Some Problems with Technical Interviews
One of the largest complaints I have heard about technical interviews is that they do not adequately measure a candidate’s skill. While I agree, I believe it is to a lesser extent than most complaints make them out to be.
After interviewing with 4 large technical companies, I have found that while the interviews may vary in format, difficulty, and time to decision, there are certain problems of the process that are quite the same across them all.
Disclaimer: these are representative of my experience and may not be representative of everyone’s experience. However, I do think they are valid for a significant number of individuals considering these same elements were present at four different companies. Furthermore, keep in mind that I received offers from all 4 of these companies; I am certainly not complaining just because I did not receive offers.
In general, the interview process attempts balance an accurate analysis of a candidate and resource cost (e.g. employee time taken to conduct the interview and provide feedback). While there will never be a “perfect” interview, we should always strive to do better. Here are some issues I think we can improve on.
- Lack of (Gender) Diversity — At each of these companies, I spoke only with male engineers. Many of these companies tout promoting women in tech, but it seemed that no women were given an opportunity to analyze me as a candidate. I feel this issue is further exasperated by the hosts and/or receptionists: they were all female. Unfortunately, this has the effect of subtly communicating that either a) females can not/should not be engineers and should instead be hosts/receptionists or b) female engineers are somehow less able to conduct interviews than male engineers. I know this is certainly not true, but I can easily see that mentality being encouraged as a result of this process. From this spreadsheet we can see that tech companies have an average of about 12% female engineers. The larger tech companies for which this article is about actually have slightly higher percentages (range of 15%-20%). There are certainly problems of women getting into to tech, but I specifically want to address this problem that occurs after they get there. I had a total of 20 interviews between these companies, and have had a single female engineer interview me. There are significantly less female engineers interviewing me than there (in principle) should be. There are many reasons this may be: female engineers may be asked less frequently, may reject more frequently, or some combination of the two. Nonetheless, I feel it is a problem when the already low proportion of women in technical roles is further compromised by lack of participation (for whatever reason) in evaluating potential employees (who may be their future co-workers!).
- Lack of Transparency — At the end of your interviews, you will receive either a congratulations or sorry. Of course you can bug a recruiter and sometimes get a small bit of information. However, in general it is very difficult if not impossible to get any real feedback. Without information to correct performance, we cannot be sure of what skills we need to improve on. It’s very difficult to introspectively determine what you may have done wrong in an interview. As seen in this Quora post, Gayle Laakmann McDowell (author of Cracking the Coding Interview) believes there is no relation between how you think you did after an interview and how you actually did. That doesn’t leave us with much to work with. With the constant cry of “lack of tech talent”, I would think that there would be an easier process from which we can improve on the skills in which we are lacking.
- Lack of Data/Lack of Training or Experience — For these companies, you will probably have a round of ~5 interviews in total. In most cases, bombing a single interview can destroy your chances. How do you bomb an interview? Is it always “your” fault? I will concede that most of the time it is. However, interview training processes do not seem sufficient. Some interviewers conduct interviews quite well; they clearly present their question(s), ask clarifying questions about your solution, and only interrupt your process if you are moving too far in the wrong direction. However, there are a significant number who make mistakes such as not clearly stating the question, not asking clarifying questions (and therefore making assumptions about your solution), and interrupting the process too often. Sure, you can say that it is up to the interviewee to make sure that his solution is clear, but that is simply not 100% true. In any communication setting, it is up to the receiver to voice any questions or concerns. A person could tell you wonderful things all day and never know you don’t understand until you let them know that. There are worse problems, though. Some interviewers will even talk on their phone throughout an interview or simply leave the room! While one bad interview can ruin your chances, it could be the lack of experience on the interviewer’s part that causes one bad interview. Having more interviews (i.e. data) or improving the interviewer training process could resolve such issues.
In the end, technical interviews have to prefer Type II errors (false negative) to Type I errors (false positive). They would much rather an applicant be rejected due to a flaw in the process rather than have an applicant without the necessary skills slip through the cracks. However, training processes could certainly be improved, a more transparent interview process could be developed, and most importantly, more female engineers could start conducting interviews as well.