On Oct 1, 2009, at 12:37 AM, Raistmer wrote: >> Which means, for what little it matters, that those metrics violate >> the standard as defined in the BOINC documentation >> >> A moving average is not a standard ...it is however a way to deflate >> the credit awards over time.., why this is seen as a good thing has >> still not been explained. > > Yeah, interesting point. > It seems such method used by credit-correction script on SETI change > absolute value of credit indeed. > This make any historical comparisons based on RAC impossible :( > Not sure it's good thing indeed (it makes whole credit thing useless > as performance estimation IMO). > > The only positive aspect I see it's more "fair" treating of oldest > project members. > For example, I participate in SETI (though under another account ) > from ~few months later its beginning. And in BOINC-SETI from its > start. One of my those PCs still online. > But what I see: > its TOTAL credits were outnumbered by my new quad after less than > half-year of quad operation. > In credits aspect it seems somehow discouraged. > > But if we consider credits not as performance-measuring tool but as > social engineering tool that provides additional MOTIVATION for > participants, then recalibrating credits/sec vs FLOPS on current > median looks VERY senseful. It still enables concurence (if I have > PC over mean I get more credits and go up on ladder), but also it > values hard work of early days participants much more then constant > credit rate giving them good feeling that all their work was needed > indeed and they made right choice to spend 1000$ on PC ten years ago > and not just wait these ten years and spend same 1000$ on order of > magnitude faster PC now. That is, such credits put more value in > veteran's work. > Not bad, even good, from social engineering point of view.
Well, I don't even look at RAC it is unstable and other than self comparison it seem pretty hopeless. As far as a historical compare on RAC even more useless. The economic uses of credit awards and using it to guide behavior is pretty much been abandoned by everybody but the participants ... The problem is that Eric's script will mean that over a 4 year life of a computer the CS it earns in year four will take more processing time to obtain. That is the flaw ... as the performance average increases the award goes down. So the CS I earned when I started BOINC 5-6 years ago, what ever it was, took less processing power over time than the same CS today (all other things being equal). That means that the earnings of a new person cannot be "fairly" compared to mine. _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
