Hello!

You've landed in the archive of the Cambrian House community. We've kept some pages here for posterity but the community is no longer active. Now we market the technology that made our early crowdsourcing a success.

Can we help you get to Cambrian House the company? – Come on over.

Are you seeking crowdsourcing technology? – Check out Chaordix by Cambrian House.

Thanks for dropping by
The Cambrian House Crew

Close [x]
Cambrian House

The success of any kind of social epidemic is heavily dependent on the involvement of people with a particular and rare set of social gifts.
Malcolm Gladwell

Cambrian House began as a crowdsourcing community using a wisdom of crowds based approach to discover new business and technology ideas. These pages are being kept online as a technology demo to showcase Chaordix™.

Looking to harness the power of your crowd? Find out about Chaordix™ - technology that enables enterprises to get the most out of crowdsourcing.

Sell your idle CPU time (like SETI@home)

Dig_Dug
Dig_Dug is offlineSend a Message to Dig_DugAdd Dig_Dug as a FriendSend a Hat Tip to Dig_Dug
  • Submitted by: Dig_Dug
  • Created: Jul 22, 2006, 12:17 am
  • Share on Facebook
  • Promote
 

Join Cambrian House

People

Ideas

Businesses

Connect with talented people. Collaborate on ideas. Realize your vision.
Not freeish. Not freesque. It's free!

The Idea

Just like SETI@home, or Einstein@home where users download data to process while their computer is idle. In this model, users are paid to use the client, automatically download a set of work and process it while their computer is idle. The company sells the distributed processing power to companies needing large tasks completed with better throughput. It could possibly be extended for users to seed a torrent for companies who need to distribute content. Essentially, paying users for their idle CPU time and selling the collective CPU time to whoever needs it. A customer would implement an data processing object to an interface which we define. They submit the object along with their data set via FTP. Our system recognizes the submission and queues it into the grid. Our system uses the grid to processes the work, organizes the results, performs various error checks, packages the result and makes it available to the computer via FTP. All automated. CPU cycles on demand - and make it cheap

I thought of this idea when I was...

Thought about this back when I was an active SETI@home user - they created a supercomputer out of a network of idle computers. I thought that it was pure genius. I found that enthusiasm waned over time with the projects and thought that a revenue model might alleviate that problem.


Comments Posted

Aidan
Aidan Posted: August 1, 2006, 4:42 am

I think this idea has some merit. I'm just wondering about the type of customers who would use this? How much would they pay, and for how long?

It would be useful to see some back-of-the-envelope calculations to see how much money the average user would be making with their idle cycles. I'd imagine it would be something on the order of < $10 per month. Is it likely to be cost-effective versus just buying/renting a few high-powered servers?

smeier
smeier Posted: August 5, 2006, 7:50 am

I like this idea very much. Why not make money out of something that would otherwise be thrown away (processortime)?

Buddysaul_BillyBob_Taladega_Johnson
Buddysaul_BillyBob_Taladega_Johnson Posted: August 16, 2006, 9:47 pm

I like the idea too, it seems that heaps of processor time is wasted.

However I think this might be difficult to get off the ground. The CH approach requires that cost for the 1.0 product be kept to a minimum. Only after the market has proven a concept will more investment be made.

I think building a grid, the interface framework, and the custom programming for the first project would be big, big, big. And probably too expensive for CH.

But if I'm wrong I want in on this one!

Buddysaul_BillyBob_Taladega_Johnson
Buddysaul_BillyBob_Taladega_Johnson Posted: August 16, 2006, 9:52 pm

A few other thoughts:

If this idea truely got off the ground and people started making some real money off of it the idea would spread like wildfire. Then everyone would want in - does the demand for grid computing match the avalanche of supply that would happen? Would people still be happy recieving a small check, say 25 cents for 1 years spare processing time? (Assuming supply is vastly greater than demand).

adiuvat
adiuvat Posted: September 14, 2006, 10:14 pm

This is an extremely abstract idea, not without merit, but frankly in dire need of more thinking.

First, what processing tasks require so few instructions to be passed along with the dataset that the total transmission is cost-effective?

Second, what datasets are so small that they can be passed to each individual processor at a bandwidth-cost-effective rate?

"Processing" implies the machine doing the work knows what to do with the raw data. Ergo, you have all the supplementary data you need, and the instructions to actually munch the data itself.

For those applications where this works, it's a killer app. There will be no generalized distributed processing software because the topology of the various types of problems is so varied specific solutions are required.

scrollinondubs
scrollinondubs Posted: December 17, 2006, 10:08 am

This is a great idea- i was about to submit the same one and thought to run a search on "seti" first...

Here's the things that would need to change:
-need to make retrieval of the work product that is produced by your CPU cycles automatic. asking the Yeti@home (for the sake of having a name) user to FTP the files somewhere is way too intrusive.
-need to encrypt the data that resides on yeti@home users' computers. Jobs that utilitize CPU-intensive tasks who would want this service will be animation studios and pharmaceuticals running massive statistical calculations. Neither company would do it if they thought the results of their jobs could be compromised
-adiuvat points out a valid concern - the need to figure out the economics of distributed computing so you know when pushing it out to the edge is valuable. there is an excellent article here on that subject by John Gray of Microsoft-> http://dslab.epfl.ch.../gray-economics.html
-auiuvat's last concern is incorrect - xgrid is an example of a system capable of executing distributed instructions without running special applications local to the machine. it would make sense to evaluate how xgrid for Mac does it. you could potentially just solve the billing portion and leverage something already existing like xgrid for the distributed CPU part.

it's a killer idea though and one i've considered. a lot required to pull it off. i wonder if xgrid exposes some sort of API to tap it- i've seen somewhere on Apple's site that showed the services that can ride on xgrid... I would recommend pricing be structured as an auction model - as CPU availability and demand for cycles changes flux, prices reflect the market. This could be awesome. Built correctly it would allow other projects to ride atop of it.

sean

 

Post A Comment

Got something to say?
Log in to post a comment.