Sadly, you're spot on
These guys are in the hot seat now because people actually care about their results. Most bits of science are not like that, and software support is hopelessly underfunded across the board, even when the science depends on it. An application rarely has more than one developer at a time, often just a phd student, and quality control is typically nonexistent. Data analysis is even done in Excel. It is just depressing. So yes, software-dependent results in science should be treated with extreme caution.
e.g. http://www.americanscientist.org/issues/pub/wheres-the-real-bottleneck-in-scientific-computing
"For every successful simulation of global climate, there were a dozen or more groups struggling just to get their program to run."