I was whining over the phone to a close friend and life collaborator yesterday. I was talking about how I wasn’t sure I was working on the right things, what’s my value proposition to the company, yada yada. I was having a little moment. I started pointing out how sometimes it feels like I do work very quickly, and maybe this is just a sign of expertise? I’ve seen a lot and maybe I can just work more efficiently now than ever and haven’t really recognized it. His response was, “sounds like the parable of the ship repairman.” 

 

Hm, never heard of it. But read it here. 

 

I think there’s so much to unpack in this short story, and I’ve been thinking about it all night, both in the context of higher education and what it really means to be a data scientist. Early in my data science career, I thought companies hire me to make models. In other words, tap the hammer. I can get paid a lot and be so valuable and happy if I can tap the hammer well! 

 

No. Lots of people can tap the hammer, several even tap the hammer well. Very few know when to either not tap the hammer or where to precisely tap it.

 

Which begs another question: If tapping the hammer is the easy part, then why do universities just train us how to tap the hammer instead of spending time on knowing where to tap the hammer? Wouldn’t we better prepare students if we focused more on the latter? I think the simple answer is that you must know how to tap the hammer before you can learn where to tap the hammer, and universities are set up to provide deeper technical skills (and perhaps rightly so). Knowing when and where to tap the hammer can only be learned on the job, and it’s a real challenge to replicate that in a university setting. I think for a long time, I thought college was simply a replication of what I would encounter in the ‘real world.’ I now understand that college is but one slice of the pizza. It gets you started, but to become a valuable data scientist, you need to move forward and start learning the next parts. When and if to tap the hammer. You already know how to tap the hammer. 

 

So data scientists, listen up. Companies don’t value your model creations. They value the skill of knowing when to not create one in the first place.