I remember running into this early on at Quickoffice. I was tasked with essentially re-writing the entire application stack as the existing code was in pretty r

I remember running into this early on at Quickoffice. I was tasked with essentially re-writing the entire application stack as the existing code was in pretty rough shape. C++ was mandated by the platform, but I spent a lot of time getting the STL and Boost up and running. The entire thing relied heavily on templates, including some of the more advanced meta-programming techniques. In each case the 'cleverness' was warranted as it greatly reduced the complexity and redundancy of the code. My boss at the time raised concerns over how maintainable this all was. How we're junior programmers going to be able to work on it? Being all of 22 I naively responded "well, lets just hire people who can handle it." Interestingly, that's exactly what we did. We were much more thorough in our hiring process precisely because we needed folks who could rise to the level of the code we had written. It seems to me that people (and companies) tend to rise to the level of the expectations that you set. In this case, making the decision to use LISP means that you're consciously making the decision to hire the caliber of talent that can use lisp. That may be a good thing (it was in the case above for sure). I suppose it really comes down to making sure that the complexity of the application warrants the use of the more advanced abstraction. A simple web-site for someone selling tractors might warrant a different ("easier") tool than something inherently more complex.