Thanks everybody for the entertaining discussion in this thread, I learned something new about SegWit (WITNESS_SCALE_FACTOR). Thanks Lauda (and Sipa on IRC) for explaining.
You are welcome. You shouldn't believe the Segwit == doomsday propaganda from randoms when the super majority of developers/experts are in full support of it anyhow.
Absolute bollocks. If SWSF becomes a contentious soft fork, you would still need replay protection.
Correct.
You only need to implement replay protection if you want to cause a bilateral split, otherwise people will eventually unite behind a single chain, the one which has the most proof of work. The uniting behind one chain will happen sooner rather than later otherwise it is a complete clusterfuck.
BU refused to add that to their implementation, which led us to where we are today. If one is so confident about their position, a bilateral split rather than a hostile split with no replay protection is the right way to go. As it currently stands, without replay protection a split would cause a lot of daamage.
It cannot be reversed since it would be an 'anyonecanspend' free-for-all.
A lot of soft forks are "anyonecanspend" for nodes that haven't updated. Segwit is no different to this. Try reversing P2SH as a comparison (the difference being that SW has another UTXO set); it is not impossible, just *very hard*. This is just related to the propaganda that is exaggerating the downsides of SW, which have been clearly written out on the Bitcoin Core website.