| You're looking at this through a purely technical lens. The guy who "pokes around with Excel" probably operates in a business context. He interacts with people who have no clue about data science, and is able to use the data to tell a convincing story. This can be dangerous if he doesn't know what he's doing, but 90% of things people want to use "data science" for are pretty trivial technically and probably can be done in Excel. The guy designing a novel algorithm probably operates in a technical context. People like this tend to be very "in the weeds" and incapable of succinctly explaining their findings to people without the same context they have. This is a universal problem -- people who are extremely technically skilled often have trouble explaining their craft to, say, a marketing exec wanting to know how a certain characteristic is derived. In fact, the marketing exec will probably call in the Excel data scientist to translate. Does this mean the guy designing the novel algorithm is somehow lesser? Absolutely not! But when you choose a deeply technical career path, you run the risk of losing the external context. This is why many companies have managers in engineering who aren't super technical -- they're technical enough to understand the jist of the concept, but their core skill is communication. If they're doing their job well, the engineers are left alone to do their job without senior business people sticking their noses in everything. Coincidentally (or maybe not), I think the "soft skills" are sorely missing in this skills matrix. Every engineer will have to give a presentation or work with an external team at some point in their careers, and some are better at it than others. In my opinion, the guys with hardcore engineering skills are great, but someone with solid engineering skills who can communicate well is a rock star. You can replace a badass engineer, but you can't easily replace the cross-team relationships that a good communicator has built that can often short-circuit requirements problems before they get turned into code. |
In the CS encryption is probably the best example of this where the basic algorithm can be identical between a system protected from the NSA and something trivial to break for the average researcher.