J2D3

Distributed Lottery

If I had a huge computing task that could only be accomplished by a massive network of computers working together over time, I’d be worried about how to convince millions of people to let me use some time on their cpus.  Creating a botnet like storm is one way… I wouldn’t have to ask… however, that’s hugely illegal and as such presents an enormous potential downside.  Another way would be to just convince people to donate compute-time out of the goodness of their hearts, which was easy with SETI@home… everyone wants to help find aliens!  But what if my project is less exciting and just as important?  I could pay everyone their share of the computing project income… but no one wants a check for $0.28… when they could win $10,000!

So here’s the idea:

There are 2 groups of people involved… those defining and scheduling tasks on the platform (buyers of compute-time units), and those whose computers are completing these tasks in small chunks.  There would be relatively few schedulers and many millions of users.  Say a scheduler has a task that needs 1,000,000 compute units and the cost is $.01/unit.  This project would cost $10,000.00, and 75% of that would go to “the pot.”  Say another scheduler buys 2,000,000 units, then the pot would be up to $22,500.  (This figure would be posted on the front page of the site.)  All participants who are computing would be gaining “entries” per compute-time unit they complete to win the pot.  As a user performing computation, if I had completed 3 compute-time units before the end of this compute project, I would have had a 1:1,000,000 chance of winning $22,500.   Oh, the remaining 25% would go to the company building the platform software (aka me.  and maybe you, too… wanna help?)  These figures are all merely illustrative.

The client end of a distributed compute platform, like a voluntary storm virus, would be downloaded and installed by each participant who signs up.  The user would have a screensaver interface (like SETI) showing real time statistics about his cumulative participation in the platform, the status of the pot, and a countdown clock until the next drawing to see which user wins the pot.

I think the great benefit here is that millions of people will throw a screensaver onto their computer with the hopes that they will win money.  They will not do it for a guaranteed but equal (and therefore tiny) share of the money.  Also, knowing that more participation = better chance of a win will cause even more usage… I just don’t think adoption would be much of a problem.  And it’s not like this software doesn’t exist already… it’s just a matter of turning it into a lottery, a game show, put it on TV!  I bet a lot more people would be trying to fold proteins and cure diabetes with their spare computing time if they could WIN MONEY!  Argh I have to get out of Hollywood.