News:

Forum changes: Editing of posts has been turned off until further notice.

Main Menu

Statistics

Started by Valamir, April 05, 2002, 08:50:34 PM

Previous topic - Next topic

clehrich

I just did a little breakdown of the top 30 posters, where columns included among other things:
Total number of posts
Length of membership (in days)
Time since last post (in days)

Then I did a little projection of the last quarter, like this:

(90 days - Time since last post) * (Total number of posts / Length of membership)

In other words, (How many days of last quarter active) * (Posts per day)

I don't know anything about statistics, but I could hand over the raw data if you like.

The numbers seem to me quite interesting.  It looks like there is not a whole lot of correlation between the factors Number of posts, Posts per day, and Recent posts.  For example, Lxndr comes in sixth if you sort by Posts-per-day or by the 90-day projection, but 24th in total postings.  Clinton (Nixon) is 8th in postings, 14th in per-day, and 16th by the projection.  And so on.

If I did the calculations right, I think that there is a 15.33% margin of non-correlation between Total and Per-day, and an 18.44% margin between Total and the 90-day projection.  [I just took the absolute value of the difference between the two rankings, averaged the list, and divided by 30 (the number of entries).]

As I say, I don't know anything about statistics, but it looks to me like this suggests we have a significant amount of spurt-drift behavior, if you get me; I mean that a lot of heavy posters, at least, seem to post a whole lot in a relatively short period, then post little for a while.  Of course, screwing this up are LeJoueur, Jared Sorensen, and Christoffer Lerno, who haven't posted in 300+ days each.  And with a sample of only 30, 3 people way off is a problem.  If the procedure is correct, or can be corrected, it might be interesting to do similar samples in other ranges of total posting, but now this is starting to sound like you have to know what the hell you're talking about.

Ralph or M.J., since you guys know what you're doing, would you be interested in this raw data?  Or can you tell me what stat things to apply to it (it's in a spreadsheet now, so I can pretty much tell it to do things, or upload it as a chart)?
Chris Lehrich

M. J. Young

I'm trying to figure out what 90 days minus days since most recent post gets you (now that I think I've figured out how you get that number). What was your thinking about that particular piece of the equation?

Maybe Ralph sees something I missed; I'm pretty much self-taught in statistics. (I'm a generalist, you know--I'm learning less and less about more and more until one day I hope to know nothing about everything.)

--M. J. Young

clehrich

Quote from: M. J. YoungI'm trying to figure out what 90 days minus days since most recent post gets you (now that I think I've figured out how you get that number). What was your thinking about that particular piece of the equation?
My idea was to project, on the basis of average posts per day over the course of one's membership, how many posts the person would likely have made during the last 90 days.  By subtracting from 90 the number of days since last posting, the theory was that (for example) Jared wouldn't come up as posting anything, and someone who hasn't posted for a week would be assumed to have posted at his usual rate prior to that week, i.e. 83 days.  It's very rough, and would presumably have to be done quarterly or whatever, after which it might stabilize.  

The problem is that there isn't really any way that I know of to determine how many posts someone has actually made in a given span of days, since the total posts listed under someone's name is updated continuously.  That is, if your 1000th post was still marked as such, I could just count back 90 days and see that you'd made, let's say, your 1100th post, meaning that you've posted some 369 times in the last 90 days.  That would enable us to figure out how often people post during specific periods, such as quarters.  Instead, I created this obviously rather dubious projection equation.

Any suggestions, from you or Ralph (or anyone else who knows something about statistics), about how better to do this?
Chris Lehrich

John Kim

Quote from: clehrichThe problem is that there isn't really any way that I know of to determine how many posts someone has actually made in a given span of days, since the total posts listed under someone's name is updated continuously.  That is, if your 1000th post was still marked as such, I could just count back 90 days and see that you'd made, let's say, your 1100th post, meaning that you've posted some 369 times in the last 90 days.  That would enable us to figure out how often people post during specific periods, such as quarters.  Instead, I created this obviously rather dubious projection equation.
So I whipped together a little Perl script (i.e. a little program) which can sift through a user's log of posts and collect them into forums and month/year.  So I've put a sample of the output at

http://www.darkshire.net/~jhkim/rpg/theory/theforge/members/clehrich.html

Now, there are various sorts of statistics which one could collect from all this, but I'm not sure what things to look for.  Is this useful?  What data are we looking for?  One measure might be the standard deviation of posts per month?  (i.e. how regular a poster is that person).  

Also, I could share code and/or raw data.
- John

Lxndr

How long does that take per user?  How much load would it put on the forge's machine to run it for all 3000+?
Alexander Cherry, Twisted Confessions Game Design
Maker of many fine story-games!
Moderator of Indie Netgaming

John Kim

Quote from: LxndrHow long does that take per user?  How much load would it put on the forge's machine to run it for all 3000+?
Well, it takes one page load per 15 posts.  For the sample on clehrich, I spaced out the 65 page loads with one every 5 seconds, so it took a few minutes to finish.  If I were to do a complete statistics collection at that rate, it would take about a day (i.e. 127000 total posts means roughly 8500 page views, with one every 5 seconds would be 11 hours).  That isn't very much for a website like this -- I'd guess it wouldn't be any more load than just a few regular Forge readers.  

So as long as I space it out over a day or so it shouldn't be noticeable.  And after I collect the statistics once, for the future I would only need to refresh by checking the latest posts -- which is much less.  But I want to test a bit to make sure I'd only have to do it once.
- John

Valamir

Hey, just got back from a week's vacation where I was out of touch.  Its good to see there is still interest in these quarterly updates I've been doing.  I'll continue to do them as long as folks see some value in it.

John, that output looks amazing.  I don't have any solid thought's on it yet, but I can well imagine there's alot of interesting things one could try to pick from such data...especially if the data output is in a manipulable array format.

...off the top of my head,

it would be possible to track the trend in forum postings to see if there is any discernable cycle in the rate of posting in  GNS or RPG theory activity.  

to determine which forums the newest members are posting to most frequently as a gauge of what's attracting folks to the site and where their initial interest lies.

to find out how many members we have who predominantly post in a specific game forum rather than participate in the general forums.

To see how many of the top posters not only post in the general theory forums but also down in the game forums in direct support of a featured publisher.

To see how many of the top posters to the RPG or GNS theory forums are also regular posters to the Actual Play forum (as a way of motivating a greater focus in application).

Others could probably come up with a few more uses.

I think all of these could be gleaned from the output format you have now, but it would be rather manually intensive.  How manipulable is the data you've pulled, John?

Rob Carriere

Ralph,
The scanning process John describes could collect all the things you mention in a single pass and/or extract the raw timing data and then do off-line stat with that.

Which makes me wonder whether it'd be possible to run this directly against the database? It'd probably be a single SQL query to get all the timing/user info for the entire site.

As for the stat side of this, I'm seeing interest in trends and periodic behavior. (which is good since you have to de-trend the data to get a decent handle on the periodic part anyway :) However, the stuff is currently being binned by the month. That means you're setting up for the detection of multi-month trends and periods and I think that the strong growth of the Forge pretty much guarantees that at any time the majority of members will not have enough history to be able to say anything useful about them. For example, the data for Chris equally support the hypothesis that he is a steady poster with a single lull and the hypothesis that he is a periodic poster who only posts in the first half of the year.

Of course, trends/periods for the Forge as a whole would work fine at this scale.

The period stuff I know how to do pretty well at various levels of ambition, but it's been a while since I've had to de-trend anything (which would be crucial in this case.)

SR
--

Valamir

Ok, time for another quarterly update.

We had 1114 members in January 2003, 1420 in April (+306; +28%), 1796 in July (+376; +27%), 2151 (+355; +20%) in October, and 2460 in January 2004 (+309; +14%). That's a 121% increase year over year.

We had 2889 in April (+429; +17%),  3266 (+377; +13%) in July, and now have 3603 (+337; +10.3%)

Membership growth continues to decline as a % which I think is a good thing for the continued quality of the site.  Interestingly, new members in absolute terms seems to be pretty consistantly between 300 and 400 members per month.

The top 10 posters have made 25% of the total posts (down from 40%, 38%, 37%, 35%, 33%, 31%, 30%, 29%, 28%, and 26%)
The top 20 posters have made 34% of the total posts (down from 60%, 53%, 50%, 46%, 44%, 42%, 40%, 39%, 37%, and 35%)
The top 40 posters have made 46% of the total posts (down from 75%, 69%, 66%, 61%, 59%, 55%, 53%, 52%, 49% and 47%)
The top 80 posters have made 59% of the total posts (down from 90%, 85%, 81%, 76%, 73%, 69%, 67%, 65%, 62%, and 60% )

This curve is really starting to flatten out.


There are now 29 members with more than 800 posts (up from 9, 10, 15, 15, 19, 23, 28 +1)
There are 35 members with between 400-800 posts (up from 18, 23, 24, 27, 25, 29, 31 +4)
There are 71 members with between 200-400 posts (up from 20, 29, 42, 47, 52, 58, 65 +6)
There are 102 members with between 100-200 posts (up from 42, 55, 64, 69, 83, 84, 92 +10)
There are 153 members with between 50-100 posts (up from 58, 60, 70, 93, 88, 116, 146 +7)
There are 196 members with between 25-50 posts (up from 72, 84, 100, 118, 155, 162, 178 +18)
There are 375 members with between 10-25 posts (up from 112, 151, 206, 243, 268, 315, 340 +35)
There are 327 members with between 5-10 posts (up from 120, 147, 179, 202, 223, 265, 295 +32)

In the April quarter 89 new members joined the botton two tiers (at least 5 posts), and in July 51 members climbed to the next tier or higher (at least 25 posts).   That's 57% of fresh new posters sticking around long enough to post at least 25 times. (yes, some of those will be fresh new posters from several months ago but for trend purposes it works)

In the July quarter 55 new members entered the funnel of the bottom two tiers.  Now in October 46 members climbed to 25+.  That's 84%.  Overall that's 67%.  No doubt the spike in this quarter is a result of some of last quarter's new members taking longer to process.

In the October quarter we've added 67 new members to the bottom two tiers.


There are 113 more people with at least 5 posts than there were in April. This compares to the  122 increase April to July, 139 increase January to April, 88 increase October to January, and the 113 increase from July to October.

Curious that this is the exact same number from a year ago.  The last three quarters have been in a decided down trend.



There are 405 members who've made only a single post (that's up from 150, 169, 207, 259, 300, 347, and 375 but down in % terms from 13% to 12% to 11% back to 12%, 12%, and 12% again, dropping back to 11%. And 11% again)

There are now 1387 members who've never made a single post (up from 312, 434, 588, 738 and 866, 1035, and 1221. In % terms that's up to 38% from 28%, 31%, 33%, 34%, 35%, 36%, and 37%).

This seems to be holding fairly steady. It would be nice to be able to eliminate accounts that have been inactive for more than a year or whose sole post was spam.

We're maintaining 16% (from 20% 18%, 18%, 17%, 15%, 16%, and 16%) of the total membership having more than 25 posts. This means that the number of people finding the site useful enough to post with some relative frequency (25+ posts), has stabilized. I wonder how this compares to other sites and what the usual decay over time for discussion forums like ours would be.


We've dropped to 6% of the total membership having at least 100 posts (from 9%, 8%, 8%, 7%, 7%, 7%, and 7%), In order to maintain 7% we'd need 1 new 100+ poster for every 14 new members. With 337 new members we'd need 24 new 100+ posters. We got 21.  We've been just shy for the last several quarters and rounding was no longer able to prop it up.  To maintain 6% we'll need 1 new 100+ poster for every 16 new members.  That's the ratio we've actually been getting the last couple quarters so it seems reasonable to expect to maintain that for a while.



The number of lurkers or largely non participating posters (less than 5 posts) stands at 64%, from (from 59%, 61%, 61%, 62%, 63%, 64%, and 64%). We continue to gain active participants (at least briefly active) at roughly the same proportional rate as we gain new members. This is a promising number


Continuing with the new stat I found of looking at the recent members (Joined since Julyl 2)

Of the 337 new members who joined since the last update:
38 have posted 1 time – 11% (15%, 16%, 16%, 13%)
46 have posted between 2 and 5 times – 14% (16%, 21%, 18%, 18%)
28 have posted between 6 and 10 times – 8% (10%, 6%, 7%, 7%)
12 have posted between 11 and 20 times – 4% (5%, 3%, 6%, 3%)
6 have posted between 21 and 30 times – 2% (3%, 2%, 3%, 1%)
6 have posted more than 30 times – 2% (3%, 4%, 5%, 3%)

Just in the last quarter.

This time around the numbers have dropped slightly. Only 40% (from 52%, 52%, 54%, 46%) of the newest members have posted at least once. 29% exactly (from 36%, 36%, 39%, 33%) posted more than once.

This quarter has broken the 1/2 of all new members post at least once, 1/3 post more than once rule.  

Of the 6 who posted more than 30 times, 1 posted 81, 1 posted 114 and 1 posted 178.  Seems like every quarter sees a couple very active posters.

Valamir

Its that time again:

We had 1114 members in January 2003, 1420 in April (+306; +28%), 1796 in July (+376; +27%), 2151 (+355; +20%) in October, and 2460 in January 2004 (+309; +14%). That's a 121% increase year over year.

We had 2889 in April (+429; +17%), 3266 (+377; +13%) in July, 3603 (+337; +10.3%) in October, and 3956 in January 2005 (+353; +9.8%).  That's a 61% increase year over year.  

Membership growth continues to decline as a % which I think is a good thing for the continued quality of the site.  Interestingly 2004's increase is almost precisely half of 2003s increase.  I continue to be struck by the consistancy in absolute numbers of new members.  Its been 300-400 a month pretty steadily for 2 years now.  

The top 10 posters have made 25% of the total posts (down from 40%, 38%, 37%, 35%, 33%, 31%, 30%, 29%, 28%, 26% and 25%)
The top 20 posters have made 34% of the total posts (down from 60%, 53%, 50%, 46%, 44%, 42%, 40%, 39%, 37%, 35% and 34%)
The top 40 posters have made 45% of the total posts (down from 75%, 69%, 66%, 61%, 59%, 55%, 53%, 52%, 49%, 47% and 46%)
The top 80 posters have made 58% of the total posts (down from 90%, 85%, 81%, 76%, 73%, 69%, 67%, 65%, 62%, 60% and 59%)

As expected, this curve is really starting to flatten out.  I wouldn't expect to see the needle on these numbers move much from this point on.  


There are now 31 members with more than 800 posts (up from 9, 10, 15, 15, 19, 23, 28, 29 +2)
There are 41 members with between 400-800 posts (up from 18, 23, 24, 27, 25, 29, 31, 35 +6)
There are 77 members with between 200-400 posts (up from 20, 29, 42, 47, 52, 58, 65, 71 +6)
There are 109 members with between 100-200 posts (up from 42, 55, 64, 69, 83, 84, 92, 102 +7)
There are 149 members with between 50-100 posts (up from 58, 60, 70, 93, 88, 116, 146, 153 -4)
There are 208 members with between 25-50 posts (up from 72, 84, 100, 118, 155, 162, 178, 196 +12)
There are 413 members with between 10-25 posts (up from 112, 151, 206, 243, 268, 315, 340, 375 +38)
There are 352 members with between 5-10 posts (up from 120, 147, 179, 202, 223, 265, 295, 327 +25)

In the April quarter 89 new members joined the botton two tiers (at least 5 posts).  In July, 51 members climbed to the next tier or higher (at least 25 posts). That's 57% of fresh new posters sticking around long enough to post at least 25 times. (yes, some of those will be fresh new posters from several months ago but for trend purposes it works)

In the July quarter 55 new members entered the funnel of the bottom two tiers. In October, 46 members climbed to 25+. That's 84%. Overall that's 67%.

In the October 67 new members joined the bottom two tiers.   In January, 29 members climbed to 25+.  That's 43% or about 60% overall of initial posters continuing to post at least 25 times.

Now in January we've added 63 new members to the bottom two tiers.


There are 92 more people with at least 5 posts than there were in October.  This compares to the 113 increase July to October, 122 increase April to July, 139 increase January to April, 88 increase October to January, and the 113 increase from July to October.

The last 4 quarters, basically the entire year, has been in a decided down trend.  Since the number of new members has been fairly consistant throughout the year I think these numbers indicate fewer new members who are actually posting (at least 5 times).  This could be a source of some concern if it continues.  Last year there was a big spike in the first quarter.  Maybe we'll see that again this year.
There are 463 members who've made only a single post (that's up from 150, 169, 207, 259, 300, 347, 375 and 405 but down in % terms from 13% to 12% to 11% back to 12%, 12%, and 12% again, dropping back to 11%, staying at 11% and now climbing to 12% again.)  Pretty consistant

There are now 1547 members who've never made a single post (up from 312, 434, 588, 738 and 866, 1035, 1221, and 1387. In % terms that's up to 39% from 28%, 31%, 33%, 34%, 35%, 36%,  37%, and 38%).

While the single posters seems to be holding fairly steady, the no posters has increased by nearly 40% from January 2003.  If this number were staying the same it would indicate that the same percentage of new members never post.  Since it trending upwards that means that even though we get a consistant number of new members each quarter the % of those new members finding something to post about is declining.  This is consistant with my earlier source of concern.  Is this simply indicative of casting a wider net due to the increase recognizability of the Forge as a site to check out.  Basic marketing teaches us that the less focused your advertising the lower your %age of responses.  I would hope that that is the case here.  

We're maintaining 16% (from 20% 18%, 18%, 17%, 15%, 16%, 16%, and 16% again) of the total membership having more than 25 posts. But this is due to rounding up and if the current trend continues will probably drop to 15% next time around.  

We're holding at 6% of the total membership having at least 100 posts (from 9%, 8%, 8%, 7%, 7%, 7%, 7% and 6%), In order to maintain 7% we'd need 1 new 100+ poster for every 14 new members. With 353 new members we'd need 25 new 100+ posters. We got 21.  To maintain 6% we'll need 1 new 100+ poster for every 16 new members.  



The number of lurkers or largely non participating posters (less than 5 posts) is now 65%, (from 59%, 61%, 61%, 62%, 63%, 64%, 64%, and 64%). We continue to gain active participants (at least briefly active) at roughly the same proportional rate as we gain new members. This is a promising number but it slipped this quarter along with the others noted above.


Continuing with the new stat I found of looking at the recent members (Joined since October 4)

Of the 353 new members who joined since the last update:
58 have posted 1 time – 16% (15%, 16%, 16%, 13%, 11%)
59 have posted between 2 and 5 times – 17% (16%, 21%, 18%, 18%, 14%)
16 have posted between 6 and 10 times – 5% (10%, 6%, 7%, 7%, 8%)
16 have posted between 11 and 20 times – 5% (5%, 3%, 6%, 3%, 4%)
3 have posted between 21 and 30 times – 1% (3%, 2%, 3%, 1%, 2%)
5 have posted more than 30 times – 1% (3%, 4%, 5%, 3%, 2%)

These numbers continue the trend of reduced posting we've seen above.  The first 2 lines moved up to the high end of their range (meaning more people with fewer posts) and the rest moved to the lower end of their range, including 2 lines setting new lows (meaning fewer people with more posts).

Consistant with this, 44% (from 52%, 52%, 54%, 46%, 40%) of the newest members have posted at least once.   This is up from last quarter.  But only 28% (from 36%, 36%, 39%, 33%, 29%) posted more than once which is down from last quarter.

The 1/2 of all new members post at least once, 1/3 post more than once rule seems to be in jeopardy.

Of the 5 who posted more than 30 times, 1 posted 48 times, and 1 posted 61 times.  Even this is down from previous quarters where there was typically 1-2 new posters in excess of 100.

Mike Holmes

I think we may be being overly pessimistic about the "downtrend" and related news. That is, I think that it's not the quality of the "advertisement" or the product not being attractive so much as the possibility that we're reaching a signal horizon. That is, at some point, people have to start to feel that their voice is going to get lost in the ever increasing amount of signal. So the more members we have posting, the fewer people see this as a viable channel.

I bet if you projected this trend, and compared it to, say, RPG.net's ratio of posters to non-posters, you'd see that we're still ahead of the curve with regards to this.

With with more and more posting, every person reads less and less of the total content. This is why I don't read RPG.net. Not because I don't like the content, but because I have this sense that for every post that I read, I miss seven just like it. That everything being said there is being said a million times. Even less than I read there, I post there extrememly little because of this (only a handful of times last year).

So I think this curve may simply be inevitable. Basically there's only so much room in one channel for posters, and the rest are going to start to become, more and more, readers.

Mike
Member of Indie Netgaming
-Get your indie game fix online.

Green

Quote from: Mike HolmesI think we may be being overly pessimistic about the "downtrend" and related news. That is, I think that it's not the quality of the "advertisement" or the product not being attractive so much as the possibility that we're reaching a signal horizon. That is, at some point, people have to start to feel that their voice is going to get lost in the ever increasing amount of signal. So the more members we have posting, the fewer people see this as a viable channel.

I think you're on to something, but pinning the cause solely on statistics undermines what the Forge can do to reverse or minimize this trend.  Expecting everyone to contribute equally to everything is unrealistic, yet I think being aware of Forge attitudes, habits, and tendencies that push people away can help more people contribute more consistently.

I don't think this thread is the best place to go into details, but I could discuss this further in PMs or start a new thread if more than just me and another person is interested.

jdagna

I also agree that the number of non-posters isn't something to get down about.  In some ways it's a good thing.

I keep pretty close track of my own forums, so I know that there are people who have no posts to their name who still log in at least once a week.  In fact, you can expect 15 unique people to check back in on an average week and about 5 of those rarely, if ever post.

I take this as an indication that they find the site a useful resource - they're either finding information that answers their questions or see new information that they like reading.  With an informative/support site, these active non-posters show that it's fulfilling its purpose.  

The Forge's primary function seems to be a resource for indie game designers, and it can do that job marvelously whether they post or not (provided that someone is still posting, and there's no problem there).  Only a secondary function of the Forge (refining and inventing theoretical models) really requires that people post and participate.

Knowing just how many non-posters still frequent the forums would be a very useful statistic and I'm sure the raw hit logs would reveal just how many unique visitors the Forge gets every day.  I'm sure those numbers would be very encouraging.
Justin Dagna
President, Technicraft Design.  Creator, Pax Draconis
http://www.paxdraconis.com

Mike Holmes

Well, just to be clear, the issue is not that people do not post. We expect that some proportion of people are here to read, and not to post. The "problem" is that the proportion of non-posters seems to be increasing. The thought is that this could be indicative of less arrivals here finding the site useful. That is, presumably, of the people who do not post, some of them are the ones who arrive, look around, and decide that the site is not useful. If the proportion of non-posters grows, the argument goes, then it's possible that the proportion of new arrivals who find the site not worth coming back to, is increasing.

Or, at the very least, that of the people who arrive that a smaller proportion of those find the site not worth posting to. Which would be problematic, in theory, because the eventual outcome of that trend would be that we'd have few or no new blood posting. Which even I'd agree would be a bad thing. Basically, we don't want the Forge to be a place that people feel they just can't get into at all in terms of participation. No, that doesn't mean that we make it easy for people to participate, generally, we have a relatively high bar to entry. We just don't want there to be factors other than we consciously have decided to put in to setting that bar to be keeping people from posting. So, for example, it's fine that trolls are discouraged from posting, but not fine if people with something to contribute feel, say, that people here at too closeminded to talk to.

I don't think that the latter or anything like it is happening - my point here is that I think that there are other forces conspiring here to bend the curve the way that it has gone. Again, I think that it's simply an overload of information, that growth at the same proportions is just untennable. But that's just my theory. What's most important is to realize that the statistics could be saying many things. This particular trend might indicate something to worry about, and it might not.

Which means to me that I think it's far from time to panic, but certainly time for people like Green who have theories about any potential problems to put them forward for consideration.

Mike
Member of Indie Netgaming
-Get your indie game fix online.

Valamir

Its that time again:

We had 1114 members in January 2003, 1420 in April 2003 (+306; +28%), 1796 in July 2003 (+376; +27%), 2151 (+355; +20%) in October 2003, and 2460 in January 2004 (+309; +14%). That's a 121% increase year over year.

We had 2889 in April 2004 (+429; +17%), 3266 (+377; +13%) in July 2004, 3603 (+337; +10.3%) in October 2004, and 3956 in January 2005 (+353; +9.8%). That's a 61% increase year over year.

We have 4401 in April 2005 (+445; +11.2%).  Interestingly, there seems to be a pattern of First Quarter surges, with the January to April period posting higher than average membership growth.  


The top 10 posters have made 24% of the total posts (down from 40%, 38%, 37%, 35%, 33%, 31%, 30%, 29%, 28%, 26%, 25% and 25%)
The top 20 posters have made 33% of the total posts (down from 60%, 53%, 50%, 46%, 44%, 42%, 40%, 39%, 37%, 35%, 34% and 34%)
The top 40 posters have made 45% of the total posts (down from 75%, 69%, 66%, 61%, 59%, 55%, 53%, 52%, 49%, 47%, 46% and 45%)
The top 80 posters have made 58% of the total posts (down from 90%, 85%, 81%, 76%, 73%, 69%, 67%, 65%, 62%, 60%, 59% and 58%)

As expected, this curve is really starting to flatten out. I wouldn't expect to see the needle on these numbers move much from this point on.


There are now 34 members with more than 800 posts (up from 9, 10, 15, 15, 19, 23, 28, 29, 31 +3)
There are 40 members with between 400-800 posts (up from 18, 23, 24, 27, 25, 29, 31, 35, 41 -1)
There are 81 members with between 200-400 posts (up from 20, 29, 42, 47, 52, 58, 65, 71, 77 +4)
There are 113 members with between 100-200 posts (up from 42, 55, 64, 69, 83, 84, 92, 102, 109 +4)
There are 161 members with between 50-100 posts (up from 58, 60, 70, 93, 88, 116, 146, 153, 149 +12)
There are 214 members with between 25-50 posts (up from 72, 84, 100, 118, 155, 162, 178, 196, 208 +6)
There are 445 members with between 10-25 posts (up from 112, 151, 206, 243, 268, 315, 340, 375, 413 +32)
There are 367 members with between 5-10 posts (up from 120, 147, 179, 202, 223, 265, 295, 327, 352 +15)

In the April 2004 quarter 89 new members joined the bottom two tiers (at least 5 posts). In July 2004, 51 members climbed to the next tier or higher (at least 25 posts). That's 57% of fresh new posters sticking around long enough to post at least 25 times.

In the July 2004 quarter 55 new members entered the funnel of the bottom two tiers. In October 2004, 46 members climbed to 25+. That's 84%. Overall that's 67%.

In the October 2004 quarter 67 new members joined the bottom two tiers. In January 2005, 29 members climbed to 25+. That's 43% or about 60% overall of initial posters continuing to post at least 25 times.

In the January 2005 quarter 63 new members joined the bottom two tiers.  In April 2005 only 28 members climbed to 25+.  That's 44% or about 56% overall of initial posters continuing to post at least 25 times


There are only 75 more people with at least 5 posts than there were in January. This compares to the 92 increase from October 2004 to January 2005, 113 increase July 2004 to October 2004, 122 increase April 2004 to July 2004, 139 increase January 2004 to April 2004, 88 increase October 2003 to January 2004 , and the 113 increase from July 2003 to October 2003.

The last 4 quarters, basically the entire year, has been in a decided down trend. Since the number of new members has been fairly consistent throughout the year I think these numbers indicate fewer new members who are actually posting (at least 5 times). This could be a source of some concern if it continues. Last year there was a big spike in the first quarter coinciding with a spike in membership.  This year we had a first quarter spike in membership but the number of new posting members declined another 18.5% matching last quarter's decline.



There are 507 members who've made only a single post (that's up from 150, 169, 207, 259, 300, 347, 375, 405 and 463 but down in % terms from 13% to 12% to 11% back to 12%, 12%, and 12% again, dropping back to 11%, staying at 11%, climbing to 12% again, and remaining at 12%.) Pretty consistent

There are now 1829 members who've never made a single post (up from 312, 434, 588, 738 and 866, 1035, 1221, 1387 and 1547. In % terms that's up to 42% from 28%, 31%, 33%, 34%, 35%, 36%, 37%, 38% and 39%).

While the number of single posters seems to be holding fairly steady, the non-posters have continue to ratchet up.  As Mike mentioned last quarter, concern over this trend might be premature but at the very least it indicates that the bar to entry into Forge conversation has increased substantially over the last year.

As expected the number of total membership having more than 25 posts dropped to 15% this quarter (from 20% 18%, 18%, 17%, 15%, 16%, 16%, 16%, and 16% again).  What's notable is that it almost dropped all the way to 14% (after rounding).  

We're holding at 6% of the total membership having at least 100 posts (from 9%, 8%, 8%, 7%, 7%, 7%, 7%, 6% and 6%), In order to maintain 6% we'd need 1 new 100+ poster for every 16. With 445 new members we'd need 28 new 100+ posters. We got 10.  Looks like we'll probably be slipping to 5% in this category next time around.


The number of lurkers or largely non participating posters (less than 5 posts) is now 67%, (from 59%, 61%, 61%, 62%, 63%, 64%, 64%, 64% and 65%).  That's now 2/3s of our membership having never posted more than 5 times.

Continuing with the new stat I found of looking at the recent members (Joined since October 4)

Of the 445 new members who joined since the last update:
48 have posted 1 time – 11% (15%, 16%, 16%, 13%, 11%, 16%)
54 have posted between 2 and 5 times – 12% (16%, 21%, 18%, 18%, 14%, 17%)
11 have posted between 6 and 10 times – 2% (10%, 6%, 7%, 7%, 8%, 5%)
11 have posted between 11 and 20 times – 2% (5%, 3%, 6%, 3%, 4%, 5%)
1 has posted between 21 and 30 times – 0% (3%, 2%, 3%, 1%, 2%, 1%)
3 have posted more than 30 times – 0% (3%, 4%, 5%, 3%, 2%, 1%)

These numbers highlight the recent trend rather dramatically.  Last quarter the first 2 lines moved up to the high end of their range (meaning more people with fewer posts) and the rest moved to the lower end of their range, including 2 lines setting new lows (meaning fewer people with more posts).   This quarter all but one line set a new low and that 1 (the first) tied for its low.  To put this in perspective, a few quarters ago 5% of all new members had posted more than 30 times in their first quarter of membership.  To duplicate that this quarter would have required 22 members.  The actual number was 3.

To hammer the difference home, we'd been holding at approximately half of the newest members posting at least once.  At one time I'd considered that a rule of thumb.  This quarter, however, that number is 29% (from 52%, 52%, 54%, 46%, 40%, and 44%).  Similarly about one third of the new members had posted more than once.  This quarter, however, that number is 18% (from 36%, 36%, 39%, 33%, 29%, 28%).

The "1/2 of all new members post at least once, 1/3 post more than once" rule seems to now be broken, at least for the short term.

Of the 3 who posted more than 30 times, the highest was 51 times.  Even this is down from previous quarters where there were typically 1-2 new posters in excess of 100.


Make of it what you will.  I invite discussion.