The Google memo is not ‘science’ and here is why.
Okay… let me rephrase that in a slightly less clickbaity manner: there are two fundamental problems with James Damore’s memo. One is a mistake in its reasoning that disqualifies it from being called science. The other is that the preconceptions of the author clearly clouded his judgement and that resulted in the aforementioned logical fallacy.
There are many more problems, but I’m just going to concentrate on these two to keep some focus and because I feel these are the most important points on which it should and can be objectively criticized. I’m not intending to take down every part of his argument, my sole purpose is to point out the part where he most blatantly ditches scientific rigor in favor of expressing an opinion.
Is it science?
The memo makes the same basic mistake that most pseudo-scientific mumbo jumbo. It sounds very scientific, it does cite seemingly valid sources, but at the most crucial point — where it makes an extraordinary claim — it fails to provide any evidence. In this case the claim is that the lower representation of women in IT is partly due to statistical differences in traits due to biological reasons.
His reasoning is based on a list of traits that he claims to have different averages among women and men, and that the personality of the average male is more likely to seek a career in IT. However for this reasoning to be scientifically accurate he should have included all relevant traits in the list with evidence that shows that these traits are the most important factors indeed. Short of that his claim does not meet the standards scientific rigor.
He also uses cherrypicked results that he thinks imply that women are less likely to pursue a career in IT successfully, and ignores data that would balance these claims out. What’s even worse is that some of the articles he cited are actually outliers and are in contradiction with the rest of the literature.
That is still not the end of the road though. In scientific terms all of the above — even when done correctly — only constitutes a hypothesis and not a fact. For it to become a fact studies would have to be conducted and controlled for external effects. None of that has been done.
Traits of a great developer
I was struggling if I should discuss his reasoning in more depth since the lack of evidence is enough to show that his claim is not some kind of scientific truth. However he also made two claims about software development that are fundamentally contrary to what most agile coaches would argue for. Since these are common stereotypes I think it is important to point them out, and make sure that people understand that even the scientific facts that were cited, were interpreted incorrectly.
Our existing knowledge of what makes a good software developer is mainly coming from the agile software development community. Books like Clean Coder explain these traits in much detail. For me it seems that peer reviewed validation of agile software development is not complete yet, but there are surveys that confirm part of the claims. On the other hand according to the regularly performed Standish Group Chaos Reports teams following agile principles are decisively more successful.
I wouldn’t claim that the science is in on agile, but it certainly is the most successful methodology used so far, and the existing research is promising. I’d be happy if there were scientific studies that unquestionably proved these claims, but short of that the success we see in the industry and the existing research in academia is good enough for practical purposes.
The first of his incorrect claims is that software development is not inherently people-oriented. One of the core values of agile software development is human interaction. Developers need to have strong communication skills to gather requirements, to communicate with their teammates during daily stand-up meetings, and even more crucially they need to be able to write code that conveys intent. All of these require someone who is a people person and great at communication. If I had to name the most important trait I look for in candidates, it is the ability to communicate both verbally, and in code.
The second claim that I found particularly misguided is that competitiveness is more important than cooperativeness. Software development is a team work where team members need to cooperate not just among themselves, but with other teams and management as well. In a functioning team there isn’t much space for ego and for competitiveness between members of the team. Too much ego — putting one’s own interests in front of company interest — is one of the biggest enemies of a company’s success. In fact the management guru Patrick Lencioni points to ego as one of the dysfunctions in his book titled “The Five Dysfunctions of a Team”.
Ironically even though Mr. Damore has cherrypicked the results that he believed fit his agenda, he accidentally included two results that actually — when interpreted with enough knowledge about what is important in software development — imply just the opposite of what he intended to prove.
I find it incredibly misleading to compare the average person in a group with the average person in another group, since any such comparison can easily be interpreted as a slur against one of the two groups. But if I were in a team where two of the team members had traits that James Damore describes as the “average women” and the “average man”, I would argue that they are both fit for the job and for different reasons. They might not play the same roles despite having the same position in the company, but they would both be critical to the success of the team.
To be honest I’m kind of sorry that we need to have this discussion instead of a discussion about constructive criticism of diversity programs. I’m pretty sure there are ways we could improve those. We are still pretty far from making software companies a welcoming environment for women. The really sad part is that valid criticism of the current diversity programs would have been perfectly possible without even mentioning the whole topic of biological differences.