Hacker News Reader

Comment by jankedout | Hacker News Reader

There's data scientist. You don't need a PhD to pursue that. I mean, you probably won't be researching cures for cancer, but you'll still be applying research methodologies. You'll be getting paid more and probably won't have to wait half your lifetime to get a plum position. I abandoned grad school after seeing the rigor mortis that has set into academia and private research companies.

View on HN

I work as a data scientist at a nonprofit research organization (rti.org) and have to apply far more skills than the PhD researchers here. While we fall back on them for depth in subject matter expertise, we also have to understand conceptual design, research methodology, stats, newer data sci methods, software development, 3+ programming languages, user experience design, tech stacks, database design and deployment, data visualization, business consulting, public speaking, etc... Often, we lead client engagements, determine project deliverables, co-author academic papers, present at conferences, schmooze with key stakeholders, and so on. ALL of us, however, would lose our minds if we had to focus on a single domain area or spend a protracted amount of time on one problem. I worked on > 50 projects across the biological sciences, social sciences, and humanities in 2015. It can be exhausting, but it is a real blast most of the time.

So... Consider the path of a research data scientist if that sounds fun. It's also very lucrative, which is nice, too, though I'm not personally motivated by the finances of it.

I'm confused. All the skills you cite are exactly the ones I would expect a PhD researcher (esp. in math, comp. sci, stats, etc...) to be proficient in (although, perhaps not the schmoozing). You're saying your PhD researchers aren't skilled in research methodology and stats??

Ehhh. I suspect an information asymmetry problem -- that is, the GP doesn't know what his/her colleagues actually know -- coupled with a tendency of researchers to let work drift by when it isn't in their wheelhouse.

While I've known a lot of PhDs who know nothing about statistics or programming or some other specific skill (nobody is an expert at everything), it's pretty hard to make it through a good graduate program while knowing none of these things. What generally happens is that domain experts try to keep their minds focused on their domains (where their value is highest), and let non-specialist work fall to generalists. Generalists then (sometimes, incorrectly) assume that the specialists are useless outside of their niche.

I'm not going to say that there aren't incompetent PhDs, but it's a bad assumption to make, in general. You don't assume that your CEO doesn't know how to clean a toilet simply because she lets the janitor do it.

What comes with the PhD positions is a requirement to respond to RFPs with concept notes, proposals, etc... and a lot of high level project management. They provide subject matter expertise from an analytic angle (usually without getting involved in the technical aspects) but almost never are involved in modeling, development, or anything else that I listed above. We're seeing change wherein data scientists more often are involved in the proposal and business development process, but not wherein PhD-level researchers are involved in the technical implementation of projects.

These aren't your grandfather's PhD researchers.

Reply on news.ycombinator.com