Moishe Lettvin

I love coffee, bicycles, cameras and code. I mainly write about technical interviewing here.

The Myth of the Potato

01 February 2021

A few years ago, when I was teaching interview training classes at Etsy, my co-teacher Tim came up with a great metaphor to describe the goal of an interview: by finding the bounds of a candidate’s knowledge, you’re discovering the shape of a “lumpy potato” that describes their knowledge. Every candidate’s potato is unique, and your job as a team of interviewers is to discover the shape of that potato. No individual interviewer can find the whole, but you put your slices together after the panel of interviewers and get the whole potato.

There are lot of great things about this metaphor. The most obvious, of course, is the word “potato”.

Another is that it embraces the fact that what you’re looking for is boundaries of knowledge – how far along the potato-vector can a candidate go before they run into an “I don’t know” moment. And because you’re looking for boundaries, there’s a necessary discomfort or tension there – reaching the border is uncomfortable.

It’s also a neat metaphor because it embraces the redundancy and also independence of each individual interview – none of them can describe the whole shape, but you put them together afterwards to build a shared model. One interview may describe the piece the potato associated with coding ability, one interview may describe the piece of the potato associated with system design, etc. but all of them are also dealing with the part of the potato associated with communication and collaboration, and that part of the potato is in many ways the most important part, so you want every interview to gather data about it.

But here’s an interesting thing: we wouldn’t expect that the communication piece is shaped exactly the same based on data from the coding interview as it is from the system design interview as it is from the phone screen. Maybe the phone connection was bad, or the candidate (or the interviewer!) doesn’t like talking on the phone. Maybe the candidate was super excited about the particular coding question and made lots of independent progress while narrating and was so excited they missed some details you told them, but got stalled on the design question, and ended up asking lots of questions and listening and incorporating the answers. The shape of the communication piece will differ in each of these 3 interviews, partly because of noise and also partly because of data that simply conflicts. We can easily conclude from this that the potato maps are just maps, and a map is not the territory.

But what if I told you there is no territory?

Here’s the thing: no person acts in a way that’s independent of the environment they’re in. No candidate behaves in an inviolable way, or conveys an idealized cloud of information, in isolation. The act of discovering what a candidate knows and the skills a candidate possesses changes how a candidate shows what they know and what their skills are (and, in the best cases even modifies what they – and their interviewer! – know and what their skills are).

There are some exceptions to this — eg. does the candidate know the definition of a Turing Machine? — but those exceptions are boring and don’t concern us here.

So, setting aside vocabulary questions and their ilk, the potato-shape you discover is a shape that only exists for the duration of your interview, and is molded by you, any other interviewers participating with you, and the candidate. It’s subject to the mood each of you is in, how much energy each of you have, the time of day, the weather, whether you’re talking over video conference or in person, and so on.

When people say “interviewing is noisy” it implies there’s an underlying pure signal there, but there just isn’t.

You may think all hope is lost.

It isn’t.

It’s a profound realization that every interview is about the interviewer as much as the candidate. This is the essence of interviewing.

Think back to your best interviews, as a candidate. I’d be willing to bet the interviewer drew something out of you that you didn’t know was there; I’ll bet you finished the interview realizing that you’d learned something, and perhaps that you’d taught the interviewer something as well.

Be that interviewer. The interview is inescapably a cooperative act if it has any hope of success. The upper bound of success is governed not solely by the candidate but also by you.

It is your job to be generous, and kind, and provide the best possible environment for you and the candidate to do the best you can. Sometimes, you and the candidate together despite your best efforts can only create a potato that doesn’t match the shape you need. Sometimes, you’ll create a potato that fits some of what you need and is lacking some of what you need. That’s okay, this is the nature of interviewing. But you need to finish each interview knowing that you gave that potato all the starch you could. It’s hard! You will absolutely never, ever be perfect at it. But it’s your job to do it well, and get better at it every time.