You Should Have Two Different Kinds of Hiring Interview

Most companies and managers today are making a fundamental mistake in their hiring process. They are conflating two different types of filter—the positive filter and the negative filter.

When most people interview they’re actually trying to do two things simultaneously.

  1. They’re trying to filter out bad candidates.
  2. They’re trying to select the very best.

Whiteboards and puzzles

By now you should have heard that requiring people to code on a whiteboard, or do brain twister puzzles is a really bad way to find talent. It’s bad in both directions, i.e., the people who pass are not necessarily good at the work you will be having them do, and it also leaves your best people behind.

But what I just realized recently is that these types of exercises are also bad because they try to smuggle in both filter types into one activity. If you fail to answer the brain teaser correctly you’re part of the discard pile, and you’re also not part of the keep pile. Beautiful!

There’s a better way

I believe the better way to hire is to clearly move through two separate filters, which are configured in completely different ways.

  • The first is the negative filter, which looks for fundamentals and a lack of obvious no-gos. So, you should probably know some basic computer science fundamentals to be a programmer, not be horrible to work with, not be a collector of human skin suits on weekends, etc. This filter should be relatively large-grain.
  • The second filter is the positive filter, and it’s not a list of checkboxes like the first one. In fact, the entire concept of this filter is that nobody would really know what that list of checkboxes would be. As such, the best way to filter here is using the Evolution Method. That is to say that you put them through a series of real-world tasks, for as long as both sides can tolerate, and make your selection based on who performs the best.

In Design Thinking a few people think they have the best questions, the best checkboxes, and the best nose. They can just pick em’, in other words. Which is largely crap. Google couldn’t, and they had millions of dollars and a couple of decades to spend on the problem. The first step to better hiring is accepting that you are not all-knowing, that you don’t have the perfect nose, and—most importantly—that talent and ability will surprise you.

Evolution Thinking is the system you go to once you’ve accepted this. It works by taking your pre-filtered group and trying them out as much as possible. There are usually limitations here, which is unfortunate because the longer the period the more it will map to reality, but even short periods can be quite instructive.

Create sets of real-world tasks and give people real-world environments, time-periods, and resources to accomplish them. Hybrids are best. Have someone describe a problem, how they’d approach it, have them create some or all of a solution, have them communicate it to a peer, and then have them summarize the project and its results to others.

Smart readers might notice that these too sound like checkboxes. Nice catch.

They’re like checkboxes, but not quite. Actual checkboxes would be attempting to capture the exact je ne se quoi of the activity, like “describing the problem”, or “presenting to others”. The power of this approach is that when you model reality in the task it actually has thousands of opaque and inscrutable checkboxes nested within. Checkboxes that you and I aren’t smart enough to capture.

The other side of evolution

Evolution is the metaphor we’re using here, and we’re using it because it is brilliant at producing results. But there is a dark side of evolution that must be applied if you want to receive the benefit.

You have to try lots of things, and only keep what works.

Evolution’s core concept is descent with modification, coupled with selection. Meaning, the ones that don’t die off continue and then mix among each other.

The goal is to get as close to this with hiring as possible, within the bounds of human decency, the law, etc. So you have to try lots of people if possible. The more the better. And you need to have the tests be realistic. As realistic as possible.

And then you have to have your selection mechanism be as accurate (mapped to your actual position’s goals) as possible as well. People who don’t make it don’t make it.

Even if they went to the same school as you, or had the best sense of humor, or you just had a feeling about them.

If you designed your selection criteria correctly, and the process worked as it was supposed to, you have to take the people who produced the best output and discard the rest.

That’s Evolution-based Hiring.

Summary

  1. Too many are confusing negative and positive filtering when hiring.
  2. They often think that whiteboard coding and brain puzzles are a good way to do both simultaneously, but they’re not.
  3. You want to break out hiring into both phases—negative and positive—which are quite different.
  4. In the negative phase, use a strong checklist for filtering the rocks from the coffee, and little more.
  5. In the positive phase, use Evolution-based Hiring and put them into a pool of people doing real-world work.
  6. Don’t forget the essential selection phase which has to be unbiased and well-mapped to reality.
  7. Pick who does best at producing that output, and trust the system.

Notes

  1. The way I define a negative filter is that you’re pulling out the candidates that should not be in the process at all. If we’re mapping this to the physical world, this would be a liberal mess that allows a lot of different types of dirt through, but not large rocks. A positive filter would then be used on the second pass to allow only the smallest (finest) grains through.
  2. There are probably many industries where this model won’t work. I’m in IT and Security, so that’s where my experience is.
  3. There’s another meta-phase as well where you should constantly re-evaluate your work samples, selection criteria, etc., to work out the flaws and update according to changing requirements.

__

I do a weekly show called Unsupervised Learning, where I collect the most interesting stories in infosec, technology, and humans, and talk about why they matter. You can subscribe here.

Source: http://feeds.danielmiessler.com

Leave a Reply