| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: Suggestions / Feedback
Thread: Misleading Global Statistics |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 17
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The "Statistics" tab shows Global Statistics by Team and by Members. One would assume that a member is an individual, on his/her own or a member of a team. If you check out the Top Crunchers, you'll find that #1 has over 3,000 devices running; same for #2, #3. Sounds like a team to me.
I think we should have apples to apples comparisons. It will give the little guy like me who joined a team and is ranked 57,000 as a Member (3 devices) a chance. Before anyone tells me, yes indeed we are crunching for humanity. but WCG purposely set up rankings to generate some fun competition. Happy Crunching, Sid Gibson |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello Sid Gibson,
A member might be an individual, contributing his own computer time, or it might be an entire university department, contributing the spare departmental computing capacity in accordance with an official policy. Either way, we are grateful. I have recommended that we revamp the statistics to show active devices rather than registered devices for most purposes. If this happens, you will be able to see how you are compared against members with an equal number of computers and can stop comparing yourself to members with ten times as many computers. Lawrence |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
members with ten times as many computers. More like 3,000 times |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
This is true. But sometimes, one individual does have a mind-boggling number of computers to contribute. This is partly because it doesn't make sense for, say, a school to create seperate member accounts for each computer. It's much easier to register them all with the same username.
I see the problem, but I fail to see a solution. The reason is, where do you define a cut-off? There are perfectly individual individuals with dozens of computers. One possible solution would be to firstly fix the discrepancy between registered and actually active devices, then allow comparisons between people with the same number of computers. I know SETI@Home compares individuals with a single computer, and it's a lot fairer than comparing arbitrary numbers of comuters. What do you think? |
||
|
|
davidhobbs
Senior Cruncher England Joined: Dec 30, 2004 Post Count: 152 Status: Offline Project Badges: 
|
Didactylos,
I agree very strongly that only currently active devices (say those that have returned a result within the last four weeks) should feature in the statistics. I am somewhat embarrassed that the number of devices credited to me is almost twice the real number. Similarly, members should only be included in the statistics if they have at least one active device as defined above. I think this would give a much more balanced (and probably more impressive) view of how the Grid is performing. David. |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7849 Status: Offline Project Badges: 
|
I agree with Mr. Hobbs. In addition some provision should be made to keep including devices which have returned results but been retired due to any number of reasons. Those devices which are still listed but never returned a result could surely be eliminated. I understand this would be a low priority item because the programmers are tied up with more important things to do. Just a thought.
----------------------------------------Cheers Sgt.Joe Minnesota Crunchers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I clearly see your point on a school member ID. I support the plan proposed by lawrencehardin to create a "like-for-like" comparison. I have 3 devices at 100% throttle running 24/7--I'm sure other members who have a competitive spirit, but have only one device at the default 60% setting, would like to feel like their "in the game".
Please let me repeat, the stat comparisons are fun, but the main reason for the amount of resources I donate is a commitment to help rid the world of deadly diseases--I'd like my granchildren to think of "Pop-Pop" as a humanitarian. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Since Sid belongs to the same group (SeniorNet) as this registered device, I have looked at his stats - he says 3 installations and the WCG says 4. I am finding what I consider to be MAJOR errors in the algorythms that are computing statistics - just within the SeniorNet group. A single device reported more than 2 days run time in the first day of registration along with inconsistancies like 3 results returned, avg. 1 result per day, with NO results returned in the last 22 (as I recall) hours. The actual return was 1 result as reported the next day. I also find the average points per result to be suspect with a range of ~100 to 600 points per result amongst the 13 members of the SeniorNet group.
|
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7849 Status: Offline Project Badges:
|
SNLCSJwillows-VM:
----------------------------------------I don't know the answers to all your concerns, but I can tell you the number of points per result can vary widely, depending on the size of the work unit. As I understand it the number of points is proportional to the number of floating point operations performed or to put it another way, the amount of work done. I think it would be illustrative if you specify the agent(s) being used, the operating systems used and the types of machines being used and which projects are being done. The number of devices in use is NOT a good indicator of the presently active devices. I have seven devices listed, but three of them resulted from abortive attempts to get BOINC running, which I was eventually able to do. It appears as though once a device is registered with the system, it stays listed forever. Perhaps one of the CA's will chime in with a more definitive answer. Cheers Sgt.Joe Minnesota Crunchers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I was formulating a different response, and checking on some details for a couple of the group members when I found that the questionable results come from those that obviously have been on the grid for some time and have recently joined the Seniornet group. Looking at the individual characteristics, the calculations DO NOT come close to matching the result reported for the group. The individual report shows ~300 points per result while the equivalent reported for SeniorNet is ~639 points per result - doesn't make sense, especially when it yields 572 results per day for SeniorNet versus 1.55 for him as an individual.
THERE ARE PROBLEMS. |
||
|
|
|