Hacker News Reader

Comment by jackvalentine | Hacker News Reader

I read his memo (There is a copy on Gizmodo) and while I saw a lot of what I could educatedly guess looked like flaws in logic or even flaws in his reading of the research I've been unable to find someone who has put the effort in to debunk it thoroughly.

Rather all I've been able to find is the equivalent of how dare he say these things.

Does anyone have a link to a good debunking of the whole thesis?

View on HN

> Rather all I've been able to find is the equivalent of how dare he say these things.

I completely agree. The original memo may be wrong, but it was a cogently and constructively stated position (if a bit thin on references) and it deserves to be at worst ignored, more likely respectfully rebutted, and (dare I even say this?) perhaps taken seriously. Because I think one of the reasons people are so up in arms about it is that they are afraid that if they start to try to rebut the argument they will find that it is not so easy to do, and possibly even that parts of the argument might even be correct.

> a bit thin on references

I've seen mentions of the fact that the Gizmodo version removed most of the links/references/graphs that were included in the original.

Oh dear. If that's the case then I'm very confused.

That what, Gizmodo might mislead people for clicks?

The original is here:

https://assets.documentcloud.org/documents/3914586/Googles-I...

At least I'm pretty sure that is the original. It has the author's name on it.

Wow, you're right, per the lisper link. That is extremely shady of Gizmodo.

Sigh, it wasn't gizmodo being shady, it's just there's multiple leaked versions and some of the leakers only supplied html/text.

"The text of the post is reproduced in full below, with some minor formatting modifications. Two charts and several hyperlinks are also omitted."

Per the article itself.

There are a few, and I'd suggest you look further, for starters scroll up here: https://twitter.com/GlennF. Arguments about gender differences in CS should address the large decline in women's participation around 1984, the relatively better gender balance in other engineering/science majors, or the better gender balance for CS degrees granted in other countries.

Annie Lowrey and others argued institutional effects far outweigh biological or IQ effects; basically, there's a much larger environment of harassment and bias, and Google has so many other levers in its control that the arguments advanced in the paper aren't really salient or important to the situation at Google.

Other folks have argued the vast majority of engineers at Google are doing the equivalent of string concatenation; it's not exactly a difficult or demanding job.

I think one reason people have been so angry about this is that women in the tech industry generally deal with much more harassment and bias than men do, see for example elephantinthevalley.com. To argue these are the biggest issues requires looking past everything in that survey, and implicitly declaring it not that important or salient. Imagine going to Flint, Michigan and arguing that a lack of charter schools are the community's biggest problem; you're going to get some bad reactions.

People also tend not to like it when you suggest they're dumb or inferior.

I think it is telling that the only actual reference you provided was to a live Twitter feed. Twitter is not the right venue for any kind of substantive debate about anything, let alone one where the opening salvo is a ten-page document.

Twitter is an excellent venue for discussion. It's one thing to complain that you don't like it, and another to pretend it's widely agreed upon as an invalid source.

> Twitter is an excellent venue for discussion.

I didn't say it wasn't. Is said it was a bad venue for substantive debate. It's a fine venue for banal chitchat and voicing unsubstantiated opinions, which is what makes it so popular because that's a lot easier and more fun than actually diving into the depths of an issue.

Absurd. It's a fantastic venue for substantive debate, technical political and mediatic. You're speaking confidently about something you're ignorant about.

Can you supply examples?

Twitter, by virtue of being Twitter, is not a good medium for substantive debate. The forum is the most populated one in the entire world. It is amazing for equalizing influence and allowing democratic propagation of information (which would otherwise be under the control of media firms), but this also means the bar for voicing one's opinion is in the Mariana's Trench. Anyone can butt in an derail the debate.

Twitter moderation is also very clearly biased in one direction (if you were to simply compare how often each side's individual account's are banned) and thus not an open medium where those debating can speak freely without consequence.

There's also the greater scheme of digital communications which are notorious for not being in anyway productive.

Debate online just simply does not work with the way human communication biologically evolved.

I like it because it essentially forces "If I had more time, I would have written a shorter letter" — I've read more than enough "limitless" threads of utter rubbish, longer garbage is still garbage. There are positives and negatives of every platform.

the large decline in women's participation around 1984

From what I've seen, that's a decline in the proportion of women enrolled in CS, not a decline in the absolute number. So this could be adequately explained by men crowding into CS rather than women fleeing. Do you have the numbers to show otherwise?

That's not the argument made by the author, however, his argument is the gender imbalance is not concerning because women have, on average, smaller brains and lower IQ's.

> his argument is the gender imbalance is not concerning because women have, on average, smaller brains and lower IQ's.

Where does he say that? Or are you trying to be sarcastic?

This guy takes a crack at exactly what you propose. The factors you mention do not undermine the manifesto's thesis. https://nintil.com/2017/08/07/why-so-few-women-in-cs-the-goo...

> ... the vast majority of engineers at Google are doing the equivalent of string concatenation ...

wait, what? is that even remotely true? that sounds unbelievable on its face.

String concatenation is slightly facile, but in the sense that most people are writing web servers, which are largely just composing templates, and not, e.g. Working on the distributed lock service.

I don't know that there's a debunking of the thesis, because there's nothing very original in there -- it's repeats and retreads of other things that have been debunked over and over, so it all merges into a mashed up mess of points and rebuttals like a badly maintained wiki.

There are a couple of books which I can recommend that talk about why you see so few women in computing (as opposed to biology, math, etc) and also why you see so few black men and women:

    * Unlocking the Clubhouse: http://a.co/8oXDROX
    * Stuck in the Shallow End: http://a.co/9HXgvKW

These are not only well written, but they're also well researched and well thought out. They're worth reading on their own merits, but you may find them a useful debunking to the memo.

TBH I'm not really interested in the wider issue at all. For starters I'm not even in America and we have our own issues here.

But this memo has dominated a lot of the internet that I use and figured surely someone has done the work to pull it apart.

I'm not going to read two books just to figure out some google guy's 10 page memo - call me lazy but perhaps nobody is going to read two books just for that purpose.

> I'm not going to read two books just to figure out some google guy's 10 page memo - call me lazy but perhaps nobody is going to read two books just for that purpose.

Well, if it's any consolation, I wouldn't read it for that either. The blog posts and memos do accumulate, though, and at some point it's worth it just to have the sources to refer back to.

Good point.

Also, the larger problem with debunking is that it just doesn't work all that well. It's like arguing with an antivaxxer -- they may ask for logical discussion but it has no effect. Ultimately, what convinces them is social pressure or a sudden intrusion of reality i.e. their child gets sick, they have a female friend who encounters discrimination etc.

My view is that in the absence of accessible ways to fact-check the memo people will only see the memo and the gnashing of teeth surrounding it. That leaves a really big hole to be filled - the memo is already out there in a big way.

The is Google lol, not some guy from i hate women dot com. Perhaps based on his platform alone we can consider it an honest attempt at conversation and debate it with our own honesty.

So you assume he worked this through from first principles and is thinking critically and has checked his sources... because he works (well, worked) at Google?

I can tell you that the norm among engineers, even at Google, is to receive lore and anecdotes from people slightly more educated as if it were gospel, and to repeat it uncritically until someone else more charismatic comes along. This happens so often that someone wrote a book called [The Leprechauns of Software Engineering](https://leanpub.com/leprechauns) about the unsourced commmon knowledge that engineers will blindly repeat without thinking about it.

They've only been "debunked" in the lazy, anti-intellectual corners of academia; that is, gender studies, cultural/social anthropology, critical race theory, etc. The factual points made by the author (re: gender differences only, no opinion on his recommendations for Google) square very nicely with scientific consensus [1]. The evidence has only piled on since the famous Pinker-Spelke debate.

[1] https://www.edge.org/3rd_culture/debate05/debate05_index.htm...

I've only seen it in comments thus far, but here's a good example:

https://news.ycombinator.com/item?id=14951042

A quick Google, have not read the below: https://www.fastcompany.com/40449844/5-debunked-gender-myths...

Of course, internet comments will be reactionary. Pretending that is proving the point of the author is a self-fulfilling prophecy. I'm not saying you are inherently saying that, but I am sure many others will see it that way, as they already have commented in many of the threads about this.

+1 for the attempt, but the Fastcompany article is very weak at best, and doesn't even cite relevant studies to debunk the 5 points. It even manages to misquotes its own sources.

Not sure anyone has written a "debunking" piece but that could simply be because so much of what he wrote is standard trope for people with his views...if you read it naively it sounds like he has a point but if you're familiar with such attitudes then you'll quickly see through the pseudo-intellectual veneer of his argument.

I supposed one could take the time to rebut everything he wrote but there will probably be someone new with similar arguments next week.

What's important is that Google or any other organization striving for fairness (that is what diversity is partly about) isn't derailed by such sentiment.

That is poor reasoning and only adds fuel to the assertion that there is an echo chamber / shaming culture given the response to the assertions.

> "if you read it naively it sounds like he has a point but if you're familiar with such attitudes then you'll quickly see through the pseudo-intellectual veneer of his argument."

Well I suppose that's my point isn't it? Expecting everyone who reads it to be "familiar with such attitudes" in order to come away with the "right" opinion is an awfully high bar to entry.

No, that is not my expectation and I doubt it is the expectation of people who don't agree with his opinion.

There is just a long history of his type of argument deployed in favor of his type of view. Frankly, rebutting such arguments can get exhausting especially for those who are the target of his post.

How would you propose that this kind of blow-up be handled then in the future? It's going to happen again and in the absence of opposition that's just ceding ground without resistance to the kind of attitudes in the memo.

I'm thinking if you're going to report on it, you need to either pull it apart with rigour or accept that you're just increasing the reach of it without opposition.

Shouldn't it be up to the author to prove his claims rather than someone else to disprove them? I see no evidence provided for his stereotypes of women, for example. Something this unsubstantiated doesn't need debunking. It's simply one man's unsubstantiated opinion, nothing more or less.

"It's simply one man's unsubstantiated opinion, nothing more or less."

That's how it should be treated. Nothing more. A lot of the pro-diversity arguments are pretty flimsy too so they should be treated the same way too.

> Shouldn't it be up to the author to prove his claims rather than someone else to disprove them?

Look, yes. It should be. But the author has made his best 10 page effort to do so and now I'm looking for someone who has the necessary background to critically question his claims.

That the manifesto has taken on a life of its own and spread as far as it has is only more reason for it to be succinctly and academically critiqued.

It's a long document, but reading his tldr he makes claims that I've rarely heard reasonable people make. Notably:

The lack of discussion fosters the most extreme and authoritarian elements of this ideology. ○ Extreme: all disparities in representation are due to oppression ○ Authoritarian: we should discriminate to correct for this oppression

I've never heard that all disparities are due to oppression before. I think the problem is that we don't know how much of the gap is due to oppression. There are studies that indicate that there are factors in our society that do oppress the achievement of girls in math/science. How do we figure out how to clear up these issues?

And honestly, as someone who works in tech, I can honestly say that anecdotally I've encountered a LOT of sexist men in the field. I can't say it is the majority, but it is not a small minority. While I can't say with any certainty how much their biases impact the achievement of women in the field, I think most people outside of tech would be surprised at how much bias there is (because it is viewed as so much of a meritocracy).

And then his other point about discriminating to fix the oppression simply is naïve and ignores the day to day discrimination that happens against most oppressed groups on a daily basis. I get that a better solution would be for there to be no more discrimination across the board -- we are all born into a world where discrimination is gone (and past fruits of discrimination disappear too), but that isn't going to happen. Sometimes the choices we have in front of us are the lesser of two evils -- and yes, looking at an evil always looks bad -- but I guarantee you that the author of this article would never trade his life in for the roll of the dice of being born a black female.

What was strange to me is that it didn't seem to actually cite any sources. When I scrolled down to the bottom to read the footnotes, I expected links to purported studies, but instead, they were just more rambling. I'd think the burden of proof should be on the author for the dubious claims within.

EDIT: I looked at a link to the document on a separate comment on this thread, which appears to have links on some of the text that was missing from the Gizmodo article. Unfortunately, they aren't clickable, so I guess the most I can actually say about the citations is that they are rather informal.

From Gizmodo's summary[1]:

> Two charts and several hyperlinks are also omitted.

It's unclear whether Gizmodo removed the links or whether the leaker did. Either way, I bet some of those links were to research or summaries of research.

1. https://gizmodo.com/exclusive-heres-the-full-10-page-anti-di...

The links are clickable if you get the original PDF, rather than using DocumentCloud's PDF viewer. https://assets.documentcloud.org/documents/3914586/Googles-I...

Reply on news.ycombinator.com