The Google memo and real bias

Submitted by cathy n on 21 September, 2017 - 11:35 Author: Les Hearn
Google offices

James Damore, the recently (and perhaps unjustly) fired Google employee, criticised in an online memo(1) what he sees as the “left bias” of Google which has created a “politically correct monoculture” which “shames dissenters into silence.”

This left bias translates as “Compassion for the weak; disparities are due to injustices; humans are inherently cooperative; change is good (unstable); open; idealist.” A right bias would hold views such as “Respect for the strong/authority; disparities are natural and just; humans are inherently competitive; change is dangerous; closed; pragmatic.”

Like all stereotypes, these caricatures have some elements of truth and Damore is keen to distance himself from both but in reality he comes down on one side.

Put simply, Google’s stated policy is to encourage groups which are under-represented in their current workforce to apply for jobs or promotion. These include: women, around 50% of the population (31% overall in Google; 20% in technical posts; 48% in non-technical posts, doubtless lower-paid; 25% in leadership positions); black people (undefined but presumably African Americans), 13.3% of the US population (2% overall; 1% technical; 5% non-technical (lower-paid); 2% leadership); Hispanics, 17.6% of the population (4% overall; 3% technical; 5% non-technical; 2% leadership).(2)

To reiterate, there is a marked imbalance in the employment of black and Hispanic people in all areas of Google and of women in all but non-technical posts, relative to the US population. Damore chooses to focus his arguments on Google’s attempts to redress the balance for women. His arguments do not deal with ethnic or other minorities (except, curiously, conservatives) but his concluding suggestions do!(3)

He then produces a series of truisms and half-truths about male-female differences which he proposes as “possible non-bias causes of the gender gap in tech [i.e. software engineering].” He himself accepts that he is talking about averages and that there is a substantial overlap between the sexes so nothing can be deduced about any individual. He therefore sets a high bar if he expects these differences to account for a 20:80 split in tech jobs.(2)

Damore refers to biological differences that he claims are universal across cultures, highly heritable, linked to prenatal testosterone, and “exactly what we would predict from an evolutionary psychology perspective.” These include, he says, “openness directed towards feelings and aesthetics rather than ideas…a stronger interest in people rather than things…empathising versus systemizing.” This may direct them towards social or artistic areas (why then are there more male composers and painters?). It is not clear how this makes women less suitable (on average) to code software programs (or men to be more suitable to be managers).

There is also “extroversion expressed as gregariousness rather than assertiveness.” Damore says this results in women being less likely to ask for raises, speaking up …or leading. Google has tried to counter the reticence of women to put themselves forward for promotion. They sent an email to all engineers quoting studies showing that (1) girls don’t tend to raise their hands to answer maths problems, though they are more often right than boys; and (2) women don’t tend to volunteer ideas in business meetings, though their thoughts are often better than those of male colleagues: the email also reminded recipients that it was time to apply for promotion. Applications from women soared, and with greater success than for male engineers. It is not clear why Damore would object to this.(4)

Damore points to evidence that women show more neuroticism than men but his source (Wikipedia) points out that this concept is not well-defined. He also says that higher status is more likely to be a male goal, using the lack of women in top jobs as evidence (thus assuming what he set out to prove). Curiously, he sees the preponderance of men in dangerous jobs such as coal-mining, fire-fighting and garbage collection(!) as part of their drive for status.

What Damore does not mention is that cultural and individual sexism and misogyny discourage some (many?) girls and women from pursuing studies and careers in areas that have historically been denied to them or away from which they have been directed by peers, family or advisers. If girls and women were encouraged to see software development as something that was open to them, where they would be welcomed, but they still didn’t apply in equal numbers, then we could perhaps start looking for other explanations.
The question of welcoming is crucial. If male employees disrespect or sexually harass them, women may not wish to stay.(5) It is likely that, with encouragement at school and college, and with a non-discriminatory working environment, instead of 20:80, something approaching balance would be achieved: it might not be 50:50 — it might conceivably be 60:40 — who knows?

According to Wendy Hall, a computer science professor, there isn’t such an imbalance in several Asian countries, indicating cultural rather than biological influences on gender imbalance in US information technology companies.(6) Professor Hall refers to a decrease in women on computer science courses in UK universities from 25% in 1978 to 10% in 1987. In the US, women’s participation in historically male-dominated fields such as medicine, law, physical sciences rose from about 10% in 1970 to between 40 and 50% in 2010; computer science followed the same trajectory from about 12% in 1970 to about 37% in 1985 but thereafter declined to around 18% in 2010 (from blogger Faruk Ateş).7 We have to look for explanations for these changes, other than biological ones.

Ateş points out that many pioneers of computing and programming were women but that, from the late 1960s, women were actively discouraged from going into computing by professional organisations, ad campaigns, and by aptitude tests that favoured men. Stereotypes of computer programmers as awkward male nerds appeared in films in the 1980s. Ateş and Hall also refer to the marketing of video games on home computers, such as Sinclair and Amstrad, preferentially to boys in the 1980s, giving an impression that “technology is for boys, not girls.” Other scientists have also argued against Damore, including Angela Saini,(8) and Erin Giglio.(9)

A number of scientists have weighed in on Damore’s side, claiming that his views are in line with research findings on sex differences. Thus males tend to be “thing-oriented” and females to be “people-oriented” and women’s and men’s interests tend to match job preferences. Therefore, we should expect imbalances in gender ratios for jobs. (The fact that “women’s” jobs tend to be paid less is just a massive coincidence.) One study asks subjects about their preferences for these jobs: “car mechanic, costume designer, builder, dance teacher, carpenter, school teacher, electrical engineer, florist, inventor(!), and social worker.” No doctor, lawyer, bus-driver, para-medic, politician, accountant…

A closer look at many jobs show that the duties do not easily split into either “thing-oriented” or “people-oriented,” being more a mixture. Further, the proportions of men and women in some occupations have varied enormously over history: examples include physical labour occupations during wartime, or in other countries, and the medical profession from the 19th century, when women were banned, to now when a majority of entrants to medical school are women.

What is disturbing is that these scientists choose to investigate sex differences to explain observed gender imbalances in occupations when we already have a perfectly good explanation — the different experiences of boys and girls. Boy and girl children are treated differently by their mothers and significant others right from birth and, even in the supposedly egalitarian societies of the West, sex roles and expectations are reinforced throughout childhood and beyond. It may be that the “natural” ratio in software engineering is not 50:50 but we will never know since we don’t have a Planet B for comparison.

It is also disturbing that the research itself does not clearly show many statistically significant differences between the sexes that are relevant to suitability for software engineering. For every study showing some effect (such as higher general intelligence (“g”) scores) in men, there is another not showing this. Further, where there are well-documented differences, for example in visuo-spatial skills such as mental rotation, these can be reduced or removed with training.

To Damore’s credit, he suggests ways to make software engineering more woman-friendly (making programming more people-oriented and collaborative, fostering cooperation, making tech and leadership jobs less stressful, offering more part-time work, and, intriguingly, freeing men from their current inflexible gender role, allowing them to become more “feminine”).

However, Damore incorrectly sees Google’s encouragement of applications from historically under-represented groups as discriminatory, failing to recognise that, even if women would not necessarily take up tech jobs in equal proportion to men, there is no reason other than discrimination (not just at Google) for black and Hispanic people to be seriously under-represented in Google as a whole and especially in tech and leadership jobs.(3) In the absence of any better policies, his proposals would perpetuate the present unfair treatment of African Americans and other oppressed minorities.

There is bias in Google and in the job world in general but it’s against women and minorities, not against white men like James Damore.

3. Damore’s suggestions include “Stop restricting programs and classes to certain genders or races.” One programme cited is BOLD. Google states that “The BOLD Immersion program is open to all higher education students, and is committed to addressing diversity in our company and the technology industry. Students who are members of a group that is historically underrepresented in this field are encouraged to apply.” Another is CSSI. Google describes this as being for “graduating high school seniors with a passion for technology — especially students from historically underrepresented groups in the field.” It is odd that Damore interprets this as “restricting … to certain genders or races.” He also mentions Google’s Engineering Practicum intern programme which states that it is for “undergraduate students with a passion for technology — especially students from historically underrepresented groups including women, Native American, Black, Latino, Veteran and students with disabilities” [all emphases mine]. I suppose it is an occasion for rejoicing that Damore doesn’t oppose Google’s encouragement of veterans and people with disabilities to apply. This is in the context of only 2% of Google’s employees being Black (population average 13%) and 4% Hispanic (18% of population).
5. This survey reveals that 87% of female tech staff responding had experienced demeaning comments from colleagues and 60% had received unwanted sexual advances. Individual stories range from infuriating to sick-making:
8. Angela Saini, author of Inferior: How Science Got Women Wrong, deals with some of Damore’s points in the Guardian:
9. Erin Giglio, a PhD student in evolutionary biology and behaviour and a graduate in psychology and genetics (and blogger), cites peer-reviewed evidence contradicting Damore’s arguments: