Modern bioinformatics are based on sharing shitty scripts 20-30 years ago. Sometimes you have this good but complex interface piece of software (e.g. https://evolution.genetics.washington.edu/phylip.html) that newbies find impossible to use, so someone in the middle creates a dirty script that does a very specific thing in a single command, in a way that you just replace the input and output filenames.
I can name a number of really bad pieces of software that are widely used because nobody has time/funds to do it properly. Lots of those programs are made by pre-doc students that are learning the basics, and as soon as they get competent at coding, they are absorbed by the private sector.
Yeah, the adage "nothing is more permanent than a temporary solution" exists for a reason. A lot of industries that use software but aren't actually tech themselves tend towards this kind of result. E.g.: If one guy figures out how to script some medical imaging task, he'll just share that with other people who then rely on the same script. I've seen it in academics as well, especially around things like image processing.
I can name a number of really bad pieces of software that are widely used because nobody has time/funds to do it properly. Lots of those programs are made by pre-doc students that are learning the basics, and as soon as they get competent at coding, they are absorbed by the private sector.