| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Community
Forum: Chat Room
Thread: The Average WCG Cruncher |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 15
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Many posters have freely bandied about the term 'Average User'. Some have used the more offensive term 'Normal User', which I will ignore. So I thought I would do a little research into the current WCG statistics to try to nail down who is the Average user. Note all statistics were retrieved on 24Jul09 and are taken as presented by the WCG and BoincStats websites. I know that the statistics updates are not in sync with each other, but this is the best information a cruncher can get (i think).
The WCG stats report the following for 23Jul09: Points Generated 218,399,982 or 3,199,999 credits. Boincstats reports the active WCG user count is: 102,298 but on more detailed investigation it reports 55,027 users reported at least 1 credit yesterday (or 7 points or 14.4 minutes per day). Therefore the Average User could be defined as crunching 218,399,982 points / 55,027 users or 3969 Average points per day (567 credits). Now you can argue that the top 25 monster crunchers, and their 26,199,138 points, inflate the stats, but you can also equally argue that the 989 users that crunched less than 70 points (1 hour) deflate the statistical average. If you reduce the total points and users for the top 25 and you reduce the users by the 989 smallest, you get (218,399,982-26,199,138-98,900) points / 54,013 or 3,556 points per day (508 credits). This can be loosely translated as 2 days and 2 hours per day. There can be no doubt that the Average has significantly increased over the last year since every hour crunched by EVERY user increases the average. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
There can be no doubt that the Average has significantly increased over the last year since every hour crunched by EVERY user increases the average. Good! |
||
|
|
stwainer
Advanced Cruncher Joined: Nov 21, 2005 Post Count: 128 Status: Offline Project Badges: 
|
Can you figure out what the median is? It might be a better indicator than the average.
---------------------------------------- |
||
|
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3716 Status: Offline Project Badges: 
|
The median is an interesting element, and I think that true statisticians (I am not one) would like to introduce some percentiles at the top and at the bottom of the list but this would be much cumbersome considering the sources we use.
----------------------------------------One simpler thing which will make a brute computation less arguable is to do it on a 30-day period of time. An active user for BoincStats is one who has got credits over the last 30 days. You have found 102,298 which is consistent with what I have found the last time I wanted it. Over the 30-day period going from June 24 to July 23 WCG has "produced" 6,546,500,025 points, which gives 6,546,500,025 / 102,298 / 30 = 2,133 points or 304.7 credits. This result is not so different from yours, the good number (if ever there is one) is probably somewhere around them. But it confirms that 1. it is not that simple to give a right number 2. the average member's daily contribution is far below what one might think when reading our forums every day. Regarding how to name this normal statistical member I don't know an ideal name. Everything one can use in similar situations can be interpreted in a pejorative way by anybody who wants to or who is not in a good mood: normal, average, basic, lambda, plain, simple, etc... Personally if/when I type "normal member" I am thinking of the target (not nicer!) participant that WCG and other grids are addressing: - somebody who wants to help with its own computer(s) - while it is normally switched on (i.e. the same way as it was before starting to participate) - by offering the CPU power not used by its machine(s) - without having to take care of it or be disturbed by it. Strictly speaking all the above means - no 24/7/365 operation if this machine was not on 24/7/365 before - no dedicated machine - reduced CPU usage (%) if the machine is too hot or too noisy - no need to haunt forums to get it working. With the above I do not mean to exclude anybody (me first) from being a normal member, I am just showing why I have neither bad feelings nor a particular level of contribution in mind when I am typing "normal user". For me normal members are the very wide (and silent) base of our community. Jean. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Interesting.
I don't think you can discount the 47000 active members who *didn't* return a result yesterday. That just means their contribution (mostly) is less than one result per day. Recalculating the mean, that gives 2135 points (305 credits) per day (the same as JmBoullier got with an entire month of data). The top ten contribute a lot, but in fact the head of the data shows an exponential relationship - the top 1% produce 40%, the top 0.1% produce 20%, and the top 0.1% produce 10%. Coincidentally, the top 0.1% are the top ten members - and they produce 10%. I don't think there's enough data to calculate the median - and that's a pity, since I believe it is the most useful average we could calculate. We can calculate the mode easily, though: it's those 47000 members returning less than one result per day. |
||
|
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3716 Status: Offline Project Badges:
|
I don't think you can discount the 47000 active members who *didn't* return a result yesterday. That just means their contribution (mostly) is less than one result per day. It may even be more. A dual core crunching in the evening only can well need two days for crunching a WU, and if both cores are more or less "synchronized" it will return two results every other day. Theoretically it's true also for quads (or higher) but 1. the probability that all cores be starting a WU about the same time is lower 2. the probability that a quad runs only a few hours in the evening is lower too, I think.  Didactylos, can't we determine the median from BoincStats? As I said I am not a statistician, therefore I may misunderstand something. I would believe that it is the number of credits got by the 51,149th member in the list sorted on Credit/month. That would be 1,886 credits right now, i.e. 13,202 WCG points per month, or 440 WCG points per day. But again I may be wrong. Jean. |
||
|
|
mikaok
Senior Cruncher Finland Joined: Aug 8, 2006 Post Count: 489 Status: Offline Project Badges: 
|
I think I have seen some statistic about computers used to return BOINC results. It might been in boincstats.com... If those computer feature statistics were compared to median CPU times and points, you would probably get, at least, nice statistic of the average cruncher. That way you would also get some information about the wall clock time used for crunching as well
----------------------------------------
to infinity and beyond
---------------------------------------- [Edit 1 times, last edit by mikaok at Jul 25, 2009 11:30:33 AM] |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7854 Status: Recently Active Project Badges: 
|
Didactylos, can't we determine the median from BoincStats? As I said I am not a statistician, therefore I may misunderstand something. I would believe that it is the number of credits got by the 51,149th member in the list sorted on Credit/month. That would be 1,886 credits right now, i.e. 13,202 WCG points per month, or 440 WCG points per day. But again I may be wrong. Jean. No, you are right on. Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I don't think you can discount the 47000 active members who *didn't* return a result yesterday. That just means their contribution (mostly) is less than one result per day. I have two points to make. If you are going to include the 47,000 (that maybe returned a result yesterday) then you have to include ALL the points from yesterday and divide by 2 days. You cant just add the crunchers without adding points from those crunchers. The second point is that some portion of the 47,000 have actually quit yesterday (the actual count may be impossible to determine) so those that quit are pulling the average down, when they should not be included. |
||
|
|
Ingleside
Veteran Cruncher Norway Joined: Nov 19, 2005 Post Count: 974 Status: Offline Project Badges:
|
The second point is that some portion of the 47,000 have actually quit yesterday (the actual count may be impossible to determine) so those that quit are pulling the average down, when they should not be included. A few likely has stopped, but these has been offset by new users that haven't got any validated results yet... Seeing it can take multiple days before results is validated, due to waiting on the wingmen, saying only someone that's got credit last 24 hours is "active" will grossly mis-represent how many is really active. A more realistic minimum is active users last week.  "I make so many mistakes. But then just think of all the mistakes I don't make, although I might." |
||
|
|
|