Comment by A_D_E_P_T
You write something at T, under normal background conditions of increasing entropy. Entropy at some T+n fluctuates to a minimum, and at T+n+1 begins to rise from that minimum. None of this appears to necessarily reverse what you've written at T?
If you've made some state change that amounts to irreversibly changing information, then you've increased entropy, full stop. I believe TFA is obviously correct that closed timelike loops involve forgetting, thus no time travel paradoxes.