Why is hiring broken? It starts at the whiteboard.
Quincy Larson
1.8K105

If not whiteboard tests, then what?

It’s no secret that everyone hates whiteboard coding interviews. They are incredibly stressful, don’t measure day-to-day skills, and don’t predict success very well. Question is, if not coding tests, then what?

  • Take home coding tests are fine, only they have most of the same disadvantages of whiteboard coding tests. They are puzzle-solving, which is not a day-to-day skill. Plus they are generally longer. A company may weaponize this practice by asking candidates to spend hours on a coding exam as a precondition of a phone screen. Developers with choices may feel they don’t need to spend hours to prove themselves before even finding out if they are a candidate.
  • Chatting about projects is a pretty random way to evaluate candidates. It’s so random, in fact, that a developer belonging to a protected class may well feel they were rejected based on their protected class instead of their interview, and file a lawsuit which is difficult for the company to defend. Big companies have the most trouble in this regard, because they have the deep pockets that attract lawyers.
  • Coding tests do have some advantages. They filter out complete poseurs. You can google the infamous “FizzBuzz” example for a coding test that is simple and stupid, but still rejects a significant percentage of candidates. An algorithms test also filters out candidates who never went to college, no matter what it says on their resume. And yes, there are posers, people with no skills whatsoever who know if they get hired they can take home a developers’ salary for 3–6 months before their fraud becomes sufficiently evident that a big-company HR function will permit them to be fired. That’s worth two years on welfare. I’ve personally met several in my career.
  • Some companies do temp-to-hire. I kinda like temp-to-hire because I have no difficulty proving myself. But there are companies who have weaponized this practice as a tool to hold down salaries and benefit cost, and have little intention of retaining the temp after the six month trial period.

Coding tests are just the least offensive of a long list of bad alternatives. If you can find a better way, productize it. It’ll be worth millions.