Dissecting the Google Employee’s Anti-Diversity Manifesto, Piece-by-Piece

Part 2: Facts claimed, facts maimed.

Following on from the first post on the Google employee’s anti-diversity manifesto, we start to look at the data and conjecture within the manifesto and find a very weak empirical position.


Possible non-bias causes of the gender gap in tech [3]
At Google, we’re regularly told that implicit (unconscious) and explicit biases are holding women back in tech and leadership. Of course, men and women experience bias, tech, and the workplace differently and we should be cognizant of this, but it’s far from the whole story.
On average, men and women biologically differ in many ways. These differences aren’t just socially constructed because:
- They’re universal across human cultures
- They often have clear biological causes and links to prenatal testosterone
- Biological males that were castrated at birth and raised as females often still identify and act like males
- The underlying traits are highly heritable
- They’re exactly what we would predict from an evolutionary psychology perspective
Note, I’m not saying that all men differ from women in the following ways or that these differences are “just.” I’m simply stating that the distribution of preferences and abilities of men and women differ in part due to biological causes and that these differences may explain why we don’t see equal representation of women in tech and leadership. Many of these differences are small and there’s significant overlap between men and women, so you can’t say anything about an individual given these population level distributions.

Now, there are a plethora of factual inaccuracies here anyway.

- They’re universal across human cultures

A study was conducted a number of years ago to consider the component of cultural effects across a world standardised performance scoring system. The PISA math level 5 is an OECD standardised test of mathematical performance at age 15. The data across all countries using the PISA international student assessment showed significant enough cultural variation which the authors of the study considered relative to genetic variation.

This graph isn’t easy to read. However, it compares the number of boys and girls above level 5. Note, there are less girls that achieve level 5 than boys (25% and 30% respectively) but crucially, the variation due to culture is influential for boys to a significant enough degree for those boys to yield a benefit in the ratio and the countries within the sample include the countries which do not provide girls with the same educational opportunities as boys.

For the data the graph was pulled from, check out [1] L. Guiso, F. Monte, P. Sapienza, L. Zingales, “Culture, Gender, and Math” Science 320, 1164 (2008). referenced.

The paper showed that gender-neutral environments led to the disappearance of the gap. The key to note is that the data assumes the lower bound for the genetic element and the higher bound for the maximum variability of the gap. The greater the variability, the greater the cultural influence.

This was indeed, pretty much consistent even in non-gender neutral environments when reaching middle and high school ages. The variability is consistent across culture where non-gender neutrality exists.

This starts to cast doubt on the veracity of the manifesto’s point and also the following statements.

- The underlying traits are highly heritable 
- They’re exactly what we would predict from an evolutionary psychology perspective

Another problem with the above perspective is that it fails to account for any epigenetic effects as well as the inference that the fitness function is always trying to achieve the same utility. In essence, they are attributing causal results to insignificant differences, which have a covariate in the very mechanism of assessment. Not every person in Google has to fit the profile. Indeed, innovation itself thrives on diversity of thought and perspective, which is both claimed by the author, but equally, also rejected by any inference that women are not suited to tech.

There are several further pieces of information in that alone which refutes cultural universality. Yet, they are not examined and worse, we’ve seen this before!

“As with the scientific racism of the 1980’s and before, such authors as Rushton et al, failed to account for confounding factors. Not only was the conclusion completely false, it was also completely bad science.”

Being Facetious

The manifesto author neglected to demonstrate which of the multitude of engendered human traits are different enough to exclude female Google employees. If we look at absolute physical strength, there’s a factual argument to be made.

Source: Layk et al, hand grip strength

Or how about their height?

However, that is likely not what he’s getting at. So let’s look at IQ (whatever that means — note, I’ll leave aside that it isn’t even a measure of intelligence and also leave aside Gardner’s multiple intellects until the relevant point in the manifesto).

This graph reappears time and time again in each study, sometimes with a skew either side, yet it shows a consistent point. That women’s standard deviation is different to men’s. Whether we like it or not, that is a fact.

Yet, there’s a key thing to note here, why does it matter? Google and other companies are not aiming to recruit nor develop average developers or engineers. They’re looking for the very top end of the spectrum and crucially, deliver teams that work well together. Some may see this as naturally targeting males, but this has two fundamental flaws.

  1. Performance — teams requires both capability, the skill to do the job and the capacity to carry it out. This latter point is why a team is greater than the sum of individuals, even if the team consists of slightly “lower skill” developers. So even if the manifesto author was correct, meritocratic assessment of developers would still have the same performance from men of the same skill as women.
  2. Communication/Collaboration — team collaboration is essential to effective communication, which itself is a precursor to effective collaboration.

As you go further to the right of this graph, you get a higher proportion of men versus women, as a percentage of the population, but an interesting thing happens at the very extreme right hand side of the distribution. They level out again! It’s an asymptote after all.

Yet, despite all that, even here, gaps in the attainment levels have been debunked, since examination of females ages 15 and below showed exactly the same aptitude and indeed, higher aptitude for mathematics and numeracy than boys. Although this was not the case 20 years ago. This temporal change implies that we cannot assume that the education system is not highly influential in the process of generating Google style workers and pretty much puts pay to the acceptance of any alternate hypothesis attributing the effect exclusively to evolutionary processes, whether genetic or psychological and in the latter case, any inference that such evolutionary psychology would somehow stop now.

Collaboration: Google Drops Exclusive Coding Skill Requirements

A few years ago, Google stopped relying solely on technical excellence in recruitment. This was due to the number of great coders who could not function as a team and as a result, impacted team performance.

In order to get past that, Google began seeking those with greater “soft skill” requirements over pure engineering skill. The technical requirements were still as high, though the preference had shifted to those who could collaborate. Here, there is some evidence that refutes the author’s claim, but twisted, also supports it.

Using the Myers-Briggs scale, there is some empirical evidence to support inherent over representations in certain types of each personality trait. Yet, as we know, all of these personalities develop throughout our early life, influenced by the interaction with peers, schools, neighbours, friends, clubs and family. Personality research does not provide evidence that it is in any way, a genetic trait.

Note, I’m not saying that all men differ from women in the following ways or that these differences are “just. I’m simply stating that the distribution of preferences and abilities of men and women differ in part due to biological causes and that these differences may explain why we don’t see equal representation of women in tech and leadership. Many of these differences are small and there’s significant overlap between men and women, so you can’t say anything about an individual given these population level distributions.

There are a number of issue with his claims. They made an unobjective judgement and also not commuted it. As a male, the author can only claim definite experiences of being male and his male at that (we are not tackling racial segments here, but that would be a pertinent observation).

Furthemore, we’ve already questioned the robustness of the biological argument (i.e. you cannot say there is one). What I am surprised at, is that this is then attributed to genetic factors.

No. Tech leadership limitations for women happen for a number of reasons. Let’s suppose that we work with the assumption that Google doesn’t have an issue with diversity. That means the 69:31 split is justified. That means on chance alone, with the same steady state population of women in tech, women can expect to make the above ratio into upper management.

Yet, we know that is not represented in the data. Not least because women, who were qualified enough to enter the profession and indeed, made it through into Google on merit, are 45% more likely to leave than women in non-STEM careers, especially early in their careers, with 51% leaving their skills behind altogether (personally, I think that’s a tragedy). Women are simply are not staying long enough to reach management.

Hence, the chance of promotion and development is lower, simply on that. i.e. temporally the probability of women achieving senior managerial status declines over time. Again, nothing to do with genetic or biological factors.


This is part of a series of articles critiquing the Google employee’s anti-diversity manifesto.

Show your support

Clapping shows how much you appreciated Ethar Alali’s story.