I think that the idea is that the miners should be mining before you issue the challenge.
The challenge is to show how much the total network hashing power should reduce by when you stop hashing and agree to be idle. Then you observe if block production reduces and only pay the bonus to the idlers if this reduction happens. This agreement is enforced by a combination of nlocktime and multisig.
I don't know if this will work with just one pool doing it (i.e. only a small percentage of mining power) because all of the other miner's on the network will benefit and not just the miner paying the bonus and the slow down in block production will not be noticeable enough.
I think it's really interesting from a theoretical point of view but wouldn't work in practise unless nearly all miners were involved (probably wouldn't work then either - I can think of loads of reasons why)
Here's the video of a talk on it: http://www.youtube.com/watch?v=QN2TPeQ9mnA
If you compare the increase in mining hardware to an arms race then this is the equivalent of a cold war. All of the miners are in a huge stand off, all of them ready to switch on their huge asic hash farms the minute someone breaks the agreement resulting in apocalyptic energy consumption. In bitcoin mining the apocalyptic war has already started so it's probably more likely to end up in mutual destruction than a negotiated peace.
It's a great idea though.
^^^ I will agree with this.
It's an excellent idea, however, getting it to work.......
Remember, what drives the economy is greed (unfortunately)......