October 15, 2000

6:12 PM

Piggybacking off of SETI's success...Anyone downloaded this FightAIDSAtHome program? I figure it's a bit more practical/useful than SETI@Home, and I'd rather cure a disease than search for intelligent life. Let's get cracking! [via Salon.]
posted by gramcracker (22 comments total)

I'm not sure I can explain why, but this makes me feel extremely uncomfortable.

SETI@HOME is an undeniable success, at least in establishing broad-based volunteer computing to accomplish a collective goal. It wasn't the first, but it's the largest by a LONG shot, and sometimes a difference in degree becomes sufficient to become a difference in kind. (For instance, it makes the RC5 effort look tiny by comparison.) They've had to solve problems never solved before to sustain the capacity, and they've moved that particular state of the art to an entirely new level, and even if they don't find anything in their target science, the project has to be considered a significant success in computer science.

It was inevitable that someone else should try it. It bothers me that it's a for-profit company, irrespective of the nobility of their goal. There is a large, but ultimately limited number of people willing to participate in these kinds of efforts. RC5 recruited people by promising to split the reward money with whoever was lucky enough to have his particular computer find the key.

SETI@HOME recruited people with the sheer grandiosity of their goal and the obvious coolness of it all.

These new guys are appealing to our morals, and they may not be wrong. But when do #3 and #4 and #5 and #6 appear? What will they be analyzing? Will it be as noble?

Will there come a time when these kinds of projects have to start offering prizes or bribes to recruit people? Will there be saturation? Will they have to get into what amounts to bidding wars with each other?

Ultimately they're competing for the same population base, because most people won't be interested. Even SETI@HOME, by far the most successful so far, has recruited only a percent or two of existing computers capable of participating -- which is still a huge amount of compute power.

I confess I'm tempted to switch despite my own argument because at the time I wrote that there wasn't any alternative. But I worry that this could rapidly become abused, recruiting people for far less savory distributed computing projects than this one.

And when does the first one appear with a trojan in it?

Perhaps what makes me sad is that I see evil in the future. There was probably no way to avoid that, but it still bothers me. This started out so well, but it will eventually turn sour.

posted by Steven Den Beste at 8:13 PM on October 15, 2000

I'd rather find a cure for an extraterrestrial disease.

There are lots of distributed computing organizations, I put a lot of time into GIMPS, the great internet prime number search looking for the largest prime number. They actually have results, and huge prime numbers (millions of digits long) are useful in future crypto applications, not just mathematical trivia. RC5 just proves the point that enough time and processing will break any encryption, not terribly surprising.

I stopped participating with SETI and GIMPS mainly because I wanted a more stable system and wanted to enjoy the power-saving benefits of shutting my computer down at night.

From what I've seen SETI and such are mostly geek equivalants of drag racing. Look how fast my PC is dudez. Or my team of nerds can beat your team of nerds anyday. Which is great for SETI, but I doubt this crowd will quit the sci-fi-esque SETI to join this project. SETI just has such a huge appeal. I've been to companies where helpdesk or admins set up their entire lan to run it so their team gets a little push and some of these machines were 486s taking a month to do one workload. Scary.

The trojan question is kinda scary, as SETI refuses to release the source code for security reasons. Call me paranoid but imagine someone from the CIA offering SETI a new nice big fat radio telescope for just 2% of their computing power and no one's the wiser. You could be attempting to crack your own 128-bit transmissions.

posted by skallas at 8:36 PM on October 15, 2000

The below sheds some (dark) light:

From the http://www.fightaidsathome.org/ home page:

"Some of the time, Entropia's software will be running commercial tasks on your computer."

From the http://www.entropia.com/ home page:

"From time to time, Entropia runs commercial applications on your computer. At all other times, Entropia runs important applications for leading research organizations around the world to help solve some of the greatest problems facing humanity!"
posted by mmarcos at 8:47 PM on October 15, 2000

"Commercial tasks"? Like what? Cataloging my software for marketing organizations?

Unless I know exactly what commercial tasks they're running, that thing's getting nowhere near my computer.

posted by Steven Den Beste at 9:02 PM on October 15, 2000

So, I'm supposed to give this company access to my computer for free, so that if they ever find a treatment, using my free computer power, they can charge AIDS patients $200 a daily dose for it? What a deal! Corporate altruism rocks!
posted by Optamystic at 11:41 PM on October 15, 2000

The amount of potential processing power available if everyone worked collectively is mind boggling.
posted by grank at 12:25 AM on October 16, 2000

From the standpoint of benefit to mankind, if you're going to devote your spare CPU cycles to something, I'd rather it be a search for a cure for AIDS than a search for extraterrestrial intelligence. SETI is a romantic notion for anyone who grew up watching Star Trek, but it's basically a wild goose chase. There's precious little intelligence here on Earth; what makes anyone think there's any of it Out There at all? I'd be happy to make a bet that SETI@home will find nothing in my lifetime, except that at that point I'd be dead and unable to collect. On the other hand I think it's quite likely that AIDS is curable within my lifetime.

The idea of turning my computer over to unspecified commercial jobs, however, does give me pause.
posted by kindall at 12:40 AM on October 16, 2000

Popular Power is a commercial "spare cycles" system...
posted by owillis at 1:10 AM on October 16, 2000

Have we ever "cured" a virus? I'm sticking with SETI, I'll consider changing if they need help curing cancer. I think it is pretty small potatoes to be judging people on what they choose to let their processor crunch while they are eating dinner.
posted by thirteen at 1:27 AM on October 16, 2000

How about, say, switching off your computer overnight, and donating the money you save in electricity to an AIDS charity?
posted by holgate at 4:47 AM on October 16, 2000

Before people get the wrong idea about this program, the research effort itself is non-profit -- it's run by a group at the Scripps Research Institute, an academic, degree-granting research institute in La Jolla, CA. It's not a company.

Dunno what the software means by "commercial tasks", though.

Owillis -- most of our success in fighting viral diseases has come from vaccines. Smallpox is probably the most notable success. Polio, influenza, measles, mumps, and rubella also come to mind.
posted by shylock at 10:27 AM on October 16, 2000

Not to mention there's a difference between a cure and an effective treatment of the symptoms. Treatment sure beats nothing.
posted by skallas at 10:34 AM on October 16, 2000

Not gonna load it. An easy decision for me since they only have a Windows version available.
posted by Sal Amander at 12:32 PM on October 16, 2000

Unless I know exactly what commercial tasks they're running, that thing's getting nowhere near my computer.

I bet they don't even have any buyers right now. They don't even have a potential teraflop, just 850 potential gigaflops. I don't work in supercomputing but I think you'd need more than a few teraflops before big commercial, academic, or government applications can feasibly be run.

posted by skallas at 1:09 PM on October 16, 2000

Shylock, much of the effort associated with trying to find approaches for AIDS seem to be concentrating on creating a vaccine. The problem is that it's not theoretically clear that a vaccine would help.

What has to be understood is that the life cycle of HIV regarding AIDS is extremely complex (and not totally understood). The initial infection of HIV is fought off successfully by the body, and by the end of it, the amount of HIV in the blood is negligible. But during the interval in which HIV is present in the blood, it still does its retrovirus thing and writes its genes into T4 cells. Then years later they activate and the processes leading to AIDS begin.

Suppose you have been vaccinated. What that means is that your immune system is primed to fight the disease already. When it appears in your blood, it's attacked immediately and destroyed much more rapidly. But that process is a statistical one and it doesn't prevent you from being infected, and it may be that some of the HIV will still manage to infect and genetically alter T4 cells anyway.

They're working on vaccines more because that's the only thing they know to do against viruses generically, than because there's any reason to believe that one will work in this case.

The point you need to keep clearly in mind is this: The initial infection is successfully fought off by the immune system anyway. All a vaccine would do is make that process faster. But it won't prevent infection and it won't prevent HIV from being in the blood and it's not clear that it will prevent HIV from modifying T4 cells anyway. And current theory suggests that if any of your T4 cells have been modified, you'll get AIDS anyway -- eventually.

There's no theory to suggest that lowering the level of HIV in the blood makes it any more safe; there may be NO safe minimum level of HIV in the blood. And all a vaccine does is reduce the maximum level achieved by the virus; it doesn't prevent infection completely.

In terms of the lifecycle of HIV/AIDS and the theory of how vaccines work, I don't believe that they are theoretically capable of preventing it. I think all the efforts being applied to finding a vaccine are a waste of time.

posted by Steven Den Beste at 1:51 PM on October 16, 2000

Skallas says "I think you'd need more than a few teraflops before big commercial, academic, or government applications can feasibly be run."

That depends on what you're trying to accomplish. If what you're trying to do is to inventory the files on each computer in the system, you're not even using floating point. Why shouldn't the BSA contract with these guys to run an app which hunts for pirated applications?
posted by Steven Den Beste at 1:55 PM on October 16, 2000

Probably this privacy statement.

Someone else pointed out that you could just send them a check for the power you saved when your PC was off during the night. I doubt the extra 5-bucks per volunteer would pay for the supercomputing they want. But what if they only ran commercial projects and donated the money to other AIDS foundations that aren't doing research with supercomputing? Are their algorithms and mass-processing trickery more in tune with reality than chemistry based research?

Personally, I don't think any conscientious person will participate without a statement promising that their PC won't be used for certain government or commercial purposes (virtual nuke tests, breaking of encryption, etc) If they're going to pull the, "Oh btw, you're going to be paying the bills for the next 1000 workloads" card they should hire an independant auditor to verify what work is being done and "bill payers" should recieve email that this workload is not AIDS related.



posted by skallas at 3:06 PM on October 16, 2000

Steven, I'd have to disagree with you, for a bunch of reasons. But the best evidence for the plausibility of an HIV vaccine is the existence of those people in Africa who seem to have been innoculated against HIV because they were previously infected with a less virulent strain (the so-called HIV-2). There have also already been successful vaccines against the monkey version (SIV) already, and a couple of human vaccines are in clinical trials now.

Have hope. Sometimes even scientists do good work.
posted by shylock at 4:21 PM on October 16, 2000

Is there any proof that the brute force solution they're talking about even has a chance of resulting in a meaningful solution?

I don't know much about this sort of thing, but the technique they're describing sounds a little dubious. Are the processes involved understood well enough that simulations can be accurate enough to yield useful results?

-Mars
posted by Mars Saxman at 4:36 PM on October 16, 2000

Is there any proof that the brute force solution they're talking about even has a chance of resulting in a meaningful solution?

Who knows? What I do know, from having friends who are research chemists and biologists, is that the places doing this research aren't short of computing power. In fact, Sun, SGI and IBM are often willing to subsidise or donate hardware simply because high-end laboratory calculations provide a rich source of real-world benchmarks for their kit.

If money's to go to research, I think it's better spent on employing talented researchers, so that they don't end up abandoning science for management consultancy.
posted by holgate at 7:10 PM on October 16, 2000

Mars-- It's a decent idea. They're basically modelling the interaction between potential drug candidates and HIV protease. It's one of the techniques medicinal chemists develop leads for new classes of drugs; the advantage of doing it in this massively parallel manner, from my quick reading of the web page, is that they'd be able to screen lots and lots and lots of candidates at once, and systematically cover a broad range of possible structures.

The caveat is that what they're doing is, in the end, only a simulation. They still need to actually synthesize the molecule and do the appropriate biochemistry and pharmacology to make sure it does what they think it does, and that it doesn't do anything else.

So, is there proof that this will work? No, not as such. Is it still a worthwhile idea? I think so. It sounds like a useful new complement to other methods of drug discovery.
posted by shylock at 12:33 AM on October 17, 2000

If it were being done directly by the lab, the way SETI@HOME is being done directly by the researchers, and if that were the only thing that the program was capable of doing, I'd sign up in a flash.

What I object to is the fact that Entropia isn't just about searching for AIDS treatments. When you run their program, they can download anything to your computer and run it. It's the ultimate trojan.

In fact, that may turn out to be what it becomes. What happens if some k00l d00dz figure out Entropia's protocols and checksum algorithms and safeguards, and learn how to take control of any computer running Entropia's program? You don't think someone's going to try? If they convince thousands of people to run the thing, it's a tempting target.

Sorry, no thanks. I don't care to come in some morning and find all my disks wiped and the BIOS flash on my mobo programmed to all 0's.

posted by Steven Den Beste at 1:28 AM on October 17, 2000