If we were in the Wall Observer thread, I'd say "fair game" but we're trying to discuss the likelihood of a chain fork and what level of risk in that context is justifiable.

Why would you assume that? The most important thing to consider here is that miners are working off incomplete information. They don't really know how many nodes are running what implementation as it's very easy to run fake nodes. And it's nodes -- not hashing power -- that determine the validity of a blockchain. It's a more diverse and interesting question than most realize. Miners are pretty centralized. I think this is why Gavin is targeting them: it's much easier to trick a small number of highly centralized mining pools than it is to trick thousands of node operators. And if the 2MB implementation is capable of triggering the rule change based on hashing power (at 75% or whatever bullshit "democratic" threshold Gavin & Co. come up with -- 51%, etc.), then everyone else will crumble in submission, right?

Well...the dozen nodes that I run won't. The definition of "majority" and "minority" chain can change in a heartbeat; that's just a matter of miners temporarily pointing their hashing power at one chain or the other. It doesn't matter what Coinbase and Bitstamp say now, or where Bitfury points its hashing power. What really matters are the nodes that determine block validity, and what proportion of them enforce the new fork's consensus rules. Because if a significant proportion of them enforce the old rules, we will have an irreparable chain fork. These irrelevant musings about how a majority of hashing power will render all other blockchains instantly dead are amusing but not very informative. If nodes do not approach consensus, miners will have to choose which fork to built on top of. But, which one? All of the Classic/XT rhetoric says that a temporary majority of hashing power will surely solve everything. But what the hell does that have to do with nodes? What proof do you have that Classic nodes will comprise a majority of nodes -- simply because Bitfury and a few mining pools upgraded (if that happens at all)? Well, if a majority of nodes continue to enforce the 1MB rule, you may find quickly that the "majority chain" isn't a very meaningful phrase. It's all about validity. Miners will point their hashing power at the longest, valid chain. If it isn't clear which one is the longest valid chain (due to no clear consensus among nodes), we will have multiple blockchains and this will be irreconcilable. IMO, the most likely outcome of that is for mining farms to shut down en masse and for difficulty adjustment to drop significantly, as miners cannot risk expending resources to build on potentially invalid blockchains. The market would likely never recover -- probably rightfully so. For this to happen would mean that the only mechanism to enforce rules within the bitcoin protocol was broken, and all it took was the prodding of a loud minority.

By the way, you know that pre-fork coins could also be sold off on majority-fork exchanges? Particularly because early adopters might be a little pissed off at the commit keys for bitcoin's dominant implementation being in the hands of a junior dev who wants to make the question of inflating the money supply a democratic one (jtoomim). How do you know who controls millions of pre-fork coins? You can be sure that I'll be dumping everything the second Toomim gets the keys to the kingdom, and I know several likeminded people.