I am quite serious. If the longest chain contains a block greater than 1 MB by this time next year I win, otherwise you win. The longest chain is defined as the chain built on top of the Satoshi genesis block with the greatest cumulative difficulty. If Bitcoin forks, then I only win if the "large block" fork has a greater cumulative difficulty than the "small block" fork.
As for escrow, I am open to suggestions. Danny Hamilton and Jonald Fyookball come to mind. We would each deposit 1 BTC into a 2-of-3 multisig address and the escrow would hold the third key.
Suppose there is more than one chain building on top of the Satoshi genesis block and that the one with the greatest cumulative difficulty does contain a block greater than 1 MB. It's clear you (Peter R) "win." But does that mean you get the 2BTC on every still-active chain? Or only on the one with the greatest cumulative difficulty? That outcome isn't clearly specified by what you're writing.
1BTC is too rich for me, but I could put up 0.5 BTC with the following agreement: This time next year (let's say October 1, 2016) Peter R gets the 1 BTC on every active chain that includes the multisig bet output which contains a block greater than 1 MB. I get the 1 BTC on every other active chain that contains the multisig bet output. "Active chain" means that there were at least 20 consecutive blocks mined on the chain on October 1, 2016.
Unfortunately, there's nothing that can prevent BTC from being spent on any chain, unless the hard-fork makes transactions incompatible between chains.
I've read about the process for chain-specific spending in enough of these threads to think it's not so difficult.
1. Obtain coins that are chain-specific. These could either be descendants of coins mined on one chain or descendents of coins that were purposefully separated on the chains. (Purposeful separation might require several tries, but it's clear how to do it and that it will eventually succeed.)
2. Use chain-specific coins as one of the inputs to the tx being signed.
If I remember correctly, Peter R is one of the people arguing:
There won't be more than one active chain. Certainly I've seen Gavin make such an argument. However, it's clearly possible, and any "bet" should take this possibility into account.