A Harvard Business Review blog just published a rather ridiculous piece about managing “creative” people that sounded a bit like “tuck them into a playpen, toss them some toys, and expect greatness to come out. Oh—and don’t pay them too much money or they’ll turn on you like the ungrateful infants they are.” I’m not linking to it because I’m pretty sure the article was meant to elicit a response from a lot of annoyed creatives who’d get all uppity and send the blog’s click count through the roof. I’m not playing.
Then there was the article about Tableau and how it’s going to create a “data democracy” that’s free of IT and data scientists. Real freedom, apparently, is getting rid of all those pesky employees who sink hours of their day into figuring out how to best use your hard-earned information and replacing them with license fees and vendor lock-in. Because no software giant would abuse their customers by jacking up the cost of the software once they know you’re dependent on it. That article is also not linked here. I’m still not playing.
Instead, I’d like to have a word with you guys—my fellow data scientists—about the choices available to you right now. In general: You do not have to work for people who don’t respect your work. You certainly don’t have to work for peanuts, and you don’t have to slave away developing a skill set with someone else’s product in order to land a job. SAS certification, Oracle DBA training, SAP superstar—or whatever the kids are calling it these days—tools come and go. You don’t want to participate in vendor lock-in, either. You are going to work for a long time, and today’s hot new tool is tomorrow’s old news.
What you do need are basic skills, skills that let you transfer between tools. When we started Research Triangle Analysts, we described it as a “math user group”, a place for analysts of all stripes to come together and talk about techniques. I think that instinct comes from some dissatisfaction about how the analyst community is fractured around software. The SAS user group doesn’t hang with the R user group. The Oracle users laugh at the SAS people and the SAS people think Oracle is a waste of time. It’s fine for the companies and their employees to feel that way, but tribing up around a tool when you don’t make a dime from selling it is a waste of your effort. Your asset is you, and your skill set is your brand.
I think it’s hard to come to the job market from the angle, and maybe professional organizations like ours need to do a better job of communicating to hiring managers that their approach to getting talent is sub-optimal. I recently had a frustrating talk with a recruiter who insisted that I didn’t know Hadoop. I was tired and already done with that guy, so I launched into a five-minute diatribe about the mathematical implications of using MapReduce in a nonlinear environment without careful consideration for the design space. When I wound down, there was a pause, and he said, “Yeah, it’s too bad you don’t know Hadoop. The customer could really use someone like you.”
So, I’m pleased to see that LinkedIn is putting little notes in my profile asking if I have portfolio work I want to share, and I’m glad that RTA is filling up with like-minded people. Good information is too valuable; smart businesses will figure it out. In the meantime, let’s play our game—the one where we do good work and find interesting relationships in the data that can help our clients and our companies do what they do better.
And let’s stop playing theirs. Let’s slow down the incessant blogging and the linking and the tweeting. I’m not saying stop. I’m saying let’s create information that’s useful. John D. Cook’s blog, The Endeavor, is an excellent example of what we need in the analyst field, as is the Revolution Rstats blog, edited by David Smith. John Sall and Brad Jones from JMP post some great content about statistics and software development.
So there’s some examples of what to do (and a few of what not to do). Let’s get going!