Do you need to consider testability during a research phase?

30 days of Testability, day twenty-five

Research and development

Today’s challenge:

Ask your team if there are any areas of the system they fear to change. How could you mitigate that fear?

For this conversation, I talked to my good friend and colleague Amir HajiRassouliha. Amir primarily works on AI / ML researchy sort of stuff. It’s very cool, but the nature of it is, some of it’s commercially sensitive. So I sadly had to cut a bit from this video - you’ll see - but we had an interesting conversation all the same.

When should you start thinking about testability?

I always like talking with Amir, he brings such a great perspective on so many things. What I really thought was interesting was this: A lot of research and development type work involves experiments, and proof of concepts.

How much do you need to consider testing, or testability, with this kind of work?

I don’t know the answer. It probably depends a lot on the context of what you’re working on.

My gut feeling is, if you want to build something quickly that might be thrown away, then you probably don’t need to consider testability so much.

But as soon as the experiment starts to take shape to something more viable - then testability needs to be considered right away. So that it can be built in early, and you don’t end up having to try and retro-fit it later on.

Would love to know what other people think though!

Thanks

Thanks so much Amir for joining me - and as always, thanks everyone for watching and/or reading, see you for the next one!