I imagine it as the moment when the line turns back in time :).

I though of it this way too. The problem (and what think disproves this theory) is that when that pair annihilates, it forms a closed loop in time.

To surmount this, I suppose you could say there are N electrons (where N is still less han the total # observed). Assuming not all loops are closed, this could also explain the electron/positron imbalance (more forward-going strings) as well as quantum entanglement (different instances of the same electron).

Experiment to test this theory: allow the positron from a newly created electron-positron pair to annihilate with a second electron. The spin of the two electrons should always be the same.

I think of it as the same particle(electron) arriving at the same point of space at the same time but going in a different direction (in time). Pretty mangled up. And where does all the energy come from then?

It is just a change of interpretation; rather than saying the annihilation causes the release of energy, you could say the change of motion (forward time to reverse time) causes the release of energy. The effects are the same either way, remember Eliezer's dictum, "It all adds up to normality."

The energy was moving backwards in time, and was absorbed in order to reverse the electron's direction.

http://en.wikipedia.org/wiki/Retrocausality "Time runs left to right in this Feynman diagram of electron-positron annihilation. When interpreted to include retrocausality, the electron (marked e-) was not destroyed, instead becoming the positron (e+) and moving backward in time."