role of underlying metal
Bill Tindall
These coatings are very thin, only a few microns in thickness. Hence, they provide no mechanical properties to the object, only potential abrasion or corrosion resistance. For example, I hardness tested the TiN coated blade and it didn't matter whether I tested where the coating was applied or where it wasn't. If nothing else, a hard and tough material under the coating is required to prevent dulling by edge deformation and chipping.
What kind of abrasion resistance is optimum under the coating is a question worth pondering. A beaver maintains sharp teeth by employing a tooth with an abrasion resistant outer layer and a soft interior. Such a structure is self sharpening and it works well for milling wood. I was told by the coating companies that this situation was replicated with TiN coated tooling, such as end mills and the like. This is not the situation we found for the plane iron as revealed by the pictures of the TiN coated edge after planing maple.
It is becoming clearer that the topic of dulling in hand tools is complex. In these cases optimums are often found by trial and error over a long time, a variety of evolution. The current problem is that we have had a discontinuity in the "trials and errors" resulting from a pause in wide spread hand tool use(machine age) and the appearance of new technology (new metals, new fabrication, new woods, etc).
I have no doubt that a Japanese chisel fabricated traditionally is the optimum for that metal, wood material worked and process of working it. They had many years to evolve the tool and product in a stable environment. Now change the game by introducing western joinery, African wood, modern steels and fabrication, amateur craftsmen and it could well be that there are dozens of optimums for these combinations with the result that the optimum for your combination is different from another's combination.
Thankfully, they are all "good enough" as evidenced by the fine furniture being made by persons using a great variety of tooling.