Thoughts On Credits
6 Aug 2008 13:32:15 UTC
Topic 193818
(moderation:
Quote:
Dear Bikeman
My average in S5R3 was around 15.316 sec. and this was awarded with 237 credits, now i need 24.518.85 sec. which, if awarded same should be 382 credits. As in the first crunched examples visible the new awarded credits are around 160 credits instead of 382 which is less then 50% . This leads to the conclusion that it is no longer desireable to crunch for Einstein because obviously we get cheated for more then 50% of the credits.
I only crunch about 12 work units per day but you will miss them in future and i think if not changed the credit politics more will follow.
have a nice day.
A comment on your wording here, not upon the content of your concerns. Your point is otherwise well made, but the word 'cheated' denotes and connotes deliberate theft as a criminal act. It is likely to be a cause of inappropriate trouble if not flagged, as it is the type of word that can give the appearance of offence even if none is intended ...... as I trust you did not. :-)
All please note that credit issues are a valid topic for discussion, but watch the phrasing please. I apply the 'duck test' to moderation, so if you look like you are breaching the rules [ that you see on the left hand pane when you compose messages ] then I will deem as such.
So please all do continue with your polite and robust examination of any aspect of E@H's performance. Rest assured that all feedback is genuinely considered as valuable - even though you may not get the result/response that you expected.
[personal view]
Actually I've seriously & often wondered if cross-project credits are usefully meaningful. I think it really compares apples, oranges and deep-sea squids. In the long run it may be better to define credit currencies - E@H credits, SETI credits, Cosmology credits etc. What/how/when contributors think of cross project parity is then their own pleasure, you'd be free to define whatever exchange rates you like. After all, not all 'fly-buy' schemes are equivalent for much the same reasons.
[/personal view]
Cheers, Mike.
( edit )
[personal view]
Of course within project credits ( b/w science runs say ) are still requiring of attention - but you've now got a vastly simpler landscape of concerns, and with likely less mis-alignment.
[/personal view]
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
Thoughts On Credits
)
I'm not good at the math and graphs and stuff like others are, but my point over at SETI was I was starting to see where the mix of processors on a given project could have an effect on the placement of the bell curve. I think it will eventually be shown that if a project has a high density of Core2 systems, with their respective high benchmarks and quick processing times, slower systems will get the short end of the deal. If the project is more balanced, like this project is currently, both high and low will experience significant "pain", but those in the middle are relatively unaffected. However, if the project has a high density of older/slower hosts, at least temporarily a faster host can attach to that project and get a quick boost, as the median will be lower, and thus their significantly faster system will be on the flatter portion of the curve, towards the upper extrema (woah, 50 cent word)...
I think, ultimately, all this will have accomplished is to irritate the rank and file, while the true "credit chasers" will figure out how to work the new paradigm to their advantage... Meanwhile, the "doing something to 'fix' this 'issue' is better than nothing" crowd still feel that their way has made a change for the better...
What I suspect is that since all projects have not done this, there WILL now be migration based on credits. There wasn't before, at least not to any significant extent.
As for me, unless things look significantly better with my P4 tomorrow evening, I'll just have it spin looking for LHC tasks and be idle at all other times, saving a significant amount of electricity. I'll then need to decide what to do with my AMD. I may retire it as well...
I'll bravely predict, in the
)
I'll bravely predict, in the long term, one/both of two likely outcomes:
- projects will go off whatever the gold-standard-bread-basket is. One will break away, then the rest ......
- contributor's will increasingly apply, even more than they do now, some non-credit metric as a reward for their contributions. Probably they'll accentuate their use of assessments of the science merit.
I say this primarily because:
- credits aren't really worth anything in a material sense. It can be hard for all concerned to maintain enthusiasm for what is becoming an increasingly nebulous concept.
- many now associate the cross-project credit topic with a negative taste.
- you can't please everybody, or at least it is becoming harder to try to.
- an emerging realisation that energies truly are better spent elsewhere.
It would possibly be suddenly triggered by: ET being found, a gravity wave wiggling, the perfect protein gets folded .......
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
Maybe counting credit is like
)
Maybe counting credit is like counting grains of sand -- a few grains more or less is not much to stew about. Nonetheless, credit pros/cons aside, just for my own benefit I have a compulsion to know how much I contribute to a project and how much I contribute to project A versus Project B. Resource share is worthless for this purpose, should I put a watt meter on my machine and count watts or dollars or hours or tasks or days? If I find a gravity wave I won't care what the contribution is because the discovery justifies my cost.
For a long time I ran Rosetta and Einstein at 50/50 resource shares. But since Rosetta grants considerably less credit one might conclude from my credit totals that I contributed a substantial amount of my resources to Einstein. This bugs me whether this feeling is logical and warranted or not. Then there is cosmo which awarded credit up the wazoo. You would think I contributed a lot of resources there but not so much. There is no way of realistically assessing my contribution to these projects. Credit is worthless to be sure, but I do spend time stewing over this worthless entity because it matters to me and apparently matters to most volunteers. And though Mike Hewson, Gary Roberts or Bikeman are dedicated crunchers without regard to credit, you must accept the reality that many or most volunteers are credit conscious.
RE: I'll bravely predict,
)
This is what you get for having it be a voluntary system. The mechanism exists and projects can choose whether or not they wish to attempt to implement it. If you have a single holdout, and said holdout ends up granting "more" than "everyone else", that project will be demonized by certain individuals, all the while the real "credit chasers" will have figured it out and will just have their machines there.
It is too bad David Anderson can't come to that same conclusion. There was no "credit war". There was no large mass of people selecting a project based on credit vs. the other projects as the stood before "The Grand Plan". Now, my brave prediction is that you will see either large numbers of people moving to projects that haven't instituted this "plan", or people like me who have gotten to the point of realizing that the current "science" at nearly all projects is not worth the cost of electricity for systems older than about 2 years old.
* The exception to which is Orbit, as the search for Exctinction Level Event-causing comets / asteroids is fairly important, IMO...
RE: you must accept the
)
I actually do, but I just don't see it sustaining. I'm not criticising any that do - whatever floats your boat and urges you to contribute is fine by me - the science output is insulated from that. Win/win. I'm talking at a different level of detail. What was no doubt a terrific concept used to kick off the DC thing will run it's course. People need milestones when they volunteer, hence the utility of credit for that purpose. I just feel it will be superceded, morphed .... whatever, in the long term to some other measure. For example, if say GW's get regularly picked up then people will mark their value with 'my computer got a touch of that one' or somesuch. A milder version of that can already be claimed by us all, in that upper bounds on certain celestial scenarios have been set by work to date ( see the S3 and S4 reports ).
There are some pretty long term goals for many DC projects ( decade time scales ), and most are consuming large slabs of the working lives of the professional scientists involved. I think if DC is going to dominate/persist as a paradigm for public science collaborations ( wow, do the author lists of published papers have length ) then simple credit as currently designed doesn't have the legs.
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
RE: * The exception to
)
There you go, you have accentuated your use of assessments of the science merit :-)
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
Moved from Probs & Bugs. We'd
)
Moved from Probs & Bugs. We'd forked topics - partly my fault.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
RE: ... And though Mike
)
Hey!! I resemble that remark :-).
Dr, Dr, please help ... If that is true, then please tell me why it hurts so much if my RAC drops a few K :-).
I'm as credit conscious as the next person but I choose not to make a fuss about it since my challenge is not so much against other crunchers on this or other projects but rather against myself.
I think my biggest sense of satisfaction comes from tweaking somebody else's castoffs (computer-wise) and making these old junk boxes really perform. I would look at my RAC every day and figure out what I was doing right and what I was doing wrong based on the trends that I'm seeing. Learning how to really overclock Tualatin Celerons and then taking advantage of the Linux app in the latter stages of S5R3 are good examples of things that gave me a sense of satisfaction.
At the end of the day, the quantum of credit I've accumulated is not all that important (although nice to reflect upon). It's the knowledge that I've made a contribution to the advancement of science at the same time as having a bit of fun and also at the same time as having done something to help others (hopefully) that probably adds a further level of satisfaction to something that others might deem to be an unhealthy obsession :-).
I've wandered off the point. The point is that the project admins will adjust the credit awarded in what they deem to be as fair an equitable a way as possible. The following comments are not directed against any person but moreso against the ideas expressed in several messages about unfair large drops in credit awarded. Making threats about the consequences that will ensue if the admins don't "fix" the credit wrongs is rather pointless. The admins are reasonable people who will do their very best to satisfy as many people as possible. People need to simply "cool it" until we see how the new project settles down. Making outlandish early claims of a large X% drop without acknowledging the fact that we have had a very nice "bonus" period throughout a lot of S5R3 is not really the best way to win people over to your point of view. We all know and have been forewarned that adjustments are to be made to lower the too generous rate of credit in the latter stages of S5R3. We need to take that into account when assessing what the new rate of credit looks like.
Cheers,
Gary.
RE: Making outlandish
)
Well human nature being what it is... government clearly understands the psycology of instituting policy by giving something to the citizenry to gain favor and power because once given you have a war on your hands if the other party later tries to take it away. The so-called "S5R3 bonus" has become the defacto standard!
I think that cross project
)
I think that cross project equalisation is important, if for no other reason that to be able to assume that my 2.2 million BOINC credits are appoxiamately equal to Mike's 2.3 million.
We knew as Gary pointed out that Einstein's cr/time was going to come down, about 25% at the end of S5R3, and to be in line with Seti.
And those of us that also do Seti know that their rate is going to come down as they have calculated they are overpaying with the default app by about 15%.
So one has to be prepared for an average reduction of over 40%, probably a lot more if your OS is Linux. And also have patience and wait until the bosses have studied the situation to see whether there will be any further credit adjustment.
Based on performance of my computers, I think the present granted is low and my guestimate is that S5R4 units should get 260 +/- 20 credits.