Yesterday I wrote about Folding@Home and how spare computer cycles can potentially make a big difference in research. I’ve be running the client on my home PC for a little while and have already come up with some other ideas.
If computer clusters (distributed computing) can figure out the dynamics of folding proteins, what else can they do? And how would someone go about setting up such a scheme?
The basic problem seems to be that no one gives computing cycles much value. If some form of money were involved, you can be sure that the motivation would be much higher.
This is the basic idea. If you could set up a brokering system that would allow for trades between users and systems, you could potentially sell and buy computer time from just about anyone. Please let me know if such a system already exists (sometimes it takes me a while to find out about these things). If you could sell your spare cycles to someone and that someone could use the cycles along with other people’s cycles, both sides would win. The person with the task would benefit from the raw power of the systems combined whereas the users would get money or credit to do what they wanted at some other time.
The overall concept introduces the basis of an computing economy on the Internet. To give you a primitive example, it would be possible for a movie house to produce a digital animation movie using the bought computer time from the individual users. Instead of spending a fortune on in-house systems and not getting the performance they want, they could pay for a collective net of computers to do the work for them. In this way, it is not much different from the concept of Folding@Home but instead has the goal of producing a high quality film with the lowest cost. It’s a bit idealistic given that there would be a need to safeguard both sides from potential abuse. However, the model does make sense and would justify investment.
All this flows with the concept that ideas are going to become more powerful than the things they create. This is largely due to the dynamic aspect of ideas that can be applied to many different areas at once.
With the idea of using the credit model, you could build up your credit with a particular group of systems so that when you do want to do something complicated, you can do it in a burst instead of waiting for hours. In fact, this model makes even more sense that I originally thought. If the computer is idle, it can be doing work. This work is stored as credit on the net and ready for the next time the user comes back. The user can use the concentrated work during the relatively short period of time that it is needed. It’s kind of like charging batteries. It takes a while to charge but once you have a good charge you can do lots of things. In this model it is even more focused. It is more like charging for a flash. You build up over a time and then you do something amazing for a short jolt.
NOAA Lightning Page
I’m being to like this idea even more. Any one care to brainstorm on this one in comments? I would love to hear from you.
Sounds like a flashback to mainframe days when everyone was charged for CPU Usage! Just another thought though, if a modern PC is idle it’s not wasting cycles because the processor isn’t running; it shuts down and saves power.
Yes, there are some similarities to chargeback from the mainframe era. It is a bit different since the machines are not centrally owned and therefore some form of trade would need to take place.
As for the processor cycles, you are right about the power shutdown. In the past it was a lot more obvious that the processor was just wasting energy by being idle. Now the lines are a bit more blurred.
I think it is more about the fact that the computing resource is not being fully used. The processor power would be fairly limited. I’m not familiar with the power ratings of the latest workstation machines but I would guess that it is less than 20 watts.
Overall, I’m very interested in investigating the idea of injecting an economy into the computer landscape. I don’t mean to rejuvenate the model of central control from the mainframe days but rather help to create something new and dynamic.
I think it’s a great idea in principle, but I’m not sure it’s economically viable; this viability depends, of course, on how much people would expect from their computer’s spare cycles, and how much potential users would be prepared to pay. My concern is that the market would be illiquid; the minimum demanded would be more than the maximum anyone would pay.
For example, would you rent out your machine for ten cents a day? Probably not – it would only net you $36.50 a year, hardly worth the effort. Would you pay as much as $1 a day to rent (most of) a machine? Probably not, especially if you were needing to rent multiple machines for a long period. There may be a happy medium, but it will vary on a case-by-case basis, partly depending on the value of the outcome to the buyer.
I would tend to agree Hugh. It would not be a very good model if based on a bartering type system based on money. The values would be fairly low.
However, it would still be valuable if you could trade computer time. The currency would exist in the form of work.
If I could store my work effort for the sake of others and then “spend” it later on, I would derive some value.
As a primitive example, if the computer did work for one hour for other computers, it would have one hour credit stored. Concentrating this work into five minutes, I could justify having 12 other computers to help me do that work. This is assuming quite a bit but you get the idea. It’s like saying you can concentrate the power of your machine by balancing it against other machines on the net or LAN.
I think it is key to remove the concept of financial gain. The real gain is computing power and specialization. It is like money but it should not be based on money that we know of currently.
I feel like I’m talking in circles. There is potential for it being exchanged for normal money but I would rather not focus on that first.
A more recent development is that big efforts for computing are not based on money. It is more based on competition and recognition. I certainly do not have all the answers. I’m glad that people like you are taking an interest.
What would you like to see in this field?
There is a human equivalent of the specialisation aspect – in the UK we have “LETS” schemes (Local Exchange Trading Schemes: http://www.letslinkuk.net/ ) which allow people to exchange services with each other at a rate defined by a notional currency, whose name differs for each instantiation of the scheme. (In Cambridge the currency is called Cams, in Basingstokes it might be Basings, etc.) The Inland Revenue have so far been persuaded to keep their snouts out of the trough, too, which is good.
So, for example, if I did two hours of Maths coaching for person A and provided person B with an hour of Swedish massge, I could get person C to come round and tidy my garden for 3 hours. It’s a very handy way of doing things you find easy in exchange for someone else doing the things you hate. It’s not quite clear how this could translate to the IT world, though.
That’s what The Charity Engine is about to do