This is actually may be a valid reason, unless the specific code can be supported by one person and this person doesn't mind being chained to a desk. If the company has code that only one person can understand, the company has a big problem. Of course, this can be solved by hiring more people that understand this, but it may not be always easy/practical/affordable/feasible.
>>> I think the stagnation in the development and popularity of many mainstream languages underscores this point.
Slow development of mainstream language is a good thing. Mainstream should be stable. Exciting things should happen in the cutting edge and then be slowly brought into the mainstream.
Code is a form of communication too, and it must be done in a way that is understandable to fellow developers, otherwise costs of development raise immensely and the whole project comes under threat.
Speaking of "strange tools", Rich Hickey mentioned the phenomenon of helping commoditize yourself by shying away from the unfamiliar (in Simple Made Easy).
Often, to even suggest an alternate way to look at things—perhaps functionally—or to advocate a new method be learned is often met with disdain.
I think the stagnation in the development and popularity of many mainstream languages underscores this point.As a small anecdote, there has been an interview from which I was rejected simply because in a programming puzzle, I employed high-order functions to solve the problem, without:
The sad thing is that this kind of behavior wasn't local to this particular company. It runs rampant.An industry which is focused on results generally occludes importance in the path to achieve the results. And it is often the path that can be optimized by some metric by use of intelligence.