Ask HN: How to assess interviewee math skills without test questions?

[astrobase_go]

I work on a small team at a large organization that routinely hires across the engineering spectrum (aerospace, computer, electrical, industrial, mechanical), and we are shifting our hiring culture to emphasize conversation and brainstorming rather than the use of whiteboard programming exercises or test questions that are pass/fail. While we find that this makes for a more inclusive process, our team's subject matter experts can get frustrated at our inability to probe in that manner.

What tips or activities have you found useful to use with interview candidates that assesses their mathematics skills or knowledge without asking them to do a programming or pass/fail test?

[PaulHoule]

A good question for many purposes is "Tell me about a problem that you solved?" which could be about something they did for for work or something they did for fun, the later one is better.

There are many variants of this open ended question ("hard problem", "was proud of", ... always some adjective or adverb that would be emotionally positive for an honest person) and most of the time the recipients feel a sense of relief and end up feeling pretty good about it. Often you find out a lot about what a person (did|does) (know|not know).

It's actually one of the most psychologically revealing questions that will catch the liar and professional bullshitter and leave them feeling like you dropped them 150 miles out at sea into the sharks -- and they deserve it.

I doubt that it works against the C-level psychopath because they know this question is coming so they hire a writer and take acting lessons.

[karmakaze]

There are a few times where my better than average math skills were actually needed to solve a problem faced by the team I was on. These cases are so rare I'd say maybe 5 or less.

OTOH there are countless times my math sense of trends, growth, recognition of foldable cases, etc that I wouldn't be able to assess.

Can anyone give examples of math skills that would be useful day to day that a spreadsheet with simple formulas won't solve?

One example I can think of is solving a recurrence relation to get the big-O bound for an algorithm, but also rarely needed as the algorithm usually shows itself as being log, linear, poly or exp.

[karmakaze]

Of course there's "Fast Inverse Square Root — A Quake III Algorithm"

[0] https://www.youtube.com/watch?v=p8u_k2LIZyo

[afarrell]

What I'm seeing here is two needs:

1. Trust the tech/maths capabilities of hires who pass the new interview process.

2. Clearly represent that need to others in the org who are more conversation-focused.

Do I have that right?

------------------------------

1. Design interviews as close as you can to the day-to-day work -- like pair-programming and pair code-review, but for maths.

When interviews resemble the work, it is easier for the new hire to trust that they can do the work. Some of the best interviews I've done (from both ends) were pair programming and code-review based. When I started an interview by writing tests first and then got hired for that, led me to feel a higher sense of psychological safety during the day-to-day work. I trusted that my need for TDD was aligned with the org's needs and that my ability to discuss technical tradeoffs were sufficient to deliver results. If new joiners trust that their skills are a good fit for their role, it is easier for them to speak up in otherwise-tough conversations.

For more on this, read The Speed of Trust or this blog post summarising a piece of it:

http://eliteaming.blogspot.com/2016/09/core-3-capabilities-p...

2. Represent the value of a solid base of technical expertise to folks who think in terms of conversation and creativity.

First, refer to the need for trust to enable high-quality conversations.

Second, refer to the need for folks who can make technical details accessible.

"We need people with the depth of expertise that can allow them to explain deep technical issues clearly, correctly, and accessibly. That clarity is necessary so everyone in the room can play with those ideas and see unexpected opportunities -- and so they can offer an off-the-wall idea knowing that we'd clearly see its technical risks and make a high-quality decision about it."

https://kottke.org/17/06/if-you-cant-explain-something-in-si...

[giantg2]

What space are you in that you need advance mathematics? I would think that you shouldn't really need more than average skills unless you are building the calculation software that everyone else is using.

[notoriousarun]

Other than math...

Conduct Based Questions - Lead to an insight into the probable behavior of people in the workplace...

Everyone should read... https://amzn.to/3aetw0p

[pryelluw]

But is it math or programming skills you want to measure? Cause math is not programming. Hell, you don’t need to know more than high school math to program.

[jki275]

You need to know more than high school math to write software in an engineering firm though. It's not about the mechanical act of making a computer do something, it's about knowing what to make the computer do and how to do it most efficiently.

[username90]

It is a domain specific skill in many domains.

[sigmaprimus]

Are math skills no longer required to obtain a degree in engineering?

I know that question might sound a bit trollish but it is a genuine one. I thought along with specialty in a particular engineering discipline all engineering graduates were required to have proficiency in Math and Communication.

Are You talking about hiring a person without a degree? Is the competition for qualified employees that stiff these days?

Maybe paid internships for current students could help develop a group of potential future hires, most students are fine with tests.

You might be able to glean some indication of a persons interest in Math by looking into their online histories but I don't know if such a service is available let alone ethical.

[username90]

You don't need skills to pass tests. Many tests can be passed by memorizing solutions and cheating is common.

[Jtsummers]

More importantly, for those who've been out of school for a while their math skills erode quickly. I'm having to reteach people basic linear algebra because they hadn't used it since their junior year of college 20 years ago and are now writing/maintaining programs that require it. Some of them were able to recall it with a bit of prompting, but for others it was a complete blank. For the former, the mechanics were still in their head but the terms were gone and needed to be reconnected, for the latter they aren't recalling either parts. But both groups are picking it back up quickly, they don't need a full semester of retraining, just a few sessions and then a lot of practice/application.

[sigmaprimus]

I understand now and agree with that point of view, thank you for clarification.

Now that you point it out, I have seen the same problem occur with electricians. They are very good workers, get along with everyone, are great team leaders and team members but struggle with simple single variable formulas like ohms law thus preventing them from being promoted to journeyman.

[jki275]

I don't know about college, but I do know for a fact (being the father of a high school junior) that high school math is now the equivalent of what grade school math was when I was growing up.

So kids going into college are far less prepared than they were a generation ago. Maybe the engineering programs in college pick them back up to where they need to be, I've been out of the undergrad world for many years.