SETI and the internet's computing power

[Mephistopheles]

I don't know if you are at all interested in this, but I was considering the implications of the <A HREF="http://setiathome.ssl.berkeley.edu/" target="_new">Seti@home</A> project.

I mean, the project has received the equivalent data processing power of 2 million years of CPU time and 5x10^21 floating point ops.

This is of course a great idea, but I think it is unfortunate that the end result isn't something that would surely benefit us all. And while it is an open project, it's kind of silly at some point.

I mean, when asked, the users all said that they wanted to find aliens for the good of humanity. But what the hell? Shouldn't the good of humanity be less far-fetched? Like pharmacological research or fusion technology research? Or couldn't a top university issue a work unit of their latest projects, for you to run on your computer?...

...of course, it is possible that these more earthly tasks would have political implications. What if several individuals from several nations helped in running a massively parallel pharmacological study and found a revolutionary drug? And the nobel goes to... all you bold internet users?... What about money?

I don't know if this discussion interests anyone out there, it's just that it's interesting to see that the internet enables such computational power... Just thinking.... Isn't there a way to convert this power into something that will <i>surely bring us benefits</i>, not something as vague and uncertain as aliens? Finding aliens wouldn't make earth spontaneously better.

Like, for instance, developing another computer processor? More efficient? Or whatever? Using this power that Seti@home is harnessing for the CERTAIN benefit of mankind should be possible, shouldn't it???

<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i>

 

[sonoran]

You mean something like this: <A HREF="http://www.grid.org/projects/cancer/" target="_new">http://www.grid.org/projects/cancer/</A>?

 

[kchen]

Yes and no... I think the use of distributed computing is difficult to apply to things like biology. What kind of search parameters can you really specify, and how do you measure the quality of the result (is Protein A good or bad? How can you tell without an actual experiment?) If these things could be codified, why would we need to search at all? We could just use logical deduction and inference to formulate the perfect protein and create it.

The reason why SETI is a near-perfect application of DC is that it isn't really about experimenting, it's about trudging through a ridiculously large data set looking for objectively identifiable mathematical patterns. You aren't looking to define something qualitatively (are the aliens good or evil?) It's something that any single computer could do, but just take a very long time to do it. That's the kind of application that DC is for. With biology everything really needs to be grounded in lab experiments, not computations. The computing power far exceeds the number of useful computations that can be applied, so what ends up happening is that these cancer researchers and such end up feeding you a lot of junk data trials that essentially do nothing to advance research.

Don't get me wrong; cancer research is very important but I don't think that it's a field that uses DC to its potential.

 

[Mephistopheles]

That is a damned fine example of what I think should be more useful than SETI... I'm reading on it, thanks. Very nice!!!

But enter the considerations: this won't change our society the way it should. Maybe we'll be giving another already too powerful drug manufacturer another recipe... Which we'll later on desperately need...

Granted, it does look much more useful to distribute the 5*10^21 FP operations from SETI into this project too.

<i>Edit: unfortunately, the "grid" project for anti-cancer drugs looks abandoned. The latest news is from 2002...</i>

<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i><P ID="edit"><FONT SIZE=-1><EM>Edited by Mephistopheles on 08/10/04 10:41 PM.</EM></FONT></P>

 

[Mephistopheles]

Ahh, yes, you're right! DC is very hard to do effectively. Programming for that level of computing is very, very hard. With SETI, they had a computational task at hand that could be easily split into a pathetically huge amount of small work units!

What kind of search parameters can you really specify, and how do you measure the quality of the result (is Protein A good or bad? How can you tell without an actual experiment?) If these things could be codified, why would we need to search at all? We could just use logical deduction and inference to formulate the perfect protein and create it.

Believe it or not, there are groundbreaking computer algorithms that can, up to an 80% accuracy, predict the usefulness of a specific protein against cancer... I find that impressive. One of the professors I worked with explained the basics of that specific research trend.

Accelrys, for instance, the company responsible for the aforementioned "grid" project, must be employing some sort of algorithm to evaluate each molecule.

Smart algorithms are created everyday. Be it the already old neural networks or the newer genetic algorithms, many alternatives exist to create "smarter" code.

A genetic algorithm, for instance, takes a pool of elements, evaluates their "fitness", and breeds (i.e. mixes together) the best, fittest elements, into another pool of elements. Once the fitness paramater has stabilized, the GA has optimized the population. A mutation can be introduced from generation to generation. The fitness evaluation is very costly, for instance. It's a beautiful optimization method, just as an example of what can be done with a little creativity in mind!

In any case, in that particular accelrys grid project example, I'm actually quite inclined to agree with you about the cancer drug research there: It doesn't sound as if it's easy to get meaningful results by using those methods..... Since the methods themselves (i.e. their "fitness to fight cancer" function) might be debatable. It's too new to warrant so much confidence... I guess...

<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i>

 

[P4Man]

There is only so much you can do with a distributed computing model where ping-pong latencies between nodes are measured in ten or even hundreds of ms. Its like having a factory with a near infinite number of workers (or machines), but an assembly line that is no faster than 1m/hour. Probably usefull for building space shuttles, but no good to manufacture tennis shoes or cardboard boxes.. BTW, this is more or less the same problem as with dual core (or SMP) systems


As for SETI as such; its a matter of personal appreciation, but I for one, am very interested in such research, and as someone else pointed out, this workload (unlike most others) is ideally suited for such a solution. Its also pretty much "free", so why not use all that unused computing power to cruch through that data ? Odds might be extremely small, just imagine in that radio signal database there is indeed an "ET phonecall", it would silly never to have discovered it. ET might have a better cure for cancer after all

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[CaptainNemo]

I gave up on SETI because I didn't want my CPU working at 100% all the time; I also couldn't see the point of using ancient and/or slow computers and days worth of power to do something that a supercomputer could do in minutes.

Axis of Stupid = coop, Kanavit, FUGGER, SoDNighthawk, and ninkey.

 

[P4Man]

> I also couldn't see the point of using ancient and/or slow
>computers and days worth of power to do something that a
>supercomputer could do in minutes.

Could it ? Have a look here:
<A HREF="http://setiathome.ssl.berkeley.edu/totals.html" target="_new">http://setiathome.ssl.berkeley.edu/totals.html</A>
57 Teraflop per second over the last 24 hrs. Earth Simulator, worlds fastest supercomputer does <A HREF="http://www.top500.org/list/2004/06/" target="_new">~36 Teraflop</A> sustained (40 Tflop peak). So, all our computers combined are on average faster than running Earth Simulator nonstop.

I know flops is a bad measure of performance, etc, etc, but still, no there is no way any supercomputer could hack this in a few minutes. A more typical supercomputer does 5 Tflop, so to do the crunching seti@home has done over the last 5 (?) years, you'd have to run it on the supercomputer for ~50 years.

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[slvr_phoenix]

Mephistopheles, since the distributed computing aspects have already been covered well, I will instead explain the social and economic problems with your idea.

Pharmaceutical laboratories are not looking for better drugs to benefit mankind. If they were then they would set themselves up as not-for-profit organizations or offer their drugs for reasonable prices. No, they are looking to make money. Such is why drugs are priced so insanely high these days. I would sooner give the RIAA complete control of my PC than I would give free processing power from my computer to a pharmaceutical company. If any does ever find a super-drug they will charge an arm and a leg for it instead of help mankind with it.

As for fusion technology research, it is much the same. Should someone finally produce working cold (or warm) fusion technology all that it would serve to do is make the company (or person) that discovered it rich while keeping access to the technology limited and at a high price. Besides, again, the key to fusion technolgy is not as simple as running large amounts of data through an algorithm.

In fact there are very few organizations/businesses that I would trust to benefit mankind with their discoveries even if they had their key handed to them by freely offered distributed computing power. SETI, having no salable product, is one of the few cases where any results obtained through a distributed computing network might actually benefit everyone equally.

And, again, even if any company had a distributed computing network solve their problems, they would still need to apply this information to actual lab work before it could be proven to actually work. Which, again, is another area where SETI has a significant advantage.

If you just don't believe in ETs however (as not everyone does), then I believe that there are also distributed computing projects to find prime numbers and such. I'm too lazy to link, but I'm sure that a quick Google search will do. Is furthering mathematics a lofty enough goal?

<pre><b><font color=red>"Build a man a fire and he's warm for the rest of the evening.
Set a man on fire and he's warm for the rest of his life." - Steve Taylor</font color=red></b></pre><p>

 

[Obtuse]

The entire distibuted-computing-for-profit fiasco actually happened with the original Kazaa. One of the little ninja programs it installed used "idle" cpu cycles and spare hd space. Then the company had available storage and computing power it would sell to other companies. This is how kazaa would pay for itself. But, of course that got shut down. Just interesting though, because 5 years ago I'd never heard of adaware, now I run it, spyware blaster, and spybot search and destroy every week, and it seems like they always find some little crappy prog that managed to work its way into my hd.

"It's too late now anyway. That song is stuck in my head and the only way to get rid of it is to blow it out. With a bullet!! - Carl

 

[sonoran]

Pharmaceutical laboratories are not looking for better drugs to benefit mankind. If they were then they would set themselves up as not-for-profit organizations or offer their drugs for reasonable prices. No, they are looking to make money. Such is why drugs are priced so insanely high these days. I would sooner give the RIAA complete control of my PC than I would give free processing power from my computer to a pharmaceutical company. If any does ever find a super-drug they will charge an arm and a leg for it instead of help mankind with it.

So when you or someone you love comes down with a fatal, incurable disease, you'd rather them die than pay a drug company for a cure? Forgive me if those of us who have lost loved ones have differing opinions. We'd have gladly paid for those cures, regardless of how they were discovered or developed.

Like you, I'm not naive enough to believe the drug companies' efforts are altruistic, but that does not make them unworthy endeavors, or undeserving of support. The benefit for the drug companies will be short term, the benefit to mankind will be for all posterity.

 

[Mephistopheles]

Hm, yes, but what about those who die everyday because of diseases that can easily be treated by a drug they can't afford? These things happen everyday.

For each patient suffering from a terrible, incurable cancer, there are hundreds of others in underpriviledged places in, say, africa, that are suffering from trivial illnesses that will eventually kill them. And what is more concerning is that that patient with incurable cancer might need a treatment that will prolong his life by a few months or years, without healing the disease altogether, but the treatment might be <i>so expensive</i> that you could, if these resources were reallocated, save tens if not hundreds of sick people somewhere else.

The problem here, of course, is how poorly the drugs will diffuse throughout mankind. This is a social implication and is in line of the "political" problems I was expecting....

In any case, there would be an added advantage in knowing another drug, that much can't be debated, but the problem is that the advantage would only affect the few very rich individuals.

 

[imgod2u]

There are various distributed computing projects out there that go towards more realistic goals than say, Seti@home. However, they are also not as popular for various reasons. Distributed computing has the drawback that it's not always reliable in terms of how much performance you get. The performance you get depends on how popular your client/goal becomes. For many projects, the effort required to come up with a distributed computing model/algorithm and put it out there on the internet is often not worth whatever performance you may get. Keep in mind not every project becomes another Seti@home. Seti@home just happens to, for some reason, have become very popular and is able to achieve the throughput performance it does today.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.

 

[P4Man]

Wans't there a program (was it folding@home ?) that could actually execute different payloads, and where you could choose what goal you where interested in achieving ? And if it doesn't exist yet, I'm sure it will soon enough. some company will launch a 'universal DC client', let you choose what sort of applications/research you want to participate in. That way the entrance barrier would become pretty low for anyone with a worthwhile DC project.

One could even consider using a commercial approach, where the DC client gets some sort of (minor) reward for his collaboration (things like extra webmail storage if its google or MSN, maybe even Airmiles or something per completed workloads or x hours of CPU time, or some sort of lottery where you get a ticket for every completed workload.. Reward could even be dependant on the app you choose to run.

This approach would let the client provider then sell their established network of DC clients for a reasonable price, benefit the enduser and make it easy (if not completely free either) for anyone to make use of the network.

Now all I have to do is patent this idea

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[blackphoenix77]

A little bird told me we are going find to a cure for cancer in the year 2006.

<font color=blue>AthlonXP-M 2500+(12x211)</font color=blue><font color=green>Abit NF7-S</font color=green><font color=red>Kingston DDR400 2x256Mb</font color=red>

 

[sjonnie]

Or couldn't a top university issue a work unit of their latest projects, for you to run on your computer?...

Which is why I run <A HREF="http://folding.stanford.edu" target="_new">folding@home</A> from Stanford. Tom's have a bit of catching up to do if they want to get a team in the top 10 though.

<A HREF="http://www.anandtech.com/myanandtech.html?member=114979" target="_new">My PCs</A>

 

[slvr_phoenix]

So when you or someone you love comes down with a fatal, incurable disease, you'd rather them die than pay a drug company for a cure? Forgive me if those of us who have lost loved ones have differing opinions. We'd have gladly paid for those cures, regardless of how they were discovered or developed.

Like you, I'm not naive enough to believe the drug companies' efforts are altruistic, but that does not make them unworthy endeavors, or undeserving of support. The benefit for the drug companies will be short term, the benefit to mankind will be for all posterity.

1) Unlike you I never spoke other people's opinions for them. Nor did I ever say that you couldn't have a differing opinion. I made it quite clear that this was <i>my</i> opinion. Too bad that you couldn't do the same.

2) People die. <i>Life</i> is a terminal disease. Anyone who thinks otherwise is a fool. We procreate for a reason, and if we didn't die we'd have to give that up.

3) Your assumption that I would just calously let someone die is insulting. I am a healer trained in Magnified Healing, Quantum Touch, and Reiki. I've dedicated my life to helping others who need it. <i>If</i> it is someone's time to die I make their passing as easy on them as possible. If it is not, I do what the vast majority of doctors don't. I heal the mind and soul. You'd be amazed at how many physical maladies are simply manifestations of repressed emotions. Good health is not just the body, but a balance of body, mind, and spirit. Healers have known this for millenia. Only in the last few centuries have we forgotten this.

4) Your presumption that any drug company holds a cure for anything is laughable. The vast majority of the 'medicines' that we use today (aspirin, penicillin, etc.) are <i>naturally occuring</i>. All that pharmaceutical companies do is grow the source, extract the chemical component, and put them into undigestible pill forms that mostly get wasted as our digestive systems doesn't know what to do with them. <i>Nature</i> provides us cures. The few compounds that chemists have mixed up on their own that actually make it to market as 'medicines' have miles long lists of side effects. Most never even make it because of their side effects. Nature's cures in natural form work wonders. Nature's cures in pharmaceutical form work moderately well. Man's cures in pharmaceutical usually hurt as much, or more, than they help.

5) Assuming that I have never lost a loved one or watched a loved one suffer is again a gross blunder. My father died when I was 7. By now I've lost all of my great-grandparents that were alive when I was born. One of my grandfathers is barely coherant after a stroke. My closest grandpa, one whom I was named after by the way, has been rapidly degrading in health for the last four years and likely won't make it another year. And he, by the way, is so often dicked around by doctors that he's lucky they haven't killed him yet, constantly thinning and thickening his blood way too much with careless practices and prescriptions. And that's when he can actually afford to pay for his medications, which are costing him more than his food, home, utilities, and vehicles combined. It's a shame that he's too rooted in his belief that modern medicine is the only answer because there is a lot that I could be doing to help him if he'd let me. Have you any idea how painful that is to endure, to know that you could help but are refused while you watch them suffer every ache and pain?

I'm sorry, but the assumptions that you made based on my opinion are incredibly insulting and grossly inaccurate. Just because someone doesn't have the same opinion as you doesn't mean that they have experienced any less in their life. Sometimes it even means quite the opposite.

<pre><b><font color=red>"Build a man a fire and he's warm for the rest of the evening.
Set a man on fire and he's warm for the rest of his life." - Steve Taylor</font color=red></b></pre><p>

 

[slvr_phoenix]

Hm, yes, but what about those who die everyday because of diseases that can easily be treated by a drug they can't afford? These things happen everyday.

For each patient suffering from a terrible, incurable cancer, there are hundreds of others in underpriviledged places in, say, africa, that are suffering from trivial illnesses that will eventually kill them.

Too add to that, death is hardly the only form of suffering that people endure. Further, one hardly has to look as far as Africa. It's almost a guarantee that you can find people in your own city that are suffering similarly.

And what is more concerning is that that patient with incurable cancer might need a treatment that will prolong his life by a few months or years, without healing the disease altogether, but the treatment might be so expensive that you could, if these resources were reallocated, save tens if not hundreds of sick people somewhere else.

HIV/AIDS is another example of this scenario. And while I do resent the widespread suffering of others from such an unbalanced distribution of resources, there are so many other factors such as one's own personal wealth involved that I cannot begrudge their right to do so.

In the end, all I can say is "that's life" and do what I can to help where and when I can. Or to put it a different way, expend my resources on things that actually make a difference, instead of on pharmaceutical companies. If I believed in the one-in-a-billion chance I'd play the lottery and donate my winnings to actual people who need help in paying for their prescriptions or medical insurance before I'd help pharmaceutical companies.

<pre><b><font color=red>"Build a man a fire and he's warm for the rest of the evening.
Set a man on fire and he's warm for the rest of his life." - Steve Taylor</font color=red></b></pre><p>

 

[cdpage]

Wow, this has been quite the Read people.

I thought i might add my own 2 cents.

Being that the all this computational power generaly ends up being a money making deal for those who use it... as its being suggested here. (i.e., Phamaceuticals make drugs for profit, not mankind)

perhaps it should be distributed.

I realy like the idea you Have P4man, but to add to that, leading phamaceutical companies should not be able to patent any drugs procured through such a system in that THEY are not the only ones who help create the drug.

perhaps it could be used as a raceing tool too.

one company can not use it but 2+ must use it...ensureing that once company 'X' win's Company 'Y' will be right behind them with a similar or same product...

eather way Keeps Prices low and get product out faster.

<b><A HREF="http://www.digitalgunfire.com" target="_new">DigitalGunfire-Industrial EBM</A></b>
ASUS P4S8X-P4 2.4B - 2x512M DDR333 - ATI 9500Pro - WD80G HD(8M) - SAMSUNG SV0844D 8G HD - LG 16X DVD - Yamaha F1 CDRW

 

[Mephistopheles]

P4Man's idea is outstanding. That's somewhat like what I had in mind for distributed computing...

I mentioned universities earlier because Universities are traditionally non-profit. Large multinational companies wil probably bring those social and economic conflicts you all mentioned, but universities might divulge their results in a more "let's help mankind"-kind of fashion...

Even so, it's no garantee. Universities can also sell their research for exchange of resources from large companies as well, I guess....

The problem with trying to help mankind is that mankind never quite seems to take any advice... apparently. Social implications, and so on... :frown:

 

[TheRod]

ET signal is a pretext for a much larger project. SETI is the fundation of Low-Cost Ditributed grid, check the Berkeley BOINC project (now SETI is part of it). You will be very impressed by BOINC. I personnaly beta tested it, it still had bugs, but this was a very good step foward.

By the way, SETI will probably find much more than ET signals, by analysing SETI data, scientists might understand or discover atronomical phenomenom. One of their goal is to enlarge sky coverage and widen the collected data and the scope of the search.

--
A7N8X / <font color=green><b>AMD Sempron 2800+</b></font color=green>
Kingston DDR333 2x256Megs
<font color=red>Radeon 8500 128Megs</font color=red> @ C:275/M:290

 

[CaptainNemo]

SETI is ok (I suppose); I just wish they imposed a minimum hardware requirement that stopped people from using antique computers that take days to process work units.

Axis of Stupid = coop, Kanavit, FUGGER, SoDNighthawk, and ninkey.

 

[TheRod]

I just wish they imposed a minimum hardware requirement that stopped people from using antique computers that take days to process work units.

Why? Everyone could help the scientific community, even if it's take a week to process the data! And, for your information, if the results is not sent back to the server within a certain period, the WorkUnit is resent to another user for processing.

--
A7N8X / <font color=green><b>AMD Sempron 2800+</b></font color=green>
Kingston DDR333 2x256Megs
<font color=red>Radeon 8500 128Megs</font color=red> @ C:275/M:290

 

[cdpage]

the last prime number found was done this way and they gave credit to just some guy.

would the same thing not apply here?



<b><A HREF="http://www.digitalgunfire.com" target="_new">DigitalGunfire-Industrial EBM</A></b>
ASUS P4S8X-P4 2.4B - 2x512M DDR333 - ATI 9500Pro - WD80G HD(8M) - SAMSUNG SV0844D 8G HD - LG 16X DVD - Yamaha F1 CDRW

 

[Mephistopheles]

Finding a big, big, big prime number = completely useless.

It's a curiosity, just that.

It is mathematically irrelevant that, say, 2^632561-1 is a prime. (please don't check, it's an useless example). This does <i>nothing</i> to advance mankind...

...so even SETI is far, far more useful than finding monstrous primes.