There are a host of unexplored types of logic which may at some point be important. These include Konstantin Likharev's "Rapid Single Flux Quantum" logic devices as well as the work that Michael Frank is doing with "Reversible Logic".

Then we will probably have single electron switches as well as perhaps some type of photonic switching.

It is not very clear at this point what the best applications of each of these may be.

But IMO, part of the reason Eric used the rod-logic design was that it was easy to think about based on everything from mechanical adding machines to gate arrays. It is also relatively reversible logic which minimized the heat disposal requirements — which are probably going to be significant for 3D nanocomputational devices of any significant size (think of the problems that Intel is having cooling its chips now — it gets much much worse).

I wouldn't say Drexler "predicted" nanoscale rod-logic computers. He used rod-logic computers as a worst case scenario design limitation on nanocomputational power, to see if you could have sufficient computational power onboard a free-floating assembler to run it. He specifically avoids nanoelectric computers because there wasn't enough literature and information to make useful predictions, while rod-logic didn't require much more than the equivalent of the Babage design and the scaling laws developed elsewhere in Nanosystems.

Ralph Merkle wrote a paper in 1990 on logic using buckling elements. These mechanical memory elements buckle. If they put springs between them, they might start to build something like Merkle's proposal.