I've seen people who follow the strategy you prescribe, and while they did make above market while they helped companies migrate away from [dying technology] they eventually became unemployable.
It simply isn't a good long term play, for either employee or employer.
Better yet, be the guy who can upgrade a system like that from X language to the newer one that can be maintained by the your new wave of engineers.
For some, the short term payoff is the right strategy. Also how is it not good for the employer? Assuming they could fire the person (or just hire him/her as contract).
On the other hand, you can definitely demand obscene hourly rates for old, clunky stuff most people don't want to touch. Sharepoint is one example I know of. (Not to imply CF is clunky, but the worse the workflow, the higher the rate, obviously.)
Hmm. I know this is completely anecdotal, but the two software engineers I know who are COBOL experts also happen to be the best-paid software engineers I know, and not by a little.
If author wants to continue using CF for his own reasons, that's fine but it's still worth knowing why others care about popularity:
1) job prospects - e.g. knowing Javascript will have more relevance in job hunting than knowing Microsoft's VBScript that was used in Internet Explorer 8.
2) large community that has seen similar problems and can provide answers to copy&paste code : blog posts, tutorials, books, Google searches, StackOverflow, etc. (Relevant XKCD: https://xkcd.com/979/)
3) large ecosystem with lots of 3rd-party libraries, open source repos, etc so one doesn't have to invent everything from scratch
4) the language is continually enhanced with the latest technical features to stay up-to-date - e.g. Delphi didn't have 64bit compiler support for a long time (~2011?) even though C++ had that ability for years. That drove away many programmers that needed to access more than 4GB of RAM.
The common theme to all those bullet points is that programmers are using "popularity" as a rough cost-benefit analysis of investing time into using it. They know programming doesn't happen in a vacuum.