Why 4x? Well, 0.12 just made transaction validation like 7x faster I believe with libsecp256k1. The impact on relay, there are technologies that have been discussed earlier like IBLT and weak blocks. So segwit fixes malleability ,allows more Script upgrades, allows fraud proofs, allows pruning blocks for historical data, improves bandwidth usage for light nodes and historical sync, and it's P2SH compatible for old senders so that non-upgraders can still send funds. This gives us time for IBLT and weak blocks to develop, so that we can see whether the relay and propagation stuff can have time to work.
http://diyhpl.us/wiki/transcripts/scalingbitcoin/hong-kong/segregated-witness-and-its-impact-on-scalability/I think even the worst of the big-blocker-or-bust have to stfu and think before looking very stoopid now.
So, all that "concern trolling" about the terrible consequences of large blocks by Adam, Greg, and the rest of the Blockstream/Viacoin crowd was just bullshit,
and they kew it. Segregated Witness will allow 4-5 MB blocks "soon"(because they will not wait for consensus, of course, but intend to deploy it in "stealth mode", aka "soft fork ).
And they are all euphoric about it.
I wonder how the small-blockians who sincerely believed that bullshit are feeling now. Betrayed and duped? Probably not: like that speaker from
1984, they will swicth in mid-sentence from "blocks larger than 1 MB will kill bitcoin" to "4 MB blocks are wonderful". From "we need the fee market to support miners and push traffic to off-chain solutions" to "segregated witnesses is essential to let bitcoin to grow without running into saturation".
To be sure, Blockstreamers only now are coming to realize that an increase of the effective block size limit to 4 MB will have the unexpected and unwelcome consequence of increasing the effective block size limit to 4 MB. So Luke has already proposed to keep the 1 MB limit for the
total block size, including the segregated part. And Greg, in his characteristic straightforward and honest manner, has proposed to combine SW with Adam's 2-4-8 block size limit plan, but "scaled for the effect of segregated witnesses" -- which must mean setting the nominal block size limit to 0.5-1-2 MB.
But they should not worry, since the space savings will only occur if the clients start issuing transactions in the new SW format. Which will not happen right away, if SW is deployed by soft fork. And even after the clients have upgraded, the use of SW will be optional. Will there be incentives for the clients to use it?