Life In 19x19
http://www.lifein19x19.com/

Looking at DGS activity statistics
http://www.lifein19x19.com/viewtopic.php?f=49&t=10442
Page 1 of 1

Author:  Drew [ Fri Jun 13, 2014 12:07 am ]
Post subject:  Looking at DGS activity statistics

What to make of this graph? :scratch:

Attachment:
DGS_stats.png
DGS_stats.png [ 15.25 KiB | Viewed 10926 times ]


* Early users played an astonishing number of simultaneous games per capita
* 2007 saw the beginning of a long decline in activity despite more or less constant new account creation
* What happened in the summer of 2012? Games in-play went off a cliff
* Today the server activity is at its lowest point in a decade

Anyone else want to read the tea leaves? ;-)

Author:  ez4u [ Fri Jun 13, 2014 2:00 am ]
Post subject:  Re: Looking at DGS activity statistics

I haven't played on DGS in the last few years so I'm out of touch with the actual situation. However, the decline in running games without a decline in the slope of the games line would point to shorter games rather than a decline in activity IMHO. I think at some point the default time was changed from 90 (120?) days plus byo yomi to 30 days plus Fischer. Possibly that was the source of the sharp decline in running games in '12. Jug would know best of course.

Author:  jug [ Fri Jun 13, 2014 5:29 am ]
Post subject:  Re: Looking at DGS activity statistics

A look on the registration-counts of new users reveals a big peak between 2005 and 2008 of 9000 new users per year, which I guess also played a lot per capita.
Code:
| Year | RegCnt| Active |
+------+-------+--------+
| 2001 |   227 |     39 |
| 2002 |  1470 |     94 |
| 2003 |  3625 |    155 |
| 2004 |  5512 |    176 |
| 2005 |  9099 |    244 |
| 2006 |  9862 |    212 |
| 2007 |  9613 |    218 |
| 2008 |  7526 |    222 |
| 2009 |  6139 |    185 |
| 2010 |  5983 |    274 |
| 2011 |  8597 |    323 |
| 2012 |  7897 |    401 |
| 2013 |  5675 |    516 |
+------------------------
| Sum  | 81225 |   3059 |

Between 2006-2007 there was one user (Razorflame) which raised the running-games a bit by playing 1500 simultaneous games, which most over the next years timed-out because he couldn't play so much. So you can count one 10%-block of running games (which is about 2000 games) to that user alone. But the long decrease down can not be attributed to one user alone of course.

I can only speculate about the fall between 2007-2010: less users registering, perhaps doing other games/activities, using other servers, or just playing less games, etc.
Same goes about the rise between 2010-2012, which I can't explain without guessing about reasons for an increase.

In 2012/06 there was a server-upgrade with many new features, that increased the server-load of an already slow server further.
Additionally between 2011/03-2012/05 the number of RSS-requests were increased ten-fold from 1 Mio requests/month to 11 Mio (compared to ca. 2.3 Mio game-page requests/month that didn't change during that time) by polling RSS too often, which added to making the server extremely slow. The server back then had only 1 GB memory and we later found out, that because of that there was a lot of swapping with a lot of disc-I/O.
So during 2012/06-10 the server almost became unusable with a lot of max-connection-problems and net-timeouts. Because of that I think a lot of users played less. This could explain the steep fall till Oct-2012, which was abruptly stopped after a new and much faster server was sponsored and installed by the DGS-provider.

Also (like ez4u said) in the server-upgrade in 2012/06 the default time-limit was indeed changed (from Jap-byo-yomi 90d + 10x 1d to Fischer 30d + 1d). I agree, that this could be an explanation that let players play shorter games and therefore the total number of running games decreased.

I can also tell, that since the slow-server-problem in 2012 from time to time (every few months) I manually checked the number of active players. And this was holding steady between 3000-3300 users (being active within 1 week) during the last 2 years. So there does not seem a decrease in active users, but just that the per-capita games is decreasing. The small increase in 2013/08 is probably explained by the new ladder-tournaments.

All that being said, I'm not sure, that the total number of running games is a good indicator for the "health" of the server.
If you take a look on the webserver-stats http://www.dragongoserver.net/stat/ you can even see a slight increase of usage over the last year. Though it's hard to set this in correlation as an increase in web-hits can have a variety of reasons (same user base doing more requests for example). So there's an increase in activity, but hard to tell what's the reasons for it.

It's probably interesting what criteria makes a "healthy" state: number of active users, number of games, usage of features, ??
It's a todo of replacing this stats with more meaningful charts as the current one can be misinterpreted perhaps, though so far I'm not sure what make some better stats. Any ideas?

Would be nice to see numbers from other servers to have a comparison and see if there are general trends for the up and downs or some migration between servers ;-)

Author:  Drew [ Fri Jun 13, 2014 5:50 am ]
Post subject:  Re: Looking at DGS activity statistics

Outstanding post, thank you!

Would it be possible to elaborate on this:

jug wrote:
Between 2006-2007 there was one user (Razorflame) which raised the running-games a bit by playing 1500 simultaneous games, which most over the next years timed-out because he couldn't play so much. So you can count one 10%-block of running games (which is about 2000 games) to that user alone.


What on earth were they thinking? At one move per day and only 10 seconds of thought per move you're talking more than four hours of continuous go per day just to keep up!

Author:  jug [ Fri Jun 13, 2014 6:08 am ]
Post subject:  Re: Looking at DGS activity statistics

Drew wrote:
Would it be possible to elaborate on this:
jug wrote:
Between 2006-2007 there was one user (Razorflame) which raised the running-games a bit by playing 1500 simultaneous games, which most over the next years timed-out because he couldn't play so much. So you can count one 10%-block of running games (which is about 2000 games) to that user alone.

What on earth were they thinking? At one move per day and only 10 seconds of thought per move you're talking more than four hours of continuous go per day just to keep up!

Yes. During that time there were a lot of heated discussion in the DGS-forums (just search for Razorflame) and you find 8-9 threads about it.
In the end I think he accepted that he couldn't handle that many games, but for a long time he insisted he was able to.

But he was not the only one. 2 years ago there was another user with over 600 simultaneous games (though he managed at least to not let them time out).

These two cases made me implement some max-games restrictions (200 games) on DGS to avoid such "temptations" to play too much games at the same time.

Author:  happysocks [ Fri Jul 04, 2014 11:35 am ]
Post subject:  Re: Looking at DGS activity statistics

Interesting, thanks for helps place this into some perspective for us.

Still think this site has one of the loveliest binary boards online - and having received my first experiences with the game here will always hold a special place in my bowl. :tmbup:

Page 1 of 1 All times are UTC - 8 hours [ DST ]
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/