Math could help create proper incentives to existing behavior for stakeholder to change their behavior, but the EIP doesn't actually rely on math to create incentives.
It relies on wishful thinking that a non-incentified action (flagging given the risk of retaliation) will just materialize and the wishful thinking that these actions will result in a fully rational non-emotional response to flags, end in doing so create a situation where the math might end up creating the desired insentives.
If people are too scared to flag the EIP math won't kick in. If people act emotionaly on flags and start retaliating, the EIP math won't kick in.
I think there is maybe a 20% .. 25% chance the EIP might work out, and there would nothing wrong with taking a gamble on this if there was an actual contingency plan. Unfortunately though, Steemit Inc is already on to SMTs, as if HF21 already happens, and I don't believe any of the top witnesses learned anything from HF20 about how important contingency planning actually is.
Yes, the EIP most likely will backfire, but that isn't the bigest of problems. The biggest of problems is that contingency planning for HF21 seems to be even less than it was for HF20 at the moment, so if things go south, there will be no plan B.
RE: Mathematical Incentives Only Work, If The Stakeholders Use Them