gmaxwell 2017-01-21 14:46:13
Improve block transfer _at best_ reduces node bandwidth usage by 50%, so it was a nice arguing point for segwit. (though testing shows that the savings is more like 12% https://bitcointalk.org/index.php?topic=1377345.0 becuase most usage was actually relay overhead... though we've also reduced that a fair bit since I did that measurement)
gmaxwell 2017-01-21 14:46:41
(50% because all xthin/bip152 do is allow a node to recieve each transaction only once instead of twice (once in relay, once in a block))
gmaxwell 2017-01-21 14:55:46
mryandao: it's my belief that BU didn't intentionally rip off core here without attribution. They worked from Mike's proposal and mike didn't mention the origin. They were unaware of our work or they wouldn't have left in the easily avoided collision vulnerability. (well, were unaware of it for a long time, then unwilling to admit the vulnerablity and fix it, so they still have it)
gmaxwell 2017-01-21 14:57:05
and in our case there is no element of BIP152's design in common with xthin that we hadn't published before long before BU ever touched the subject.
