Comment by BinaryIdiot | Hacker News Reader

Comment by BinaryIdiot | Hacker News Reader

Parent Full thread

> I think your conception of a Ph.D. is off by quite a bit. Outside of their dissertation topic I'd expect a Ph.D. to be on par with an undergraduate, from the time period they graduated.

Maybe? Maybe not? I've had very mixed experienced in interviewing folks where some had zero college and could code amazingly, great at solving problems, etc and I've had the exact opposite. I'm not convinced school education is any type of [significant] indicator for coding and problem solving at this point.

At least this is what's happened in my own experience hiring. Not sure if there are any good studies done in coding proficiency and problem solving versus education (though I could see those being hard to measure well).

> You also said that you're hiring for an engineering position, but you seem to only care about coding. Perhaps I don't know what you mean by engineering, but I'm under the impression that there's a lot more to engineering than any single skill, even by the loose SV definition. How do your Ph.D. applicants do outside of coding?

Well, software engineering involves reading and writing code so that's a huge part of it with the other part being problem solving. Really the problem solving is the most important part, the rest can be taught with time. But essentially we had some very, very basic coding questions (not even trick questions) just to get an idea where the person was as far as their ability to write code. We didn't care what language(s) they knew (they could pick) as long as they could demonstrate some basic knowledge of coding and then problem solving.

Since we were hiring for software engineers, if a candidate couldn't write code in any language they were simply not moved to the other interviews. Not much we can really do there.

Speaking as a Ph.D. (not software or computer anything but traditional engineering), I think you're conception of what a Ph.D. signifies is very wrong, and it's leading you to improperly value what advanced academic training is worth.

Software engineering encompasses the SDLC, which includes a lot more than coding. What about documentation, quality control, or formal methods? There's many roles that I would expect a Ph.D. to be likely to perform well in, more so than someone self-taught. Seems like you're ignoring the bigger picture.

I think we programmers tend to see coding as akin to reading, if you can't code you don't even get the chance to get into other stuff because you're missing the most basic skill of software engineering, the ability to make a computer do what you want it to do. No one interviews for "software engineer" and wants someone who can barely code, writing code is virtually always the number one requirement of the job.

If fact when I hire, I require a small programming project be completed (simple, save some objects to a file and read them back out passing this unit test), that is the resume, if you can't code I won't even look at any other credentials. Rookie programmers with only a few years experience often do far better at this task than people claiming a decade of experience. The most experienced people often completely fail the test.

> I think you're conception of what a Ph.D. signifies is very wrong, and it's leading you to improperly value what advanced academic training is worth.

The job requires being able to develop software. Candidate, who holds PhD, cannot develop software despite being a CS grad. This isn't charity; what would you prefer I do when I get such a candidate? I saw two PhD holders like this which, yeah, could be a HUGE minority but I'm being honest about how anecdotes can create bias and I also mentioned that I felt I did a great job not using my bias against them (we all have bias, remember).

Not knowing the language we use in house? Not a big deal. But if you can't demonstrate fundamental understanding of code structure in any language I'm sorry but I do not have the budget to educate you. I wish I did sometimes; someone who has awesome problem solving skills could be an amazing developer.

> What about documentation, quality control, or formal methods?

No offense but when shit hits the fan, none of those matter. At all. Unfortunately. Plus I think it's good when you make developers think about and do as much of the documentation of their own code as possible. Sometimes seeing it on paper can make you realize "hey, this is too complex to call; I think I can simplify!". At least in my experience anyway. QA / QC, yeah separate group is idea but rarely doable depending on your industry.

> Seems like you're ignoring the bigger picture.

Again, business is not a charity. Big businesses or academic arenas can afford more things but start-ups / small companies? Anything that isn't moving the product out the door gets slowly trimmed away as margins get leaner. It's far worst in the government contracting space where you're legally allowed to only have a certain amount of margin.

A PhD program basically produces a researcher who is the world's expert in a very very narrow area of a subfield. As a job candidate, a PhD holder could offer, in descending order of "putting their degree to use", A) their subfield specialty, B) their ability to work as a researcher, C) being a smart human.

If you're interviewing a PhD for a nuts and bolts software engineering role, you are basically hiring on the basis of C, ie their degree is a dubious value add for you. PhDs are rarely going to be an ideal hire for a startup unless their subfield is literally the core of your business.

As an aside, I am bewildered by the the idea of a CS PhD holder who literally cannot code in any language. But short of seeing their CV, who knows. Maybe they specialized in the more theoretical math- or EE-like areas of CS, and their undergrad data structures course was a distant memory?

> As an aside, I am bewildered by the the idea of a CS PhD holder who literally cannot code in any language. But short of seeing their CV, who knows. Maybe they specialized in the more theoretical math- or EE-like areas of CS, and their undergrad data structures course was a distant memory?

One mostly did neural network research and said his coding was rusty so he has having issues. The other I honestly can't remember. But yeah I'd assuming coding was pretty important but I kinda got the impression from both of them that maybe they had underlings do the grunt work or something. But not entirely sure.

Reply on news.ycombinator.com