I actually agree with your assessment and did think about the issue you mention. I chose this imperfect set of trade-offs because I didn't want to DIY in-wall fixtures and, after reviewing the current tech and market and which "open" standards were being developed, I decided to go with an inexpensive shorter-term solution (~4 to 8 years). The dimmer switches are only $15 each delivered from Amazon (https://smile.amazon.com/Dimmer-Martin-Jerry-Compatible-Neut...) and under Tasmota are highly configurable to the extent of being able to communicate in a switch-to-switch private mesh.

Since I had to commit to hardware six months ago, I knew the kind of longer-term hardware I'd really prefer didn't quite exist yet - at least not at the multi-vendor, commodity-cost level. There are some interesting high-end, proprietary, single-source products as well as some cool experimental DIY & low-volume kit-type designs, using stuff like touchscreen displays and ESP32-type CPUs. I think this kind of flexible display/interaction end-point will be the most likely next inflection point but it will first need to go through the early-adopter Darwinian process sorts and then come down the price-curve to reach Shenzen-container, multi-vendor commodity pricing.

My choices were to either

A. Spend too much now making an early bet on something available today that probably wouldn't be a longer-term keeper.

B. Go with a traditional "dumb" dimmer for $6.50 each and upgrade in a few years.

C. Pay $15 each today on a shorter-term, standard-sized, replaceable "smart" solution that can fall back to a dedicated wifi subnet for control or, worst-case, be a stand-alone dimmer.

I decided to go with C. While I agree with your point that the wireless protocol world will eventually move on, I suspect that wifi as a protocol is so ubiquitous, so standard and so cheap that some form of backward compatibility may survive for decades through bridge devices. A likely contender for the next big multi-vendor standard in home automation is likely to be the ChIP-based standard being pushed by Google, Samsung, Apple and a couple hundred other companies. The standard has been finalized and the wireless will be based on both wifi and Zigbee (2.4ghz and 900 Mhz) as a basic required element, so there's a fair chance my current cheap "Shenzen-specials" will be adaptable through firmware since the vast majority of in-wall wireless stuff installed today is based on wifi or Zigbee. It's okay if they aren't because in my mind they are a short-term disposable solution anyway.