On Thu, May 28, 2015 at 01:19:44PM -0400, Gavin Andresen wrote:
> As for whether there "should" be fee pressure now or not: I have no
> opinion, besides "we should make block propagation faster so there is no
> technical reason for miners to produce tiny blocks." I don't think us
> developers should be deciding things like whether or not fees are too high,
> too low, .....

Note that the majority of hashing power is using Matt Corallo's block
relay network, something I confirmed the other day through my mining
contacts. Interestingly, the miners that aren't using it include some of
the largest pools; I haven't yet gotten an answer as to what their
rational for not using it was exactly.

Importantly, this does mean that block propagation is probably fairly
close to optimal already, modulo major changes to the consensus
protocol; IBLT won't improve the situation much, if any.

It's also notable that we're already having issues with miners turning
validation off as a way to lower their latency; I've been asked myself
about the possibility of creating an "SPV miner" that skips validation
while new blocks are propagating to shave off time and builds directly
off of block headers corresponding to blocks with unknown contents.

-- 
'peter'[:-1]@petertodd.org
00000000000000000327487b689490b73f9d336b3008f82114fd3ada336bcac0