Beefy Boxes and Bandwidth Generously Provided by pair Networks
Perl: the Markov chain saw

Website Users

by thekestrel (Friar)
on Apr 03, 2005 at 18:15 UTC ( #444513=monkdiscuss: print w/replies, xml ) Need Help??

     I was looking at the user statistics for PerlMonks here and was curious about some of the user statistics. At the time of writing this it said there were;
Total Registered PM Users 33512 Logins In The Last 30 Days 2660 Users Never Logged In 6337 Logged In Once, No Write Ups 19998 Users With Write Ups 7177

     I didn't realize that a website with a userbase would have such a high percentage of dead users (81%) as I've never ran a website so have no idea. This is nothing to disparage PerlMonks, but it got me to thinking is this the sort of percantages one might expect on a website out in the world? Sort of like in the mail order or spam world, they know from sending out 1,000,000 emails/catalogs they'll get say 1% of responses and sales and these numbers are relied upon for these business models to work. I figure that some of PerlMonks community might have access to similar data about other real world sites and I was curious as to if there is a general set of rules one can approximate for usage and constituency etc. I'm not interested in anyone giving the actual domain name of data they may access to, but the numbers would be interesting.
     Some numbers that might be worth comparing;
[PerlMonks example] Percentage of dead users: 81% Percentage of users logged in within 30 days: 7.9% Percentage of active/contributed users: 21.4%

     Does anyone have any insight in to other websites or has someone already written a paper on say 'expected live users from userbase etc...'. Just curious... =)

Regards Paul.

Replies are listed 'Best First'.
Re: Website Users
by kvale (Monsignor) on Apr 03, 2005 at 18:46 UTC
    Websites are used in different ways by different folks. For instance, I read Slashdot several times a week, but would be classified as a 'dead' user because I never registered and never posted a comment. To me, 'dead' has a pejorative sense that denigrates a fine use of PM: reading and learning.

    Because simply browsing the Perlmonks site is a perfectly valid activity not captured by your stats, I think that the ratio of writers to readers is even more extreme that you portray it. But that is ok. If every visitor of PM felt compelled to submit a node, I think the result would be pandemonium.

    I don't know of any participatory website statistics, but I suspect they would be similar to PM. To answer your question, ask yourself: How many websites do I visit that allow participation vs. websites that I actually participate in? For me over the span of months it is hundreds vs. 2.


           Good point, I never considered people that just read without ever logging in. I must admit for a long time (~years) I did the same thing with Slashdot, I read it religously, but rarely if ever comment. So point taken dead users is probably not the best description.
           Thanks for the input.

      Regards Paul
      I read Slashdot several times a week, but would be classified as a 'dead' user because I never registered and never posted a comment.

      No, that doesn't make you a 'dead' user. It makes you a non-user. People who have not registered are not counted in the statistics.

      We're building the house of the future together.
Re: Website Users
by caedes (Pilgrim) on Apr 03, 2005 at 19:08 UTC

    I would say that the 81% figure is very typical for websites of this genre. I have a reasonably sized website for graphic artists and photographers with about 133k users over the past 3-4 years. I also see exactly 81% 'dead' users (accounts not used in the last 30 days).

    I would say that the percentage of active users is going to mainly be dependant on the time that the site has been opperational as well as the site genre (to a lesser extent).


Re: Website Users
by cog (Parson) on Apr 03, 2005 at 19:44 UTC
    Take a look at the second graph in this page.

    You'll notice that out of 4263 PAUSE ids, 1819 were never used to upload anything. That's over 42% inactive accounts.

    Not exactly 81%, but that's probably because a PAUSE id has to be aproved, while a perlmonks account hasn't (which is a bit scary... almost 2000 people had plans, announced them, got their ids approved and... gave up?)

    But I don't think there's anything to worry, though. I have subscribed sites myself that I never got to use, has have many around here, probably.

      Did anybody else notice someone already released a module in 2026? Interesting...

      Zak - the office

        I used to know which module that was...

        The problem arises because whatever is parsing those modules just pays attention to the time of the files in the distribution... and apparently somebody thought it was a good idea to have his/her computer ahead of time :-)

      And did you see the last graph? There is one dist being uploaded in 2026! So far for statistics.


      "If you have four groups working on a compiler, you'll get a 4-pass compiler." - Conway's Law

Re: Website Users
by CountZero (Bishop) on Apr 03, 2005 at 21:51 UTC
    And what about the Anonymous Monk? It is just one account behind which can hide an unknown number of people.


    "If you have four groups working on a compiler, you'll get a 4-pass compiler." - Conway's Law

Re: Website Users
by demerphq (Chancellor) on Apr 04, 2005 at 06:56 UTC

    Every now and then this issue comes up amongst the gods and pmdev and we usually say that we will dispose of the "dead" accounts and then never quite get around to it. The only reason I personally would like to nuke those accounts is because usernames are unique and those 20k id's that are lost to the logged in once never posted crowd are thus unavailable for real users, and personally id prefer real users have priority for the cool accounts.

    One of these days one of us will have to wipe out those accounts.


      Just as long as you leave Friar paco and his famous post How do I recursively process files through directories :-)


      Added Friar paco's honorific in agreement with Anonymonk. Guess I will have to keep an eye on this node now and update as he progresses through the ranks (Friar paco that is, not Anonymonk)


      Pereant, qui ante nos nostra dixerunt!
Re: Website Users
by artist (Parson) on Apr 03, 2005 at 23:06 UTC
    You might want to see perlmonks from different lights, which I rate very high. How many times user get satisfactory answers for that question ? What encouragement and directions users get from the site etc.. For me, personally, 2660 (logins in last 30 days) is a sound number, who want to help me with Perl. Community is, and should be for the user, and considering that aspect, Perlmonks has higher value compared to other site IMHO, regardless of stats you mentioned.
           I agree PerlMonks is a great site as I have been helped by those who have a far greater understanding and insight into topics than myself on many occasions and helped fast too.
           I was not trying to cast a negative light on PM or the frequency of use, quality or material or users etc, it just got me thinking about how user activity in general on the web is. I know when you take into consideration the genre, quality of information and whether you can access this information without logging in will skew the results, but am curious to see if there are any similarities.

      Regards Paul
Re: Website Users
by TedPride (Priest) on Apr 04, 2005 at 06:02 UTC
    I haven't yet seen a properly descriptive question that hasn't been answered comprehensively within a matter of hours. Sure, a lot of people never post, or only log in once to post a single question, but we don't need the extra spam anyway.
Re: Website Users
by Anonymous Monk on Apr 04, 2005 at 02:36 UTC
    Looks like whomever maintains it added the percentage displays.
Re: Website Users
by halley (Prior) on Apr 06, 2005 at 02:43 UTC
    For a couple of years, long ago, I ran a quiz channel on IRC. It was a very popular quiz game, offering a lot of features not found in most: statistics per player, per round, per day, per month; user privilege flags to add, edit, moderate, and manage the question database; in-game message retention called gmail; questions in various categories and special rounds with funky rules. Oh, also written in Perl. ;)

    You didn't have to "sign up" to get an account; the bot would recognize you by nickname for basic statistics and would learn any new player as soon as they answered a question in the course of the game. The focus was ease of involvement for players. You had to set up a password and authorize to use special features of your account, though.

    With the transient nature of IRC, and the network involved, we were getting 800-1000 new player accounts per day.

    Since accounts were as cheap as candy to create, and I didn't want newcomers to build on the statistics of old unused players, I implemented a tiered pruning mechanism.

    If you had less than 5 points (easy first-day total) and you hadn't been seen for a week, you were erased. These players probably wouldn't even notice that the point system existed, nevermind care, if they returned.

    Other tiers were also measured on how many points per month of inactivity, and/or any special clues such as special privileges. Over a certain threshold, the game system just sent a message to an administrator to unlock the user for reaping, and wouldn't reap otherwise.

    Back to perlmonks (and other create-an-account websites): I am kind of annoyed that all the "logged in once, no writeups" usernames are off-limits for other people who could be great contributors. I would like to see the site prune most of the ancient non-contributors: if they have not logged in for six months, have not written any nodes, then reap the username and let someone else choose that name.

    [ e d @ h a l l e y . c c ]

Re: Website Users
by webfiend (Vicar) on Apr 04, 2005 at 17:34 UTC

    I'm sure there's all sorts of interesting statistical anomolies ... anomalies ... anamal .... weird number stuff. In addition to the one-time posters and the never-posted, you've got folks like me. We go into PM hibernation for a stretch of weeks or months, coming out of it long enough for a relative flurry of activity. Then we see our shadow and it's back into hibernation for another 3 months.

    I guess it takes all sorts.

Re: Website Users
by jdporter (Canon) on May 08, 2006 at 14:17 UTC
    ...dead users (81%)

    How do you figure?

    The only combination of those user stats that I can see gives 81% is "Logged In Once, No Write Ups" plus "Users With Write Ups".
    19998 + 7177 / 33512 = 81%.

    Somehow, that's not the membership slice I'd call "dead".

    We're building the house of the future together.

Log In?

What's my password?
Create A New User
Domain Nodelet?
Node Status?
node history
Node Type: monkdiscuss [id://444513]
Approved by Corion
and the web crawler heard nothing...

How do I use this? | Other CB clients
Other Users?
Others drinking their drinks and smoking their pipes about the Monastery: (4)
As of 2021-12-06 08:11 GMT
Find Nodes?
    Voting Booth?
    R or B?

    Results (32 votes). Check out past polls.