"Source?" An academic examination of the James Damore "Google Echo Chamber" memo

Photo by rawpixel.com on Unsplash

This paper will only focus on the relative weakness or strength of the Google memo. It will not discuss the relative morality of the arguments themselves.  

*A note on formatting: direct quotes from Damore are italicized, and hyperlinks are indicated by bold type or directly linked. Direct quotes from sources are also italicized.  

As I understand it, Google Memo (https://assets.documentcloud.org/documents/3914586/Googles-Ideological-Echo-Chamber.pdf) appears to have two core arguments.

The first is that a) dramatic or noteworthy unequal levels in gender for key positions, both in level (management and above) and in type (STEM, tech, dangerous work, high status work), could be explained by 1) biological 2) preferences 3) in occupational choices.

I will address the other core argument at another time. 

I don’t feel that I have sufficient knowledge of Google’s internal policies to fairly address that argument. In short, it appears to be that Google is suppressing the dialogue around the first core argument, and that because of this, Google is vulnerable in several ways. This second argument is far more subjective and morally based, and as such I feel it is beyond both my scope and ability to address fully at this time.  

A note about reproductions of the Google memo.  The Google memo is being reproduced without the original hyperlinks, which is both understandable and, in my opinion, a poor journalistic choice. Readers have been taught by experience that hyperlinked text generally indicates source material that is supportive of the text idea, related directly to the text idea, and/or credible. A majority of the Google memo’s links do not fulfill these expectations.  Reproducing them actually weakens the individual arguments the author makes and by extension, the larger arguments.  That is, if the reader clicks on them, reads them, and comes to a conclusion using a solid, accepted metric about their relative credibility. Many readers may not do so, and may simply assume that the mere presence of a hyperlink is conclusive evidence that strong, credible support exists for the linked assertion.  It is a journalistic risk to both omit and include the original hyperlinks. I can understand why some outlets chose to exclude them, since a strong argument can be made that their mere existence lends the document more credibility than it objectively deserves. However, without those links, the Google Memo reads like a blog post, opinion piece, or very long comment from a comment section, and appears to be entirely without any attempt to support its core arguments. This is also misleading. I feel that the greater risk is to publish the paper without hyperlinks, as the reader has no way of knowing which, if any, arguments the author attempted to support and in what ways.  

I clicked on each hyperlink and reviewed the credibility of the source, its main arguments, and if/how it supported the author’s argument. I also did some data analysis on the types of sources the author used  I feel that the Damore defends each part of that first core argument weakly or incompletely at best, and that because of this, a careful and objective reader should be very skeptical of his overall point as well.  

The argument has three main components, as I see them:  

· That existing gender diversity in certain roles in dictated by occupational choices primarily 

· That occupational choices are dictated by preferences in varying areas, but most specifically around status seeking and desire for work-life balance 

· That these status-seeking preferences are dictated by biology, most specifically by testosterone    

Working from the most basic or first building block of his argument, that preferences are dictated by biology, I will address the relative strengths or weakness of this part of the paper and move on to the other parts.  

First, are preferences (in this case, the paper appears to equate occupational choices with status, which could also be argued, but I’ll leave that alone) dictated solely or even primarily by biology?  Before addressing this first assertion, we must look at the opening paragraph of the Google Memo (after the introductory paragraphs and items).  

Damore writes a long paragraph with several completely un-cited or unsupported assertions:  

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 

Damore may have found citations that 100% back up these above assertions. However, it is accepted academic practice to establish one’s credibility both in relation to the general canon of thought on a topic and in specific to one’s theory by using citations to both interrogate and develop the core ideas of that theory.  

In this case, Damore does use citations relatively extensively in his paper. To be fair, he does not claim to be an academic or have a special expertise on the topic. But if he is going to use citations to prove his points, I will judge him according to that metric. By this metric, he makes several arguments that are integral to his overall argument that he does not either interrogate or support with any citations whatsoever. He presents them as well-established facts. They are not.  

Moving on to the components of the argument that he does support, let us examine individually and then as a group, the credibility of those supporting sources.   In my college education, and as far as I understand it in the wider scientific community, sources must be primary, peer reviewed, and unbiased (as much as possible) to be considered credible. If a source fails on any of these levels, it needs to act as a secondary source, and be either interrogated or developed additionally by a primary source.  

Sources such as personal blogs or webpages, Wikipedia, fictional works, or studies that are past a certain age (limits vary, but when dealing with hard science, generally the limit would be 10 years at the outside) are not considered primary, credible sources.  

The first source that is used to develop or prove his argument: Women on average have more […] is Wikipedia. The title of this Wikipedia article is “Sex differences in psychology.” One may note that this article is not a peer reviewed scientific study, or, more importantly, does not address directly or in the main biological differences that lead to preferences in occupational choices. In fact, the first sentence in this source says Sex differences in psychology or gender differences are differences in the mental functions and behaviors of the sexes, and are due to a complex interplay of biological, developmental, and cultural factors.(emphasis theirs) Secondly, psychology is not a holistic view of occupational paths dictated by gender differentiated preferences. 

This article is not only not a credible primary source, it does not actually support the assertion that is being made in Damore’s memo. 

In discussion with another party on the topic of the support or lack thereof in the Google Memo, it was raised that due to the nature of hyperlinks, perhaps the author’s intention is for the reader to click on links within links, i.e., the “biological” link that is contained in the first sentence of the author-linked source.  I am certainly willing to allow that intention. However, it is a very weak argument structure, as it allows for an unknown amount of clicking through, since we the reader are not given explicit instructions on how many or what type of links to click. One may note that within many primary sources cited in this article, multiple hyperlinks to other articles, themselves of varying credibility, appear.  Therefore, for the purposes of showing why the Google Memo is so poorly supported, we shall assume that the cited source, in situ and exclusively, was intended to be used as supporting evidence.  

Moving on to the second cited source, which is hyperlinked in this sentence:  Openness directed towards feelings and aesthetics rather than ideas. Women generally also have a stronger interest in people rather than things.  The hyperlink is on the words “people rather than things.”  This source article is sound. It is a scientific study that is peer reviewed and was published 7 years ago.  

There are several issues with the data in this study, however. The first issue is that the data is primarily drawn from self-reported results on standardized tests. Research on gender differences in personality and interests typically relies on data from standardized tests. Because such tests use self-report scales, their scores may be influenced by social stereotypes, social desirability response sets, and self-construal processes (See Feingold, 1994; Guimond, 2008). The author of the paper allows that standardized testing, by design, can be biased in a clear statement in his remarks on the results. The author of this source also spends a significant amount of time explaining social-environmental theories of personality differences in the genders, although this is a position that the author does not agree with, citing other sound sources in his remarks.  

The data is interesting. The author of the paper finds significantly higher (sometimes as high as a factor of 125%, as in the difference between .15 and -1.40) differences between the genders than other studies do, and his work is a meta-analysis of other existing studies, as well as an independent study with a much lower number than the second lowest amount (by 8000, or almost 90% lower than the next highest) of participants. His study also measures a kind of amalgam of all “Big Five” traits rather than each trait individually, for which the differences between genders range from “small” to “moderate” according to the protocols established by the study.  In short, the method and the data are not bulletproof.  

The next citation, (also interpreted as empathizing vs. systemizing) is Wikipedia. There is a warning that the source article has multiple issues, one of which is that it needs more support. As we have noted, Wikipedia is expressly forbidden for use as a primary or even secondary source in college papers. It is not considered credible in the least.  

The third citation, in this heading “Neuroticism” is also Wikipedia. The article is a definition of neuroticism. It does not support the argument that women are more neurotic than men, and that, by extension, this neuroticism is responsible for avoiding high status or high stress jobs, or choosing low status jobs.  In fact, in the study cited in “people rather than things” link, there is a noted difference in neuroticism in women. It is .34 (or 1/3 of 1 percent) higher in women than in men. According to the protocol of the study cited by the Damore to prove his other point about women being more aligned with people rather than things, a difference of .34 is considered “small.”  

It is interesting to note that the next sentence, which makes two assertions, both of which could have been proved rather easily, has no citations linked. 

It says “This may contribute to the higher levels of anxiety women report on Googlegeist and to the lower number of women in high stress jobs.” As an employee of Google, and I assume, someone who has enough access to “Googlgeist” to see a significant difference between women’s and men’s reported anxiety, Damore could have either provided an informal result such as “of the last ten complaints, 6 were by female employees, and showed the following biologically driven concerns…” or data analysis of a larger sample of complaints. Without supporting examples or data, this remark is merely anecdotal, and should not be considered “fact”.  

The next cited source is in this sentence Note that contrary to what a social constructionist would argue, research suggests that "greater nation-level gender equality leads to psychological dissimilarity in men’s and women’s personality traits."  The study cited is a meta-analysis of “Big Five” personality traits across 55 nations. Again, I would like to note the inherent issues with both self-reported data, and with standardized tests. Leaving aside that, it is important to note that the only two credible primary sources that have been cited in this document so far are based on the same metric: The Big Five personality traits standardized tests results.  

I would like to argue that if gender differences are consistent across both time and cultures, this should be reflected in the foundation, type, and method of study that the author cites. This study is from 2008, and is now 9.5 years old, reaching the edge of what would be considered a reasonably recent scientific piece. However, overall it is credible, and it does nominally support the author’s assertion.  

The most important thing I would like to raise is that again, all these studies establish is that gender-based differences in personality exist, not that they are exclusively biological in origin, or that they in turn affect occupational choices, or that gender differences in occupations is mostly due to preferential choice rather than a complex combination of factors.  

To be fair, the Google Memo states “may be due”. However, the length of the arguments and way in which they are stated seems to take many assertions ask point of fact, thereby lessening the qualifying statement the author opens with. It appears to me that this author believes there is strong scientific evidence for his main argument.  

This is a structural issue with the paper as a whole. Damore repeatedly extrapolates on what could fairly be considered strong source material and does not connect the dots for the reader with primary sources. Instead he connects it with what appears to be his opinion, which is not supported in any way.  

The next citation is in this sentence: We need to stop assuming that gender gaps imply sexism. This source is one I would consider “moderately credible”. It is well defended, structured, supported and researched piece in an online magazine. It does consist of an argument that gender differences, which are inherent, lead to different choices in careers. 

If I were the author of the Google memo I might have cited this statement in a block text format from the article:  

In other words, in prosperous and egalitarian countries, people are free to pursue their respective career interests. Since men and women are innately different, they pursue different vocations based on dissimilar interests. Therefore, equal representation is an inaccurate measure of gender egalitarianism, simply because in egalitarian countries, gender gaps are the most robust. 

This article appears to support the statement that “we” need to stop assuming gender gaps [in certain jobs] imply sexism.  That’s fair. The simple existence of a gap in gender in a job type or level is not, in and of itself, conclusive proof of sexism. I agree and concede this point, based on the author’s defense.  However, to be fair in another direction, the Google memo author’s source here also makes a leap or jump in logic, that “since men and women are innately different, they pursue different vocations based on dissimilar interests.”  That can be true, and it can also be true that the women who are outside the gender bell curve, who do pursue management and above level jobs, and/or jobs in STEM or tech, face sexism. 

One does not mean the other is not true. It is not a mutually exclusive proposition. In the most charitable interpretation of the author’s first main point, one could argue that he is simply putting forth an alternative explanation for the lack of women in higher management and/or STEM or tech jobs, one that is not based solely on sexism.  There is certainly room to consider this position, leaving aside the social, cultural, and moral implications of such an argument. However, the author’s relative weakness in supporting or developing this argument makes it less credible on its face, and does his argument no favors.  

Moving on to the next heading “Men’s Higher Drive for Status”, let’s look at components of this argument. Here’s the argument:  We always ask why we don't see women in top leadership positions, but we never ask why we see so many men in these jobs. These positions often require long, stressful hours that may not be worth it if you want a balanced and fulfilling life. Status is the primary metric that men are judged on4, pushing many men into these higher paying, less satisfying jobs for the status that they entail. Note, the same forces that lead men into high pay/high stress jobs in tech and leadership cause men to take undesirable and dangerous jobs like coal mining, garbage collection, and firefighting, and suffer 93% of work-related deaths. 

The last 5 words are hyperlinked to a source. No other citations are given in this argument.  The key components of this argument appear to be:  · Leadership is a high-status job, and one that is demanding/stressful. · Status is a primary metric that men are judged on (in a mating context)  · The same forces that lead men into certain kinds of jobs also lead them to undesirable or dangerous jobs · These dangerous jobs cause men to suffer 93% of work-related deaths.  The main issue with this is what I would call “muddy terms”. Muddy terms is where words are used in a vague, rather than precise, way. “Leadership” “status” “metric” “judged on”, “higher paying” and so on.  

This is actually a golden opportunity for the Damore to link each and every one of these terms with studies that back his assertions. The amount of men in leadership, a study that shows what status is (he later does tie testosterone to status seeking behavior, and mate-seeking behavior in women to higher status men, but not in this paragraph) and that men do indeed seek it, actual facts and figures about pay rates and satisfaction levels, and what “forces” Damore refers to that “lead men into […] undesirable and dangerous jobs”.  

The statistic that 93% of work related deaths are men proves only that men are over-represented in dangerous jobs, not the reasons men were drawn to these jobs in the first place. It certainly doesn’t support an assertion that status seeking behavior, enforced by heterosexual mating rituals, resulting from testosterone “forces” men into dangerous, and even fatal work. Again, Damore connects dots with what appears to be his opinion.  

This argument could actually be a much stronger core of Damore’s point, but it is a missed opportunity in my opinion.  The 93% citation itself is perhaps in error. It goes to a Wikipedia article that does not contain this statistic in any direct form. I did a little research and it appears that this article supports this claim. That article actually cites an impeccable primary source, the Bureau of Labor Statistics, and fully backs up the Damore’s assertion about men overwhelming being subject to work related deaths. 

However, I had to locate this article myself, the link Damore included did not cite either the opinion piece, or the piece’s primary source as his (Damore’s) primary source.  Again, a missed opportunity, if only to establish strong credibility for the point that 93% of work fatalities are suffered by men. 

I would like to point out, again, that the link between the fatalities themselves and the reasons why these more dangerous positions are primarily staffed by men in overwhelming numbers is not supported in the Google Memo, nor actually in the opinion piece Damore uses to support this assertion. 

The opinion piece makes the same leap in logic and reason as Damore ---which is that biological gender preferences lead directly to occupational choices.  Furthermore, showing that dangerous jobs are taken by men does not actually address the other half of the author’s assertion, which is that high status jobs that are not dangerous are both chosen by men because of a biological desire to achieve status, and/or avoided by women due to a desire to attain work-life balance.  

Are high status jobs filled with men? A cursory search of five high status jobs yields interesting results:  

Doctors (as of 2016) about 50% more male doctors than female.  

Lawyers (as of 2016) about a 60/40 split, favoring males  

College Professors (as of 2016) about a 60/40 split, favoring males  

Entrepreneurs (as of 2016) women own 38% of all businesses, meaning there is a 40/60 split favoring men We can see that the data is suggestive that high status jobs are filled by men if not primarily, than heavily or in a majority percentage. 

Why didn’t Damore link to these strong primary source studies? Why didn’t he show using data that across key high status jobs, men are in the majority, as late as last year? Why didn’t he define “high status” jobs, as this study does more clearly? Another missed opportunity, and one that may have gone a significant distance towards proving his initial point, which is that men are drawn to and populate high status jobs.  

The final part of his argument that addresses gender differences in relation to occupation is called Non-Discriminatory ways to reduce the gender gap.  

Again, we see a paragraph with no citations, including a few statements which are assertions Damore appears to believe he has conclusively proven as fact earlier in the document and therefore, we could conclude that he doesn’t feel he needs to additionally cite sources.  

Below I'll go over some of the differences in distribution of traits between men and women that I outlined in the previous section and suggest ways to address them to increase women's representation in tech without resorting to discrimination. 

Google is already making strides in many of these areas, but I think it's still instructive to list them: 

● Women on average show a higher interest in people and men in things 

○ We can make software engineering more people-oriented with pair programming and more collaboration. Unfortunately, there may be limits to how people-oriented certain roles at Google can be and we shouldn't deceive ourselves or students into thinking otherwise (some of our programs to get female students into coding might be doing this).

 ● Women on average are more cooperative 

○ Allow those exhibiting cooperative behavior to thrive. Recent updates to Perf may be doing this to an extent, but maybe there's more we can do. 

However, the assertion that Women on average are more cooperative is new to the overall argument, and is not cited or supported at all.  

The citation he does use in this paragraph is here: Competitiveness and self-reliance can be valuable traits and we shouldn't necessarily disadvantage those that have them, like what's been done in education.  

This link is an article in the Atlantic Magazine that is 17 years old. However, I consider it a moderately credible source, as it is well articulated, well supported, and the internal links go to other moderately or strongly credible sources.  

However, the issue with using this source is that how boys and girls are treated differently in schools argues that society is valuing one set of traits over another. That may be true. This argument is not directly related to the point that women have certain biological traits that result in conscious, proactive choices to avoid certain roles, and that this “may” be the reason for gender disparity in certain roles. It is, in my opinion, a red herring and does not need to be included in Damore’s memo.  Moving on, he states Women on average are more prone to anxiety. This is not linked or supported in any way.  

He has a footnote that has two citations in it. One is included here: For heterosexual romantic relationships, men are more strongly judged by status and women by beauty. 

This link goes to a scientific study from 1991. It is 28 years old. Science is not static, it is extremely dynamic and arguments between scientists make up a large part of the canon. Major social and cultural changes have occurred in the last 30 years, among them the aging of one generation, the arrival of an entirely new generation, more than one economic crisis and depression in the US, a major shift in certain civil rights in minority populations, major technological advances that allow for a more diverse workplace, and several significant demographic shifts in the US population.  I would argue that in light of the multiple major shifts in both culture and economics the US has undergone in the near 30 years since this study was published, this should be considered outdated and it is a very weak primary source at best.  

His next link is here: Again, this has biological origins and is culturally universal.  The first assertion “biological”, is backed by a scientific study that is very complex and rigorous. In short, it is about testosterone increasing “status-seeking” behaviors in men. This does prove that biology can/may have an effect on one aspect of male behavior, certainly. It could explain the high numbers of men in certain jobs, but Damore does not conclusively link this to evidence that prove this.  

It does not address status seeking behaviors in the workplace, and it does not address women’s biological or hormonal peculiarities that may lead them to take lower- status jobs. If Damore’s main defense for his argument is that “men seek high status jobs because they are biologically compelled to” that is only half the argument.  

Just because one half of the population does something for certain reasons and we can prove it (partly) it does not necessarily flow that the opposite, negative argument is therefore true. In this case, there is no cited evidence that women do not seek high status jobs due to biological factors. The evidence he cites is psychological or contextual (self-reported standardized personality tests, not science or biology).  Damore does not address fully or use citations or support for his “culturally universal” phase, the second part of his footnote.  The next statement with a citation is Women on average look for more work-life balance. This link is to an academic paper (interestingly, the department of Sociology, which is a type of academic department that Damore elsewhere dismisses as inherently liberal and contaminated with confirmation bias) that discusses the differences in gender as they relate to occupational choices. 

It is from 2006, so it is right on the edge of what would be considered a contemporary source, and it concerns women from the UK, so it may or may not be applicable to women in the US. However, it is a strong primary source and appropriately cited.  

Something to note about this line item is that while the article conclusively and solidly explains that women choose either home centric or flexible work up to a percentage of 80% of all women, it does not address other possible reasons such as lack of available high status options, the fact that women’s entrance into a field lowers it high status perception and high pay, hostile work environments and sexual harassment that cause women to leave or stay in lower management jobs, or (to me the most important lacuna) the extra work women do in child rearing and housekeeping despite having a partner and working a full-time job. It may very well be that women choose part time or flexible work because they are also working a full-time job in raising children and running a household, not because it is their natural, biological preference.  

The next link appears to be internal to Google and can’t be accessed. It is “Allowing and truly endorsing (as part of our culture) part time work though can keep more women in tech.” I assume it is a link to an internal job board or something similar.  

The remainder of the Google Memo discusses the “Harm of Google’s Biases” in some detail and while it is certainly worth discussion, I am not in a position to do so, as:  

· I don’t know exactly what was said in the unconscious bias class that lead to this memo in the first place  

· I don’t know exactly what Google’s current policies are  I don’t feel that the arguments in this part of the document can be effectively discussed by me, as they are inflammatory to me personally to a degree that I feel my judgement and objectivity would be compromised.  My remarks on this document:  I count 11 accessible citations in the first half of the Google Memo.  

They are:  

1. Women on average have more  

2. People rather than things  

3. Empathizing vs. systemizing  

4. Neuroticism  

5. Research suggests  

6. Stop assuming that gender gaps imply sexism  

7. Suffer 93% of work-related deaths 

8. What’s been done in education  

9. Men are more strongly judged by status and women by beauty  

10. Biological  

11. Women on average look for more work life balance  

Let’s take a look at the overall strength and credibility of those citations:    

Overall, Damore’s document is not 100% false, or misguided. It is not 100% weak. However, based on my application of rigorous academic rubrics, it is very weak and the arguments are not well supported with current, primary, credible sources.  

My suggestions:  

· Do not use Wikipedia for 36% of your citations in a position paper. It irreparably damages your credibility 

· Do not use articles that are more than 10 years old for your support, it damages your credibility

 · Do not partly support one half of an argument and expect your audience to fill in the blanks  · Do not use citations to define terms  

· Do not conflate hard and soft sciences to prove hard science theories  

· Develop each part of your argument in a multi part argument fully before going on to the next part of your argument       

Thanks for reading this unusually long article. Please share, comment, or submit your own rebuttal :)