Not so "automatically". The reason why the devs like soft forks is that they can deploy the protocol changes without having to alert everybody to upgrade (and therefore to explain *why* the change is good for them).
Are you insinuating that SW isn't good for them? They will eventually change as most users do update their wallets regularly and this has 100% consensus with devs.
I meant that, in general, the devs like soft forks because they can be deployed in "stealth mode", without alerting all the users and nodes -- and therefore avoiding "wasteful" questionings and explanations. See BIP66 last July, and BIP65 that was just enabled yesterday.
In this case too, the devs seem to be decided to roll out SW without waiting for it to be scrutinized and approved by the community. (No independent applications will break, of course, unless they deserve to break.)
I don't know whether SW is good for the devs overall, but some features at least are needed for their planned "bitcoin 2.0" products. I have yet to see why it is good for clients, or even for the relay nodes (the witnesses still have to be relayed, and blockchain pruning can be done even without SW. IN fat, that is what the UTXO database is -- a pruned and indexed copy of the blockchain.
You are assuming that we are already at 100% capacity and SatoshiDice is going to rush in to exploit the extra bandwidth. We aren't using our capacity now, what makes you think that building bigger capacity will insure that it immediately gets filled?
I am not assuming that.
I am assuming that, with a 1 MB limit, the network will each saturation in mid 2016. (Although there has been an extra increase in November, so that may happen in Q1 already.) I am also assuming that, six months after SW is deployed, half the traffic will still be using the old format. Then, even with half the traffic in SW format, the network will saturate anyway in 2016, but perhaps in the second half only.
ETA for SW is moving over from sidechain testnet to main bitcoin testnet this month and deployment in late 15 or jan 16'
I am totally confident that the core devs will deploy such a change in 1 month, with holidays in the middle.
