Folding Forum

As a game developer, I find FAH to be egregiously designed. You can't leave players so hopelessly in the dark about what anything means, especially when you have several different numbers all varying massively, each named a different variation of the same thing. What are points, score, and credit, and why is there such an absurd difference between them?

Base Credit is 6000
Est. Credit is 11842, at least they were until my progress bar filled up, then Base lost 100, and Est shot clear up to 26,010
I have a Score of 141,128, of which only 140,000 seem to be certified. Not sure what's wrong with the rest of it, but alright...
And Points Earned on the web control is 126,374,342, but my stats page doesn't seem to care in the slightest about points earned. In fact, points earned seemed to leap from 7000 to 60,000 when I registered a name, and then skyrocket from 64,000 to around 126,100,000 when I set a password.

And why do Points Per Day seem to fluctuate so turbulently? Where's my meta data? What is my CPU and GPU processing power even accomplishing? And why is none of this information available? I'd like to be able to look at a finished set and go "Oh, I contributed X amount of Y to Z." It may not seem like much to you, but I think if users could see their contribution milestones, FAH wouldn't have fallen so abysmally short of it's 1,000,000 user goal. Just saying...

FAH seems to be attempting to be gamified, but being a 23-year gamer veteran and a 14-year professional game developer, I can tell you that gamification wasn't handled even remotely well. Too much of it seems to be wholly meaningless. At least with FoldIt, you can actually see the fruits of your labor. (I will give FAH one thing, though, it doesn't try to make it into a contest, which is FoldIt's major downfall. People aim to beat everyone else's score and call it good enough, rarely ever shooting for the highest possible score.)

I've often wondered about some of the issues you bring up. Only many hours of skimming through this forum have helped me sort of understand whats up, but even after 16 years I don't fully grasp most of it.

In my head we're just calculating molecular trajectories of peptides as they approach secondary & tertiary states. I don't care about stats beyond knowing the optimum configuration I can achieve with my hardware, but would be cool if the client software included a comprehensive glossary with links or a guide - especially to help retain newbies and stats oriented folders like yourself.

One thing Vijay might want to think about is trying to connect the published scientific papers to the computational donations. To date we're at 153 papers. That's a massive achievement, but folding.stanford.edu/papers-results/ is buried in a drop-down menu. That info should be in one of the visual 'squares' on the home page that shouts RESULTS.

This is what I think I know, I am a long time donor with no insider information.

If you do not have a Passcode, you will only get base points.

If you do have a Passcode, once you do 10 Work Units, and so long as you maintain a 80% completion rate, you get a Quick Return Bonus. Depending on how quick your hardware is this might be 10 times your base score.

The client software has no mechanism to know if you have done 10 WUs with a Passcode, or if you are maintaining a 80% completion ratio, so it starts acting like you get the QRB as soon as you have a Passcode. (which is wrong) So for some days (weeks?) the estimated score is wildly wrong.

Once your WU is done the server gives you credit for it. (the server does know if you have 10 WUs with a Passcode) So far as I know, this credit is not changed later. However, the Stats server may be malfunctioning in several ways so your points may be delayed for days. (weeks?)

Another quirk of this system is that nether the client nor the server reports to you if your successful completion falls below 80%. You simply are no longer credited with a QRB.

The QRB obviously favors 24/7 folding, if you fold intermittently, the estimates are again wildly wrong. It is hard to imagine how the software could predict when you will decide not to fold in the future.

This endeavor is being run by academic bio-physicists, so it is not well documented. They are not English majors nor Communication majors. If they had money to hire folks, they would hire more bio-physicists, not Documentalists. What documents exist are largely written by volunteers not actually privy to the inner workings of F@H. (just like me) Worse, as time goes on, much is outdated, although it was true once.

(As an example, I have no idea how large the QRB could be for a new Volta card, as no one I know can afford the $3000 for that card, in my experience QRB can be 90% of the total points, but I know that percentage is outdated)

JimboPalmer wrote:This endeavor is being run by academic bio-physicists, so it is not well documented. They are not English majors nor Communication majors. If they had money to hire folks, they would hire more bio-physicists, not Documentalists. What documents exist are largely written by volunteers not actually privy to the inner workings of F@H. (just like me) Worse, as time goes on, much is outdated, although it was true once.

Academic research implies that results will be peer-reviewed and (hopefully) published it a reputable journal. Whether the research was one of the requirements for an advanced degree on the part of a student or simply an entry on the CV of a faculty member is incidental to the actual published results. Once published, the journal articles contribute to plans for future research by others.

I'm a mathematician, not a bio-physicist so the Wikipedia article is more understandable to me than most of the papers, themselves. I particularly like the Markov state model figure on the left where it shows a 2 dimensional view of a process which actually takes place in a very high-dimensional space. (I'm sure others will find something else interesting.)