I'm not like most scientists.
I love writing code. I enjoy using versioning tools like git. My use of automation tools would make thousands of dollars as a consultant in the industry. I'm aware most scientists just want the computer to do the thing.
Most scientists don't see themselves as coders.
Top Research Facilities realised this early and hired Research Software Engineers — a software person. Mind you, this is not your usual IT that sets up and fixes computers. An RSE is deeply involved in your research and helps keep your code tidy, safe, and fast.
It's not Andy in the basement; it's Rachael next door.
The Legacy Way
I've been there.
I worked in institutes, where the pivotal piece of software most research orbited around was on a single USB stick. It had a big note attached to it, to return the USB stick to the office drawer at all times.
This piece of code. This essential piece of software would never see any bug fixes. There were no updates to it or improvements. Work PhDs did on this software would end up in papers but never on the USB stick.
Other scientists have trouble reproducing any and all analyses because students use a mix of GUIs and small scripts. They do amazing work, but it will forever be lost in the aether once they move on.
This is a waste of time and money and hinders progress.
The Modern Way
Create a dedicated, well-paid position for a Research Software Engineer.
This person can help PhDs make processing reproducible. It can be so easy for someone that understands code a bit better. No one wants to be on the cover of Reproducibility Crisis Monthly.
This person can help accumulate and archive institutional knowledge! When software developments are organised in a central online place, it's suddenly easy to collaborate and further develop ideas from MSc, PhDs and PostDocs that moved to different institutions.
Essentially, it means that every student can build on previous work, suddenly enabling exponential progress within your institute or lab. When three new students don't start at zero, but one step ahead and so on.
Software only becomes more important in research. Now is the time to invest in sustaining it.