Folding Forum

It's already clear that Folding@home is an extraordinarily powerful, productive, and pioneering project that's made large strides in the fields of both software and molecular biology, and I'm not questioning that at all. However, I'm wondering what is holding this project back from going even further than it already is. Perhaps its already well-known consensus on what these issues are, but even if that's the case I thought it might be a good idea to put all of them together in one place, so that it might be easier for the PG to notice our concerns. I don't want to start a gripe session or anything like that, but perhaps we can all logically list things that we think should be addressed for the good of the project, and recognize what's already been done.

So, the main discussion topic is this: Based on your own experiences and interaction with others, in your opinion what issues are holding F@h back, and what past improvements were you particularly pleased with?

For example, here are some of my thoughts, in order of importance:
1. GPU cores which cause disruptive lag, since they are now bundled with V7. IMO, this has a good chance of turning newcomers away.
2. Heat issues related to maximum hardware utilization. V7 utilizes everything by default. Everyone has a laptop here on campus, so this is a concern.
3. Bugs in V7. There are a few important tickets for Windows, but there are comparitively more bugs in Linux and especially OS-X.
4. GPU folding in Linux. People who run Linux know what they're doing, and I think there's enough computing potential there to justify the pursuit.
5. GPU folding is sometimes technical and tricky due to incompatibilities, etc. Still an issue, but the whitelist and V7 itself are reducing the problem.
6. Awareness. Will always remain on this list, though both the PG and us donors can help reduce this, can't we?

Solved:
1. In general, Folding@home is too technical - essentially solved by the website changes, the V7 client, the Wikipedia article, and help on this forum.
2. Difficult to find out what F@h is actually doing - there've been frequent blog posts and many beta projects opened, and I'll strive to keep the Wikipedia article updated.
3. Website is confusing/technical - the FAQs could still be improved, but this is essentially taken care of.

AFAIK, has never been a problem:
1. Self-explanatory name - "Folding@home". Pretty much speaks for itself. (Volpex@UH may have this problem, and Yoyo@home unfortunately does not involve yo-yos. )
2. Project productivity - Impressive already, let's get even more done!
3. Competitive incentive for folding - Essentially figured out, despite a few rough spots here and there
4. Reasonable levels of transparency and cooperation between scientists and donors - pretty good job here I think

What are your thoughts? If you agree with my opinions above, please show your support by posting so below.

Good luck Jesse_V,

In my mind the biggest thing holding F@H back, or the biggest threat to F@H (depending on how you want to see it) is a mix of improvements in computer powersaving features, move towards mobile devices and a plateau in 'good enough' computing.

General apathy has been the main reason that my friends and family either haven't folded or have stopped. With my friends, it's seems like it's due to a combination of boredom and a lack of a tangible benefit, though the power costs would probably be the main reason now (it was so much nicer when we were on-campus undergrads and the power costs were covered under room+board). Most of them stopped because they seemed to tire of running it and weren't hooked by the potential benefits. One of my friends who never had any interest perked up a bit when I told him about the EVGA Bucks program, but he never went any further than that. None of them were at all concerned about points.

I'm not sure that's a problem that can be effectively dealt with and still have it be purely a not-for-profit academic endeavor. The points system does attract some people, but there are many for whom it doesn't really matter outside of purchasing decisions or identifying hardware problems. You could likely get more people to join and stay on if there were more tangible rewards like EVGA Bucks, but I'm not sure if the Pande Group could offer something more than points without seriously muddying the charitable nature of the project, and I'm not sure if they have the budget to offer more or set up an infrastructure to do so. Other companies could certainly follow EVGA's lead, but I imagine there aren't that many who'd be interested and fewer still who'd be willing to go along.

This is somewhat related to the other major issue I've seen- power costs. Yes, most people think of it as a charitable donation, but there are a lot of people who have quit because they either can't afford the power cost or just don't feel that they get enough of a return on what they spend. If people could be reimbursed for some or all of those costs that could be an incentive to not quit, but I don't find that a likely scenario.

Ultimately, though I think it's primarily the lack of tangible feedback. I think a number of potential or active contributors just aren't excited by the publication of papers or some blog posts, and see the points as meaningless since they have no real value. The promise of helping to find a cure for X just doesn't hold much sway for them when it's going to take years or decades to be found. I don't know what the solution to this is, if there even is a reasonable one.

Zagen30 wrote:General apathy has been the main reason that my friends and family either haven't folded or have stopped.
...

I've read that many other popular distributed computing projects have real-time visualization. SETI@home for example does this.
Do you think there should be a push for default visualizations so long as performance isn't significantly compromised or lag generated? Would that help?
I suppose the V7 FAHViewer needs to be fixed to display non-default proteins for all Fahcores.

Lack of choice. Compare to WCG where you can choose what projects you want to run, right there in a nice list: childhood cancer, clean energy, clean water, etc. Other BOINC projects have similar abililties. F@H has many different areas of research but you can't easily pick and choose those you would prefer to work on.

Jesse_V wrote:I've read that many other popular distributed computing projects have real-time visualization. SETI@home for example does this.
Do you think there should be a push for default visualizations so long as performance isn't significantly compromised or lag generated? Would that help?
I suppose the V7 FAHViewer needs to be fixed to display non-default proteins for all Fahcores.

No, I don't really think that would have helped my friends at all, at least not the ones who did fold but then stopped. Seeing some pretty graphics wouldn't have made up for the fact that they didn't feel like there was a compelling hook to keep them going. I think visualizations can help recruit people, as they're more interesting than a console output or the regular FAHControl window, but I don't think they'll help in retention. And, considering how much I talked about the project with my friends and family, I'm not sure how much visuals would have convinced those who didn't try it out at all. At this point, most people I know and who I've tried to recruit either did and subsequently quit or are unlikely to ever start.

To my knowlage there is 3 issues that is holding folding back

1: High electricity cost ( in denmark we pay around 0,26-0,32 pr KWatt and half of that is tax). That is least holding myself a bid back (I am only folding on one mashine where I can get free electricity but not on another where I pay for it)
2: Some computes are generating extra (and unwanted) noise when running folding@home.
3: A lot of people I have been talking to has a "why bother" attitude, Because a note in folding@home-network is not significant so why should they listen to the extra noise and pay the extra bills

Doing it right:
1. The results are frankly the only reason I fold. Stanford produces in quantity and quality.

2. Blogging results and slightly less secretive about project details.


Doing it Wrong:
1. GUI's It has been a long time since I encountered something as poorly laid out as the V7 client...
If a manual is required and you have to read it twice...

2. GPU folding in linux.
I stopped gpu folding just because it was a pain to run through wine... native please.

3. People not slaves.
Don't expect to flippantly assign values without proper reasoning behind it and have everyone go along with it.
When a university has donors visit they treat them quite well...

4. High E cost and Data caps on internet.
Power cost money if there is not a strong reason why people should pay for folding they won't.

Electricity is definitely a huge concern for me. Though I fold at home 24/7 and my parents don't really know what it is , I know that the electric bills are probably high :/
Oh and the heat in summer, though I have no problem putting up with that. I just don't like the fact that I'm making my parents pay high elect bills lol.

What I do like is how I'm helping scientific research, though I don't know how much significance I make.
And I like the fact that the results and data are shared for everyone (I think), although I have no idea what they are lol.

Direct information of the "About Project" of V7. Should be given much more importance. To bad example, see "About Project" of P8035. A good example the P8042.
Hire communication specialist at PG?
Different languages ​​for "About Project: V7"?
Leverage "About Project" to give news of achievements?

Communication PG> donors, is essential. I think today is very poor. Jesse_V often stressed the need to improve this. It has achieved something, but there a lot, and it must be maintained over time.

It is not that Dr Vijay Pande, is intervening all the time. He has many other occupations. This is an appropriate interlocutor between PG and donors. I know it's hard to find the right person, to understand science and have communication skills.

A computer center of a company, the jobs are sent for processing instructions. So feels the donor. And the donor is paying the electric bill, putting team updates, weather and more.

I have had the experience that one partner is working for several months, I wonder: Jorge, that's exactly what I'm doing?

The lack of a true -bigadv class computer on which to do -bigadv work unit benchmarking, which causes projects like p6903 and p6904 to get too many points* while other projects get normal points, causing donor complaints and violations of FAH Best Practices guidelines when blocking the WUs with the "lower" points.

*http://foldingforum.org/viewtopic.php?p=207086#p207086

kasson wrote:[p8101] Data are good. Thanks. We can do additional diagnostics as the WU returns come in; we're also running these on a few of our own systems.
Preliminarily, the scaling data on these [p8101] look fairly good. I think the issue may be that 6903/6904 thrashed the cache on our benchmark machine and took *much* longer (hence the big jump in points assigned). But that's speculation. As we get timing data from a number of systems, we will re-evaluate the [p8101] points assignment.

Speculation easily confirmed looking at the PPD on various points tracking charts. It would also be very easy for Pande Group to end the speculation, thus ending donor complaints.

-bigadv folders work extra hard to build, configure, and make their systems run these larger work units. System cost has never been a factor in figuring points, but the contribution of additional computing resources has (hence the extra big QRB for these WUs). And these guys/gals contribute a lot. It's only fair for them to know, with absolute certainty, that they are getting what they deserve.

Long-term

1.shift to reliance on cloud rather than desktop
2. exponential decline in desktop usage/replacement with notebooks

medium-term

1. decline in gaming desktop usage.
2. retirement/replacement of PS3 / new generation of consul units accelerating current trend
3. declining usage of gaming cards/crippling of GPU compute in consumer cards

short term

1. Nature of institutionalized research/ outcomes take years for peer review much less movement to application
2. technology has consistently out-paced the resources to program and utilize efficiently. (GPU)
3. lack of a "head-line" cause.

While the research generated to date is significant, worthy and laudatory.. F@H will not go "mainstream" unless there is a clear and obtainable goal (in some definable framework...years not generations)

For example, if there was a specific problem with a defined requirement...it would be possible to generate mass support.

Unfortunately, that is not the nature of basic research.

Before long there is not going to be a FAH as we currently know it, Consumers computers are moving to I pads and such mobile devices. My phone will do what most consumers want a computer for just tether it to a keyboard and a monitor and hey you have a internet cursing home computer. Hey I will do a WU a month on my phone, if the battery does not go dead. Before too long the only folders are going people who build specialty hardware for folding. Perhaps someone should be thinking about that.

There have been some recent announcements from the computer manufactures that there focus is no longer going to be developing performance type cup’s they are switching to the mobile market. When that happens the only advancements you are going to see are going to be in the server class or nitch markets.
.

Wow those are some interesting insights! I wasn't thinking that "threats to this project" would be equivalent to "what is holding F@h back" but I'm thinking that these issues may already be at hand. Certainly the state of technology has changed since 2000 when the "Screensavers of the world, unite!" paper said "The time is right for distributed computational biology." Perhaps we need to brainstorm some ideas for the PG to consider so that they can try to counter these current issues.

Punchy wrote:Lack of choice. Compare to WCG where you can choose what projects you want to run, right there in a nice list: childhood cancer, clean energy, clean water, etc. Other BOINC projects have similar abililties. F@H has many different areas of research but you can't easily pick and choose those you would prefer to work on.

Good point. WCG certainly has a nice-sounding names to their projects. Perhaps the PG should focus even more on disease research, and make an effort to note in their project descriptions the connection between general protein folding and the implications to diseases.

Zagen30, beer, and Meh_Lay_Lay have all raised a really interesting point about donor retention. Actually, I really understand those viewpoints; they do make sense. It's difficult to know what impact you're making when the points are made up and you have no idea what progress you made in folding a protein. It doesn't help that we have no idea what remains to be done. I'm working on 7016(1, 45, 51) but I've no idea if those last three numbers go up to a million. I mean I know they don't, but if they did it would make my WU completion seem rather insignificant, wouldn't it? I've no sense of a "progress bar" for that protein. Could the PG construct such a "progress bar" and have it displayed on V7 or something? I've read that sometimes simulations are restarted and refined to make the statistical model more accurate, and that might make a progress bar more difficult to construct, I don't know.

As k1wi, mdk777, and Grandpa_01 point out, there is a movement to more mobile devices, and I don't see that really stopping anytime soon. While that may take away from the project, I think there will always be a solid set of people who hold onto larger units simply because they can cool themselves better. Modern games use realistic graphics, which require powerful hardware, which generate heat, and so are not compatible with things like an iPad. But if the systems are designed for purposes like gaming, then that would conflict for processing time with Folding@home wouldn't it? Maybe F@h over time will simply move to the more enthusiastic folders, and it'll be less casual. The planned LTMD CPU-GPU hybrid core comes to mind, that's certainly going to be demanding and not something just anyone can handle.

mdk777 wrote:...
short term

1. Nature of institutionalized research/ outcomes take years for peer review much less movement to application
2. technology has consistently out-paced the resources to program and utilize efficiently. (GPU)
3. lack of a "head-line" cause.

While the research generated to date is significant, worthy and laudatory.. F@H will not go "mainstream" unless there is a clear and obtainable goal (in some definable framework...years not generations)

For example, if there was a specific problem with a defined requirement...it would be possible to generate mass support.

Unfortunately, that is not the nature of basic research.

What can be done to address this? Is it possible for us to know the planned paper and its publication deadline? I've read that the server is aware of the deadline and tries to prioritize WUs accordingly. Are we allowed to have access to that information? If so, surely that would help. It could just be placed in the project description or something like that.

I mean I'm still interested in seeing people's thoughts on issues that are threatening/slowing this project, but now that we have a good reasonable list, what should the PG do about them? I feel this thread is a good opportunity to offer suggestions. I'd like this to be a focus of what the community thinks the PG should address. Obviously not every folder is active on this forum to contribute here, so that's why I'm asking for your ideas based on your own experiences as well as what you've heard from others.

There were some statistics posed not tong ago showing that while mobile devices were growing very rapidly, they were not cannibalising the traditional PC market, (The rate of growth was continuing but no longer getting steeper.) The ipad market will contine to get steeper but it's a bit like mobile and landline phones. Everybody now has both with the population growth adding new landlines at about the same rate as other folks transition to mobile-only, dropping their landline.

I'm not convinced that FAH will move to cloud computing. The folks selling cloud resources will end their free testing period and start charging full rates which will cover their hardware and electricity plus profit. Will you pay more to have somebody maintain your computing resources than you presently spend on hardware for FAH or will making all of those costs visible tend to decrease your participation?

As long as there is a market for multi-purpose home or office computers that do many non-FAH things, there will be FAH running on them. (with the permission of the owner). As long as we have (mostly younger) folks that don't pay their own electricity bills FAH will not die.

The competition between game consoles and home gaming PCs will continue. When Sony came out with the PS3 and it was a viable source of processing for FAH, a client was developed. That might happened again if there's a market for a much more powerful gaming console, but I doubt it. The amount of processing power in each game console has NOT grown very rapidly -- although the number of units keeps growing. The processing power in home computers that are used for gaming does continue to grow. Today, the PS3 is just about as useful to FAH as a relative cheap PC with a low to mid-range GPU (though the were mid-to high-range GPUs at the time). People still will buy a PS3 for games and will enjoy them there, but those who have gaming PCs insist on more powerful hardware, which also is good for FAH. I don't know enough about that market to say why.