If they had enough work for a developer, how long would it take some enthusiastic amateurs?

DIY doesn’t sound like a saving here.

Phd in unrelated field means nothing. Physics don't repair their own gas boilers, and civil engineerings don't plaster their own walls. Even if they have a lot of time on their hands and an interest in developing the skill, they still stuck compared to the person who does it for their day job. The physicist will probably kill themselves and their families. The pharma phd won't be able to build a service that's usable by multiple people simultaneously, with authentication, a sound database schema, CI pipelines. It's always gonna be a shitty collection of "scripts" when what they need is software. I think the point is that they have shitty scripts already. Also, who's gonna review their code? Is there even gonna be code review? Or version control?

This pattern leads to some feature incomplete rewrite being done slowly.

You can be able to cook for yourself to survive but not be a great chef. So you might not have high engineering expectations from researchers either, as it's not their job.

Also, if you wanted to rewrite some scientific code relatively painlessly, perhaps D would be a vastly better choice than Java.

Rewriting a silly library like numpy or spark from scratch to make your life easier? Rewriting numpy in java for speed? Hell rewriting spark in java for speed?? That’s a hugely massive undertaking that wouldn’t gain you any benefit.

i have so many questions about this comment