thatd really open it up

 

[Grildon Tundy]

Any way to track overall forum engagement over time?

I don't know about you guys but the forum feels as alive as it ever has since, erm, "The Event".

I'm sure this is easily trackable via SimilarWeb or whatever. But just the feeling of the community, how quickly people reply to threads, the return of hilarious insta first-posts, great GIFs... I dunno - Neogaf feels like it's now finally "back".

Hey, Biff, see my reply above. The data i can grab is really limited. If I had a full server-side data set we could really do a deep dive

 

[Maiden Voyage]

Why not do gaming side as well?

 

[Grildon Tundy]

Why not do gaming side as well?

The Gaming data job is running as we speak Started with Off-Topic because it seemed to have less activity, and the data parser has to run in "real-time". Took two hours to download the dataset used in the OP.

 

[Maiden Voyage]

The Gaming data job is running as we speak Started with Off-Topic because it seemed to have less activity, and the data parser has to run in "real-time". Took two hours to download the dataset used in the OP.

I want to see the community section as well, if that's an option. Thanks for doing the work, brother.

 

[Con_Z_ǝdʇ]

The Gaming data job is running as we speak Started with Off-Topic because it seemed to have less activity, and the data parser has to run in "real-time". Took two hours to download the dataset used in the OP.

Community section as well?

 

[Grildon Tundy]

.

 

[Grildon Tundy]

I want to see the community section as well, if that's an option. Thanks for doing the work, brother.

Community section as well?

Good call. I don't visit the Community section much, so the lack of its reference was my bias coming through lol. Sure, I'll see if I can do that, too