Using Algorithmic Puzzles in the hiring loop
Algorithmic puzzles are brainteasers intended to test the algorithmic thinking of the candidate. Teasers asked at top companies become widely known and assembled into internet-curated lists. The puzzles usually have several solutions, with one being the optimal one. The candidate's goal is to quickly pick correct algorithms and data structures to get to it. To prepare for the interview, people often memorize the answers to a large set of such puzzles. The setting is sync and is either face-to-face or remote using online coding platforms.
Several things influenced why algorithmic puzzles became a software engineering interview filter. They are handy for teaching algorithms to the test. They test puzzle-solving skills, which might allude to inductive learning ability. And due to the bandwagon effect: many software giants started using them and set up a trend.
Algorithmic Puzzles: are they unbiased?
This interview type shares many biases with other face-to-face types.
The interview setting makes it easy for interviewers to give in to their natural biases. They have to defy their first impressions and actively go for the skill assessment. They have to be well-trained to recognize a valid answer even when it is not an expected one. They have to decide on the solution itself and not on whether they like a candidate personally. In such circumstances, our brains use all the possible help from the biases to fill in the gaps.
The puzzles themselves add some implied biases because of their limited variety. Their original goal was to detect the candidates with algorithmic thinking. With added time pressure, they start favoring the candidates who memorized the answers.
Algorithmic Puzzles: are they low-stress for your candidates?
They share the same stress factors with other face-to-face interviews, like "whiteboarding" or "pairing sessions."
The candidate has to solve the puzzle in a limited time while being judged by someone they see for the first time. That stress hinders the candidate's cognitive skills during the interview. To address that and improve their chances, the candidates learn the correct answers. Yet, many of them fail not because they can't solve the puzzle but because they panic.
Algorithmic Puzzles: are they real work?
And they are often the exact opposite.
The job of the software engineer is to solve business requirements using code. The code needs to be maintainable and easy to comprehend and extend by peers. When business requirements change, it should be painless to adapt. In contrast, algorithmic puzzles make candidates do what no employee will do. For example, to invert a binary tree or work out the square root without the standard library. They need to show the apt usage of cycles, binary search, recursion, and to get to the optimal solution on the spot. In real life, the expectation is the exact opposite: the employee must use a standard library.
Algorithmic Puzzles: are they a good predictor of future performance?
Unless solving puzzles in one attempt under pressure is the company's core business.
When hired, the candidate will perform completely different tasks from interview puzzles. They will build your product and deliver code that solves specific business requirements. This interview type doesn't measure those skills and does not predict the performance.
Algorithmic Puzzles: how to improve your hiring loop
Stop using them as a filter in your hiring loop.
Their strengths lie in teaching algorithms and training algorithmic thinking. Adding time constraints and the stress of judgment does not make them an accurate filter.
Due to inertia, algorithmic interviews are still most prevalent in the industry. A variety of books and courses exist with the sole purpose of tutoring the candidates to pass them. Using them as a hiring filter strongly favors interview-passing ability over problem-solving ability. It is a wrong metric to pick a hire, and companies that pay attention are giving up on algorithmic puzzles. Google, a company that championed the brainteasers in the first place, is phasing them out. Their data indicated that brainteasers are a waste of time, and they adapted. Buffer stopped asking technical questions in technical interviews altogether. They understood that these are not predictors and instead switched to behavioral interviews. Did your company get the memo?
AutoIterative Job Interviews to the rescue!
AutoIterative Job Interviews focus on predicting the candidate's ability to do the work. The best way to do it is to let the candidate deliver something to production.
AutoIterative Platform provides each of your candidates with a dedicated work environment. It contains a real-life challenge, a code repository, and a continuous delivery system. Instead of solving puzzles, they ship their code to the production as they would do it in real life. They can iterate on their solution as long as they need until they decide to finish. On completion, the platform will give you a performance report to aid the next steps of your hiring loop.Start hiring now
Want a second opinion?
Here is what other people say about Algorithmic Puzzles:
- Buffer ditched technical interviews for technical positions.
- Google did that too.
- Adaface explains why puzzles are not as effective for an interview as you might think.
- Candidate's perspective: I will not attend your algorithmic test interview.
- Dan Luu explains shortcomings of algorithmic interviews.