The difference between LENR and bipolar transistors is that there was a theory describing the underlying phenomena for the latter. The concepts of bandgap, Fermi level, current flow, holes, and electrons were all well established, even if they underwent further refinement, during transistor technology development.
While it's certain that I'm preaching to the choir, it seems clear that the place to expend our energy is to identify a theory that can be tested. At first the testing will be limited to confirming that nuclei can be brought sufficiently close together to achieve a reasonable probability of interaction. (The assumption here is that LENR may be described as a tunneling phenomenon.) There exist computational QM packages that should be able to do this, although the optimum algorithm is not clear to me. On the upside, the number of particles required to comprehend the LENR environment is likely to be small compared to many MM problems, and this may permit the use of fairly accurate algorithms.
The next step would be to design materials with the necessary nanostructure to permit LENR phenomena to occur. Right now, CMOS technology is able to reliably construct <10 nm features. So it would be useful to understand of these feature sizes are small enough to produce LENR effects. If the feature size is in the 5-10 nm range then it should be possible to fabricate an LENR capable surface using standard CMOS fabrication techniques.
