It's not creating new energy, right? It reads to me like causality violating configurations simply move energy around time as we'd move mass through space.
The second law, like all physical laws, does not violate conservation of energy. However, reversing causality can cause the reversal of entropy, which is what the second law deals with.
It's not necessarily about reversing causality, it can simply be breaking it (ie, the effect happens before the cause without being able to influence the cause).
I would envision two chambers A and B connected by two vents X and Y,
like Maxwell's demon but with two vents instead of one and both vents
permanently open. Whereas X is an ordinary vent, Y has the property
that any particle passing from A to B through it reaches B shortly
before it leaves A. Any particle passing from B to A through Y
disappears for a short time before reappearing in A. It seems to me
that this setup would induce a current of air blowing from A to B
through Y and from B to A through X, but I'm only an amateur and could
be full of hot air myself.
At some point a particle pops up into existence in B, increasing B's pressure.
Some time later, a particle from A disappears, decreasing A's pressure.
This pressure differential would cause flow from B to A through X.
Alternatively, a particle randomly disappears from B, decreasing B's pressure.
Some time later, a particle appears in A, increasing A's pressure.
This would cause flow from A to B through X.
The more particles that cross from A to B via Y, increasing B's pressure, the more likely it is for particles to pass from B to A via Y, keeping these two effects equal.
These two pressure differentials would cancel out for large numbers of particles.
You've essentially just hooked up the system to itself, but at a future point in time. Unless the overall pressure of the system is changing, there is nothing to induce the flow. If you are increasing the pressure, valve Y is essentially just a standard one-way valve.
What may add the arrow of time is the second law of thermodynamics: a statistical evolution of a system is not symmetric.
I wonder now how a violation of causality could either circumvent of contradict the second law.