Biology software is bad because biologists like bad software

OK, not *all* biologists.

I spent a number of years in a small town in the middle part of the US, and every time a good, interesting restaurant opened up, it would go out of business within a year. I eventually realized the lack of good restaurants wasn’t that no one wanted to open one. It was that very few people in the town wanted to eat good food.

Similarly, it isn’t that no one is writing good software for biologists. It’s that when they do write it, it doesn’t get adopted, or at least not as widely as it should. The bigger question is why this happens.

And to be clear, this isn’t meant as a dig at biologists. If I was a biologist, I’m 100% certain that I would like bad software too. In fact, in my days as an academic mathematician (many of which were spent in the small town I mentioned above) I was perfectly happy using plenty of bad software.

The reason biologists tend to like bad software is that they’re used to working in settings where the cost of creating or adopting good software is greater than the benefit.

This may sound crazy if you, dear reader, come from the world of software and data, where everyone is comfortable with the work patterns and habits that allow us to make the most of well-designed tools. But learning those patterns takes a good amount of time and mental energy. It requires new mental models and habits. And if you don’t have the time or energy to learn them, it’s natural to prefer the bad software that you’re comfortable with.

All this is to say that writing better software for biologists isn’t going to solve the problem. If that’s something you want to do, you should absolutely do it, but it isn’t enough on its own. Bad software is a symptom, not a cause. To improve the software that biologists use, we also need to help them adopt work patterns and mental models that will allow the benefits to outweigh the costs.