When I asked him "do you really believe we can get full price for a unit that slow" the answer was "that is irrelevant, the code has to be 'straight'"

Whereas I had to deal with 20 year-old code written by people who had long since left the company, and their motto must have been "Comments are for the weak.". Especially since a good part of the code was written for a totally nonstandard ASIC, in assembly, and had very insightful "comments" like "Add some gain." and "Move data". The rest was in 68k assembly (which I really got to like, that chip has a very nice instruction set), also sparsely (and sometimes wrongly) commented.

But the program itself ran very well. It took me quite a while to come up with something that worked as well. I have to admire the people who wrote it for what the achieved, but not for how they did it ... ;)