I remember when I was doing my Ph D thesis, 50% of the lab team worked on VLSI cad tools (Alliance tool suite), and a lot of published algorithms were recoded. Part of it was because it was faster to recode than to merge foreign code with our own frameworks. Part of it was to keep interns busy. Part of it was "not invented here" syndrome, with a bit of "I can tweak it faster than the morons" neurosis.

Pierre