When analysing a circuit we normally consider a wire to have the same voltage along its entire length. (There are two problems with this: voltage changes only propagate at c, and the wire has a resistance. Normally these are both negligible.) Thus we can view wires as taking a voltage and spreading it out along a line in space.

On the other hand, memory locations take a voltage and spread it out through time. So they are in some sense a wire pointing in the time direction.

Sadly, the analogy doesn't quite hold up. Wires have one spatial dimension but also have a temporal dimension (i.e. wires exist for more than an instant). So if you rotated a wire so that its spatial dimension pointed along the temporal dimension, its temporal dimension would rotate down into one of the spatial dimensions. It would still look like a wire! A memory location has no spatial extent: they're a very small bit of metal (you could make one in the shape of a wire but people don't). Thus they have a temporal extent but no spatial extent. So if you rotated one you could get something that had a spatial extent but no temporal extent. This would look like a piece of wire that appeared for an instant and then disappeared again.

Ponder only the one dimensional time for now. At every point of time, you have only this moment and nothing more. But with the memories, you have same previous moments cached. Stored somewhere "orthogonal" to the timeline.

I've heard it here: http://edge.org/conversation/a-universe-of-self-replicating-code

On a site even better than this and quite unpopular on this site, also. Read or watch Dyson there. As many others.