anybody know what
@jtoomim is talking about here?:
1|Jonathan Toomim:2016-01-20 08:22:47:and try to get a version of segwit merged and activated, either hard or soft, but without the 4x adversarial condition and the subsidy for multisig
The way the default fee policies and block size limit are coded in Core, "segwit block space" gets a 4X discount compared to "normal block space." For example, a really big transaction like the kind you might need to move coins back from a sidechain gets subsidized (at the expense of miners and those who consume non-segwit block space). This brings up a bunch of interesting game-theory questions. For example, will the miners agree to enforce this subsidy even though they would earn more profit (at least over the short term) by not doing so?
I'll paste my more detailed response about this topic from another thread:
Another thing that's interesting is that the segwit data would be "cheaper" for mostly artificial reasons. Assuming the new "effective" block size limit were above market demand, I don't see why the actual cost of producing "segwit block space" would be significantly different than the cost of producing "normal block space." Assuming that miners are rational short-term profit-maximizing agents, they shouldn't discount segwit transactions at all!
If blocks aren't full, expecting miners to enforce fee policies that discount the segwit data is the same as expecting them subsidize segwit users at the miner's own expense. For example, a rational miner should choose to include a 1kB normal TX paying $0.10 over a 2kB segwit TX paying $0.10 because the segwit TX increases his marginal orphaning risk more than the normal TX. However, according to the default fee policy (AFAIK) he would choose to include the less-profitable segwit TX. From a game-theory perspective, this means that all miners would have a "profitable deviation" by reducing the segwit subsidy.
I'm not suggesting the segwit subsidy is bound to fail. The fees are small anyways and miners might not be bothered to change their default settings. Furthermore, miners might think the subsidy is worthwhile if they believe it will help the network over the long term (in other words,
real miners aren't "rational, short-term profit maximizing agents" like we often assume when making models).
But it is food for thought.
(The situation is different if blocks are persistently full, in which case segwit block space is actually less expensive to produce, but only assuming nodes can enforce (and want to enforce) the segwit subsidy against
significant transactional pressure.)