I see many people in my timeline today reacting with outrage to the Bloomberg interview with Snowflake Inc, CEO Frank Slootman.
I read the article and watched the video and as a woman in tech for 25 years, his statements can definitely feel like a sucker punch to the gut. It would be easy for me to be outraged, upset, even ‘hysterical’ as he so put it. But the reality is, I’ve heard words like his from powerful people in tech, many many times during my career.
For those of you who have not read the article, TL:DR; The CEO of Snowflake, whilst ‘sympathetic to diversity’, believes that hiring people on merit should take precedence over hiring from under represented groups, so as to avoid ‘lowering the bar’. He believes we need a more ‘moderated’ approach to diversity and ‘there’s really no room for the hysteria and outrage’. Full article here.
Frank Slootman mentions that other CEOs he knows have the same views but find it hard to be that way publicly. For me this is the most concerning aspect of the interview. Maybe Frank is correct, maybe we do need a more moderated approach to diversity?
First let’s take a look at the notion of merit as a concept. The definition of merit is ‘the quality of being particularly good or worthy, especially so as to deserve praise or reward’. In this case, the reward is a thriving career in tech. But who decides who’s ‘good’ or ‘worthy’?
In 2016 US Researchers ran an interesting piece of research on Github, the source code repository service and their then nearly 1.4 million users. Interestingly it found that a higher percentage of pull requests made by women (versus men) were accepted, meaning the code was viewed to have been of a higher standard.
One might draw the conclusion from this that women are therefore more likely to to be ‘particularly good or worthy, especially so as to deserve praise or reward’ when it comes to writing code. So why are tech teams across the industry not filled with women? Because the results from the Github research were only the case when the gender of the code submitter was not known. When gender was known, womens pull requests were less likely (than mens) to be accepted. We can therefore conclude that ‘merit’ is only in the eyes of the beholder and bias plays a significant role in how we measure it.
The interview process for most tech roles at tech companies goes something like this: Technical interview and/or coding test, then ‘culture fit’ interview.
The technical tests tend to be drawn up by engineers or tech leads already in the company (who tend to be male). The technical test can be anything from a whiteboard session, take home coding assignment or pair programming onsite task. Usually it is the employer who decides how the candidate will take the test and how the results will be measured. In many companies the technical test bares little resemblance to the work environment at the company, how they would be expected to work within a team and the resources that they would have available to them to solve a problem. All of this leaves significant room for bias, particularly in the case of a whiteboard session, where a person’s demeanor, communication style and appearance can impact the results of the assessment.
Culture fit interviews are then a minefield of bias as a person is asked to make a subjective assessment of a candidate based on their perceived understanding of the company culture. If your company’s environment consists of pool tables, nerf guns and beer fridges, it will be all too easy for your hiring managers to assume this candidate ‘will not fit in here’.
It gets worse as women engineers progress through their career. Women hold less than 20% of technical roles in technology companies. Research from the National Centre for Women in Tech (https://ncwit.org/) shows that 41% of women leave tech careers within 10 years of starting their career, and 56% of women leave between 10 and 20 years into their career. Most of these women go on to have full time careers in other industries so we cannot put this down to ‘child-rearing’. At the top, less than 10% of CTOs or CIOs are women. We don’t see this ‘leaky-pipe’ effect even in other STEM disciplines, the tech industry is literally driving women out, but why?
Humans pattern match. It’s in our nature, it’s how we can make decisions quickly based on past experience or the information we have available to us. Unfortunately our ‘pattern match’ for engineers tends to be the male, loner, hoodie wearing introvert, white, straight, tech-bro. This is not an accident. I recently learned about the Programmer Aptitude Test (PAT) from Emma Jones (CEO at Project F). The PAT test was widely used industry wide as an indicator of aptitude for coding for many years, but the data behind the test was derived almost exclusively from men. Read more about this and other reasons for our engineer bias here.
The good news is that we can break our ‘pattern matching’ habit when it comes to identifying good tech candidates. To do this we need to acknowledge our bias and make conscious deliberate steps to change our behaviour. Like you would taking steps to improve your fitness or give up smoking, we can make the change once we acknowledge it needs to be done. But in order to do that we need to throw out this notion of merit. Merit is inherently biased by the people who decide it. More than that, once we decide we’re selecting on merit only, we lull ourselves into a false sense of security that we’re hiring the best people. The truth is, the best people might not be the ones that think like us, behave like us, look like us.
Back to our Snowflake CEO Frank Slootman. I’m disappointed that none of his CEO friends (that we’re aware of) actually privately called him out on his assumptions. Or maybe they are all in agreement with him? We need to be able to have this conversation openly so that behaviours start to change. Do we need a more moderated approach? I know many male leaders in the tech industry, who really understand this problem and want to help and do. But many many more who want change but have no idea how or are unwilling to make the behaviour changes required. Until this changes and everyone is onboard, to hell with moderation, we need to see more outrage and hysteria from all leaders in the industry if we’re going to see any level of change in the near future.
In my last role as CTO, we overhauled our recruitment processes to remove bias, changing the way we worded Job Descriptions, how we did technical assessments and changed our focus to culture add (from culture fit). In that time our balance of women to men went from 17% to 34%, and our tech leadership had 50/50. It wasn’t easy. We had lots of push-back and some roles took significantly longer to fill, but no one denies the strength of the team that we built and the positive impact that is now having on the whole business.
Imagine if the tech industry was full of people who thought differently, had different backgrounds AND had higher quality code leading to better outcomes for customers and higher revenue for the company? The Github research (and many others) show us that we can have both if we remove bias from our decision making.
Hiring more women in tech will RAISE THE BAR!