First it would seem obvious that the largest sites would be the ones with the most members This however is just too easy to be fudged with phantom registrations. There are even plug-ins(code) to produce test (phantom registrations) that this simple analysis of site size needs to embellished upon.
So if we then add online users into the mix we will get a better feel of the largest elgg powered sites. This however produced another flaw.Lets look at Brighton University as an elggample
When I started studying the figures from Brighton University, mostly presumed by all and sundry to be the biggest elgg powered site (66,000+ registered users and quite often with over 300+ logged in, something strange came to my attention. These 300+ so called logged-in users took well over 24hrs to fill one page on the activity river. WHAT ?????????????????? clearly something is not right. So we now also have to take site activity into the mix.
So we now have our criteria for analyzing elgg powered web sites is
Registrations, users online, and site activity
as far as these criteria are concerned, site activity is clearly the most important.
It is the only criteria that denotes that the site is working and being used. Again when studying Brighton's stats our best guess to explain the large number of online users compared to the extremely low site activity suggests that anyone wanting to use a computer at the university must sign in which also counts them as being logged-in to the elgg powered part of their web site even though they never even visit or utilise that part of website (Brighton is not just an elgg powered site but has many other areas as well.
Some interesting stats on the sites that where analyzed
66,000+ registered users (*1)
Average daily registration increase 1.9
Ave online users (Peak times) 300+
Ave site river time (peak) >19hrs+
3300+ registered users
Average daily registration increase 27
Ave online users (Peak times) 12
Ave site river time (peak) 2.5hrs
27505+ registered users
Average daily registration increase 42 (*2)
Ave online users (Peak times) 15
Ave siteriver time (peak) 2hrs
70,000+ registered users
Average daily registration increase 300+
Ave online users (Peak times) 120+
Ave site river time (peak) 2minutes
1* Brighton's registrations appear to be force as well as there logged on count
What does this mean Well at an ave of 2 new registration/day how did they acquire 66000+ members Quite simply anyone who registers at the university itself must clearly be given automatically a community registration also It appears that to utilise any part of the universities computer system (google the net etc) will automatically log you into the community even if you never go to the elgg part of there site (Expect a sudden increase of 20,000 new user Aug Sept time 2010)
Registrations at the the elgg powered community part of elgg.org have been falling drastically over the past 8 months and the community itself is shrinking.
In Aug 2009 elgg's average increase of new registered users was approaching the 75+/day mark, by Nov 2009 this figure had dropped to 55/day, and as of late February 2010 this figure had dropped again to 42/day, almost a 50% decrease. This maybe a response to the shrinking of the elgg.orgs site itself. Over the past 10 months elgg has been systematically amputating parts of their site. They have removed "the wire, "twitter", blogs, the "about me" text area in the profiles, etc.. Quite simply the only things they have left are groups and pages.
Why is Elgg.org doing this ? Well their are many possible answers and I may write a further Blog post on this, however it would be purely speculation.