Point inflation - Longevity Credit

[Bob_6591]

Just started folding 11 days ago. Within a week I passed the founding member of our team with my meager 150 or so work units. He had about 1,600 WUs. There are a couple of people with 8,000 WUs down the team list. There is a lot of point inflation going on. The current point system devalues the work people did in the past. I am not suggesting changes to the point system. Instead, let me suggest having an additional column on the Team Statistics and All-Time statistics page to honor people that went before. Maybe a symbol with the number of years the donor was active.

Cheers

[JimboPalmer]

My first GPU had 16 shaders, of course it did less work than a modern GPU with thousands. My Semperon did less work as well. This is not point inflation, this is more capable hardware doing better science.

[BobWilliams757]

Though it's actually more tech advancement, the idea is kind of good. When I look at the EOC stats and see people still plugging along with older gear but massive WU counts and/or reasonably high points, it makes me reflect on how much money some might have spent to get to that point on the older and less advanced gear.

With hardware advancing so quickly, any given amount of money goes just so much farther with modern gear that it's almost like a penalty for those that spent their money too early in their folding years.

[Lazvon]

Hmmm. Bob William’s comments made me think. At first I was going to reply that a “point” is something that relates how far along the research was driven ahead. I don’t know much about the science at all, but always assume a point was essentially given when the proteins were “folded” to a new alignment and checked.

Essentially, I was going to reply that science was advanced relative to the number of points gained. I have only been folding for real since Jan 2021 due to COVID needs. My account/key much older but think I barely even set up and decided not to fold a decade+ ago, but at that time it would have been single card and probably very expensive for the time.

With Bob’s comment… I don’t think there is a time-series of when the submissions points were granted… but if there was, would agree that some way to assign “donated value” points wouldn’t be bad to see. Main problem is energy costs. While I pay $0.055/kWh, someone in California spending a LOT more, and someone on solar may or may not know their per kWh because it was mostly a one time capital investment for them.

Would need to know average $ for the card series doing the submissions too, and there might be reeducations in that $ over time too… and what is the lifetime of a card to divide it by on per year/month/day/hour basis anyway? I may replace my cards every 3-years, others ever 1- or 5-years.

I replace my main gaming rig (which also folds when not playing a graphic intensive game) every time a more powerful card/system comes out… roughly every year, though 4090 looks like it will stay for a couple years. Then the old card/system gets moved to the folders. Other folders I bought completely new for the efforts.

Speaking of dedicated donations to represent, what about the thousands in cash I donated directly to the charity organization? Does that count too?

Interesting things to think about.

I think just thinking about the points representing “how much farther science was furthered” versus “how much was donated” is best and easiest.

[Bob_6591]

My first GPU had 16 shaders, of course it did less work than a modern GPU with thousands. My Semperon did less work as well. This is not point inflation, this is more capable hardware doing better science.

Point inflation is the result not the cause. I would agree that more capable hardware is part of the reason for the point inflation. However, the main reason we see point inflation is the way the F@H team has setup the system.

The addition of bonus points was one reason for the increase in points. “So in 2010 we introduced the Quick Return Bonus (QRB), which gives extra points to users who rapidly and reliably complete WUs.” Faster hardware more points. https://foldingathome.org/faqs/points/b ... ts/?lng=en

The second reason was that the F@H team is using an i5-750 as the base. As hardware gets better you will see an increase in points per work unit.

What if they gave each person only one point for each work unit and no other bonuses? The inflation rate would decrease significantly.

I do not disagree with the way they are calculating points. It is actually a clever way to reward people with fast hardware. Lazvon had a very impressive 24hr Avg of 69,476,444 points. On the other hand, we have 7 teammates that averaged less than 10,000 points (24hr Avg). At 10,000 points per day it would take each of them 6,947 days or 19 years to reach 67 million points. It is sort of like trying to dig a new Panama Canal with a hand shovel while some people have 7 Bucyrus RH 340/RH 340-B excavators.

The discussion has helped me decide to shutdown PC3 next summer. Right now the heat generated by the PC helps warm the house. Given that the GPU/system is not that efficient it makes sense to turn it off to save on air conditioning costs.

Cheers

[[Ars] For Caitlin]

Point inflation is an interesting topic. As someone who was able to take advantage of a job that allowed me to fold on dozens of enterprise class servers back in the advanced methods big units days, I was amused to see that a consumer GPU could outproduce those boxes that cost six figures. Points will continue to grow at an exponential rate.

[BobWilliams757]

I think just thinking about the points representing “how much farther science was furthered” versus “how much was donated” is best and easiest.

Either way we look at it, over time the inflation takes place. Though I agree that "how much farther the science was furthered" is valid, I think the original point made still makes sense.

That was the reason I somewhat view it as "how much was donated" since the two tie together through hardware advancements. As an example, if you took your total financial contributions and applied them 10 years ago, your points total would be significantly lower, even though the effort was the same or greater.

If points/science advancement/whatever we want to call it inflation continues at the same rate, newbies might some day be passing your total score and thinking "wow this guy only folded 35 billion points, and it took him 100,000 work units". If there were a way to signify levels of contribution over time, they might be more likely to realize the true cost of those that folded years before they started, and that the advancements in hardware give them a huge advantage in throughput and thus overall standings.



How to implement such recognition I have no idea really, but I think it's not a bad idea at all.

[Lazvon]

Oooh, forgot about the WU counts.

Maybe there is some relation with points/WUs there to show what you are looking for.

[toTOW]

A long time ago, some WUs were worth 0.6 points for 1.5 hours of compute on an Athlon XP 2000+ ... and there was no quick return bonus ...