Personally, I think s2x will be the longest chain, because BTC hash rate will be inadequate to keep up. According to Satoshi, that makes s2x the true chain. Once the chains diverge enough, and before the deluge, it will be time to dump BTC classic. And yes, I will roll gains into XMR, of course.
Nope, the longest
valid chain is Bitcoin. Barry and Jihan (or anyone else) may make an invalid chain as long as you like it to be (EDA and ASICboost help too), but it will be ignored by economic Bitcoin nodes because it violates their governing consensus rules. No amount of drama can force Core nodes to accept invalid blocks, as we already know from enjoying the lulz provided by Bitcoin Scrypt, XT, Classic, Unlimited, Bcash, and most recently, Bgold.
BTC hash rate responds to economics. Within the incentive structure designed by Satoshi, as tx fees pile up and compete "inadequate" hash rate is a self-correcting problem.
I have no idea if one or the other chain will prevail; I think both coins will survive into the foreseeable future. Thing is, my crystal ball is damn cloudy on this one.
So I'll monitor the fork until I feel it's safe to split my coins with the opt-in replay protection that's coded into Segwit2x.
First, you have to make a transaction where the input is all the coins you want to split, and with two receiving addresses in the output. The bulk of the coins go to an address that you control, and you send a few satoshis to this address:
3Bit1xA4apyzgmFNT2k8Pvnd6zb6TnwcTi
The Segwit2X nodes will ignore this transaction.
After waiting for enough confirmations that you can be confident the transaction is permanent, you can proceed to claim your Segwit2X coins by importing the keys from the old BTC address, which you emptied in the first transaction, to a Segwit2X wallet.
That's how I understand it, and I think this method poses the least risk.
If you have an HD wallet with coins spread across many addresses, this could get to be a pain. That's why I'm sending all my coins to a single address before the fork.