Statistics

[M. J. Young]

Ralph, just before you got to the part where you were giving percentages of members I was wondering about percentages of members. It seems to me that it's obvious that the percentage of posting by the top 10 posters is going to decrease as the total number of posters increases; it would be stunning if it did not. But has the percentage of posts from the top 1% of posters changed significantly? 10 members was pretty close to 1% last time (well, almost 0.9%), but it's dropped to 0.7% now.

I realize that tells us something different, but it might also be worth knowing: as the site grows, is there still the same pattern of frequent posters versus infrequent posters? Are we adding people to the top posters group as fast as we're adding people generally?

I could probably figure out some of that from the other numbers, but I thought I'd ask, as you seem to be able to manipulate the data so easily.

--M. J. Young

[Valamir]

Ralph, just before you got to the part where you were giving percentages of members I was wondering about percentages of members. It seems to me that it's obvious that the percentage of posting by the top 10 posters is going to decrease as the total number of posters increases; it would be stunning if it did not.  

Actually, given that the prolifacy of those top 10, that was hardly a forgone conclusion (especially when we were under 1000 members), although now that the trend has been established it is certainly more likely to continue than less, and mostly I continue to track it just cause its fun.

But has the percentage of posts from the top 1% of posters changed significantly? 10 members was pretty close to 1% last time (well, almost 0.9%), but it's dropped to 0.7% now.

I don't think its possible to go back and find that information historically, the closest we could likely come is to interpolate from the data I did gather.  But I will say that while I forget the exact number the 20/80 rule is still running more like 20/90 here.

I realize that tells us something different, but it might also be worth knowing: as the site grows, is there still the same pattern of frequent posters versus infrequent posters? Are we adding people to the top posters group as fast as we're adding people generally?

I did comment on this from another angle.  The percentage of low posters has remained fairly close to even last time to this despite 300 odd new members indicateing that "low posters" (under 5 posts) are becoming "non low posters" (5 or over) as quickly as new members are added (almost).  Also the number of posters hitting 100 posts as a percentage is remaining fairly steady indicating that an equivelent number of new members are sticking around to post a fair amount (again almost).

However, since the system offers no way that I know to drop from concideration members who haven't posted in over a year...the number of dead accounts may well be skewing the results somewhat.

[deadpanbob]

If you really want to equalize for the great lead that older posters seem to have, you could take a look at the pace of posts based on Cohort.

That is to say, if you have this level of access, you can group all the posters into groups based on the month and year that they joined, then look at each groups pace of posting in 30 day increments over the life of the site.

You'd be looking to see if the new posters are posting at the same rate (in terms of posts per member) over their first thrity days, next thirty days, next thrity days and so on as the older memebers.

This is probably, in my experience, the past way to equalize for time spent here, and the quickest way to show equal and fair comparisons of new members to older memebers.  It's also a great way to see fatigue in the oder members, as some drop off month to month never to post again, or as their post frequency goes down.

If your interested in doing this sort of thing, have access to the data, but still have questions after all my blather, PM me or email me at deadpanbob@hotmail.com and I can explain further.

Cheers.

[Valamir]

All of my data comes direct from the member list.  I'm not sure what additional data Clinton may or may not have access to.

[Brian Leybourne]

The thing that puzzles me is the vast number of people who have registered, but never posted.

Registered and posted once, I can understand, that's a lurker who felt strongly enough about a particular post to jump in and respond.

But register and then never post? Why bother registering if you're only going to lurk and read anyway? Simply because it lets you more easily keep track of what you've read?

Bizarre.

Brian.

[Ron Edwards]

Hi Brian,

Well, it's not so bizarre for two reasons.

1. Keeping track of what you've read is pretty useful, and apparently there are plenty of "contented readers-only" out there. How many? Probably not a major fraction of the folks you mention, just to speculate, but probably enough to grunt at.

2. The internet is full of people just cruising and looking for stuff to see and/or download. Registration-sites often carry bennies for registrants; a semi-bored, possibly ADD-afflicted surfer might register at hundreds or dozens of sites in hopes of a kewl thing to get. Maybe there's a secret forum. Maybe there are naked-lady pictures. Maybe there's a free game. Whatever. So they register, realize there's nothing like that here, and move on.

Best,
Ron

[Brian Leybourne]

Ron,

Fair comment. Do you and Clinton ever do sweeps of old, unused memberships? Actually, I can see you don't, so maybe I'm suggesting that you should. I'm guessing 1000 users is not stressing the phpBB software, but there's little point in maintaining an account that was opened two years ago and has never been accessed since (for example).

Brian.

[Valamir]

Ok, its time for another quarterly Statistics update.  I waited until just before GenCon to do this one, because my thought is to do an interrim update about a month after GenCon to try and see if our GenCon exposure had any immediate kick in membership.

We had 1114 members in January, 1420 in April (+306), and are now up to 1796 (+376)


The top 10 posters have made 31% of the total posts (down from 40%, 38%, 37%, 35% and 33%)
The top 20 posters have made 42% of the total posts (down from 60%, 53%, 50%, 46% and 44%)
The top 40 posters have made 55% of the total posts (down from 75%, 69%, 66%, 61% and 59%)
The top 80 posters have made 69% of the total posts (down from 90%, 85%, 81%, 76% and 73% )

This trend seems long past the point where there’s any chance of it reversing.  Its still fun to track, however, and I suspect that it will bottom out somewhere.


There are now 15 members with more than 800 posts (up from 9, 10, +5)
There are 24 members with between 400-800 posts (up from 18, 23, +1)
There are 42 members with between 200-400 posts (up from 20, 29, +13)
There are 64 members with between 100-200 posts (up from 42, 55, +9)
There are 70 members with between 50-100 posts (up from 58, 60, +10)
There are 100 members with between 25-50 posts (up from 72, 84, +16)
There are 206 members with between 10-25 posts (up from 112, 151, +55)
There are 179 members with between 5-10 posts (up from 120, 147, +27)

A huge spike in the 800+ poster range this time around.  Next quarter will likely see little increase in this area as the next batch of posters is in the 600 post range and are generally less prolific than the current batch.  With that many posters leaving the second tier the 400-800 range only managed to increase by 1.  

In April we had 13 new members in the 100-200 range.  It looks like those folks kept on posting and now we have 13 new members in the 200-400 range.  The bubble moves higher!  But even with that many people moving up to the next tier, the 100-200 range still managed to gain 9 new members.  I must believe that many of those were from the 25-50 range (which had 12 new members in April) who also kept on posting.

These numbers are extremely exciting.  Not only are we seeing large numbers of new posters from previous quarters stick around and begin posting regularly, but there are even more new posters starting to post this quarter than last.  That’s 141 more people with at least 5 posts this quarter than last, which was only 104 higher than the previous.



There are 207 members who've made only a single post (thats up from 150, and 169 but down in % terms from 13% to 12% to 11%)
However there are now 588 members who've never made a single post (up from 312, and 434 and in % terms up to 33% from 31% and 28%).

What this tells me is that while the number of pure lurkers (no posts) is growing as a % of overall membership, the number of people who’ve been interested enough to post just once and never came back to post again is shrinking in relative terms.  Meaning, people who find this site useful, are finding it useful enough to continue to post after their first message.  A pretty good thing I’d say.


We're holding steady at 18% (from 18%, 20%) of the total membership haveing more than 25 posts.  This means that the number of people finding the site useful enough to post with some relative frequency (25+ posts), are increasing in exact proportion to the number of new members we’re adding.   This is also pretty amazing given the volume of new members we’re adding.

We’re also holding steady at 8% of the total membership having at least 100 posts (from 8%, 9%).  I said last quarter this will be a difficult number to maintain as we’d need 12 new 100+ posters for every 12 new members.  Well, with 376 new members we’d need 31 new 100+ posters.  We actually got 28, which is pretty darn close.


As you can probably already guess from the numbers above, the number of lurkers or largely non participating posters (less than 5 posts) has remained unchanged at 61% of total members also (from 61%, 59%).   This is probably the most incredible number of all.  Gaining participants at the same proportional rate as new members is absolutely phenomenal.  Now this is not a direct measure of active participants.  There are probably plenty who posted 5 times and that’s it.  But every person who posted a few times, found the answers they were looking for and declined to stay, is still a victory.  It means people were reached in a manner deeper than a casual skim.  Even the people who left after disagreeing with what they found here were at least confronted with the idea of appying critical thought to their roleplaying, and found enough here to be worth commenting on at least a few times.

With so many numbers having remained steady (at least for a quarter) I think we’ll have a good backdrop to see how big a rush of new members GenCon does or doesn’t produce.  1 month won’t be long enough for most current members to change their “tier” status much, nor will it be long enough for new GenCon members to have posted much.  We should be able to get a fairly good read on the influx of new members and then quarters down the road whether that batch of new members stuck around like recent batches have (apologies to referring to our new members as “tiers” and “batches”) or not.  Should be interesting.

[M. J. Young]

This trend seems long past the point where there?s any chance of it reversing. Its still fun to track, however, and I suspect that it will bottom out somewhere.

I'm not sure it will bottom out.

You have to remember that the top ten posters, at the same time they represent a smaller percentage of the posts, they also represent a smaller percentage of the posters. As long as membership on the site keeps increasing, that block of top posters becomes a smaller and smaller fraction of the whole.

When there were 1114 members, the top 80 were roughly 7% of the total; now that there are 1796, that same top 80 are not quite 4.5%. In each category, for each of the two quarters reported here, that number represented a smaller fraction of the whole by roughly 20%.

In theory, to stay even, either the majority of the membership would have to post less on average, or the top posters would have to constantly post more. Since the percentage of non-posters is not increasing, and all categories are increasing in numbers, that doesn't look like it's about to happen.

Sorry, Ralph--it's not likely to level out.

--M. J. Young

[Valamir]

Sorry, Ralph--it's not likely to level out.

Doesn't that presume that the number of active posters continues to increase indefinitely?  At some point the number of people who can actively hold conversations on the Forge simultaneously and be heard will reach a critical mass.  Likely one much smaller than places like RPG.net due to more heavy moderation.

At that point the relationship between top 80 posters and a fairly stable (in numbers not necessarily individual members) group of regulars should become fairly assymptotic I would think.

[Hunter Logan]

Seems to me the numbers are imperfect, because you have members who made accounts and abandoned them. So, you count a guy who has made more than 5 posts as an active member, but that guy made his last post 2 years ago. He's not active any more, so he really shouldn't be counted in current tallies of active posters. Or maybe members made their accounts, never posted, and didn't stay, either. So you have a certain amount of dead wood in the pool. That might be skewing the results. It might be useful to track total number of posts to the site during a given period or some other measure of site traffic to accompany the tracking of members' number of posts, and to calculate the number of posts each group makes during ther given time period. Then, you could say, "Traffic is up this much, the top posters represent this much of the traffic, and other groups this much," etc

It seems to me that as traffic increases, the activity of your top posters will also increase - but only up to a certain point. Beyond that point, the top posters won't have time or energy to keep up with the increase in traffic. Then, their impact will drop significantly. The question is, can you track the posting trends with any real accuracy, and will the site continue to draw a continuously increasing percentage of active members who stay active for the time period in question? It's possible that the rates will level off as Ralph suggests because the number of new people who join and actively post could balance the number of existing members who stop posting. I don't know if it's doing that, but I think it would take a more aggressive evaluation of the available data to really find out.

You'll never have the traffic rpg.net has because you don't have something like Tangency to artificially boost your traffic - And that's a good thing.

[Valamir]

Seems to me the numbers are imperfect, because you have members who made accounts and abandoned them. So, you count a guy who has made more than 5 posts as an active member, but that guy made his last post 2 years ago. He's not active any more, so he really shouldn't be counted in current tallies of active posters.

I note that above.  Total posts is not and cannot be a measure of active members.  But the numbers of people moving from one tier to another from time period to time period shows a broad level of activity.  So its not the raw number of posters in each category that's of interest, but the number of new posters in each category that reflects traffic trends in terms of posts.

Plus, "traffic" must be adapted for Forge purposes.  Number of hits and number of views don't really say too much.  The Forge is about discussion and exchanges of ideas and that requires actual postings.  Its certainly possible for a Lurker to come an read and find things that enhance their game play, but if they're not posting than their is not exchange, the ideas are traveling only 1 way.

The primary purpose for tracking posters and the ratio of top posters to the total stems back to when I started looking at the numbers.  There was a very real possibility that all the Forge was was a group of the same few dozen people tossing ideas back and forth in a very private and insular group.  In fact, there were some folks who'd suggested that's all the site was and would ever be.  What the numbers show is that the base of total people contributing ideas and comments to the archive of knowledge and experiences here continues to broaden.  There are far more people posting ideas (both game and theory ideas) and the like now then there were a year ago.  The diversity of posters, experiences, and "ideologies" continues to expand.  It was not always certain that this would happen.  The Forge *could* have gone the other way and become a dozen crotchety old gamers shouting at each other over points of minutia.  It didn't, and that's what those "Top 20 posters have X% of the total posts" demonstrates.  I think the site has reached a mass and level of awareness where that danger is long behind us, so as I say, I mostly just keep track of that piece for fun...I like graphing trends, its what I do.

That said I would love to have more data available.  Even a column of "most recent post date" would be HUGE in terms of evaluating numbers of Casual Posters, Active Posters, Very Active Posters, and Hyper Active Posters.  I don't know whether Clinton could make such info available or not.

[Hunter Logan]

Ralph,

You're absolutely right about the utility of a "most recent post date" and in your broad assessment. The "most recent post date" would assist in determining current participation; and you've got a lot more here than 20 cranky, old gamers yelling about minutiae.

[greyorm]

I like tracking and graphing statistics and trends as well. I like finding and exploring patterns, it's an occasional hobby for me, so I'd love to have that extra data available to see how the community is evolving and reshaping itself over time.

Any chance you'd be able to provide that "most recent post" data, Clinton?
(or, gods forbid, the statistic geek's holy grail, # posts in a given month per poster?)

[Valamir]

brief update.  I'm going to wait a full month to crunch the numbers, but I just noticed that we've added 66 new members since I ran them last...11 days ago.  At that rate we'd be well over 500 new members for a full quarter...much higher than a typical non GenCon quarter.  Should be interesting.

I think when I run the numbers I'm also going to start a "if you heard about as at GenCon, post here" thread, just to see how effective our fairly heavy Forge promotion was.