Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Kolmogorov Complicity and the Parable of Lightning (2017)

slatestarcodex.com

Post reply on HN
Print view View original

71–80 of 106 posts

Re: Kolmogorov Complicity and the Parable of Lightning (2017)

#71

Post by jbay808 »

RangerScience wrote: post #46
ALittleLight wrote: post #38

Earlier quoted context omitted.

I read Damore's document and your summary and description of it seems entirely foreign and inconsistent to me. For example, you summarize it as "these are the insurmountable blockers to this kind of success for people who aren't like me". In Damore's document he discusses ways to promote diversity within Google by accounting for differences between men and women and adjusting Google's hiring and promotion practices t…

I think I can. Here's the document: https://s3.documentcloud.org/documents/3914586/Googles-Ideol... JD says: "On average, men and women biologically differ in many ways", and then has the nice overlapping bimodal distribution graph. In my read, this data is foundational to many of his arguments, but there's a fundamental flaw in his application of it, which is: why you should expect anyone at Google (or, well, any su…

> why you should expect anyone at Google (or, well, any sub-group with a filter process) would be meaningfully predictable by or representative of their biology's bimodal distribution, in the first place?

I don't think you would; however you might find you end up with a ratio of both groups that is not 50/50 but rather proportional to the ratio between the areas of both curves that lie above some cutoff.

For a maybe less controversial take, as of 2021 there are far more women than men who do, say, embroidery as a hobby. In that sense "population by hours spent embroidering per year for fun" is two overlapping histograms that skew differently by gender. Big peaks for both groups at zero. However, once we filter our discussion to "people who do embroidery as a hobby", cutting off the majority of both populations (but the supermajority of men), we can't assume that the same pattern holds. The trend might even flip; the average man-who-does-embroidery might do it much more than the average woman-who-does-embroidery, just because they've been so strongly selected for; less committed men never would have picked up that hobby at all.

But we can be pretty sure that the population of men who spend even an hour a year embroidering is much smaller than the population of women who do the same, without making any assumptions that those men who do embroidery have any lesser skill, interest, or dedication.

Re: Kolmogorov Complicity and the Parable of Lightning (2017)

#72

Post by clavicat »

Lambent_Cactus wrote: post #37
exolymph wrote: post #6

This was published in response to a lament about James Damore, in case anyone is wondering about the immediate context. The proximate issue was that women and men, on average / in aggregate, are different. But the general principles apply to other issues as well.

The problem with signing up with an underground Straussian heretic network is that if you're not actually beset by the Inquisition or by Stalinism, you're likelier to trap yourself in an epistemic filter bubble than to discover some secret truth. Damore wasn't burned at the stake or shot in the head; he was fired from one especially cushy software job at a giant, publicly-traded company. And rather than grappling wit…

>Damore wasn't burned at the stake or shot in the head; he was fired from one especially cushy software job at a giant, publicly-traded company.

How do you think regimes normally deal with dissent? Extreme violence makes for vivid imagery but it’s not typical. Usually, it’s as simple as threatening to fire you from respectable employment and education. This is how the entire Eastern Bloc kept a lid on things.

Re: Kolmogorov Complicity and the Parable of Lightning (2017)

#73

Post by joshuamorton »

ALittleLight wrote: post #55
RangerScience wrote: post #46

Earlier quoted context omitted.

I think I can. Here's the document: https://s3.documentcloud.org/documents/3914586/Googles-Ideol... JD says: "On average, men and women biologically differ in many ways", and then has the nice overlapping bimodal distribution graph. In my read, this data is foundational to many of his arguments, but there's a fundamental flaw in his application of it, which is: why you should expect anyone at Google (or, well, any su…

I read your comment opposed to Damore as something like "Damore says this distribution of traits exists in the general population, but it doesn't necessarily exist in the subset of Google." Please correct me if I'm wrong. I think that point is true and I think it is the kind of thing where, if Google were considering taking Damore's feedback, they should investigate. Is Google's population of employees noticeably dif…

> I think that the assumption that Google, with 150k (or whatever it was at the time) employees plus candidate employees, roughly approximates population trends is a reasonable starting place for a single author writing up some thoughts. If this were an academic paper we would probably expect the author to consider this at some point. If it's some thoughts the author put together on diversity? Seems like a crazy standard to me.

But we know this to be untrue: Damore's questions come up as a response to the question of "why does Google have an unrepresentatively low number of women" (and from that, the implied "how do we fix that"). To assume that Google's employees match broad population trends is to assume the entire conversation moot. In fact, everyone involved agrees that Google doesn't match general population trends, so if you're having to assume that to justify Damore's takes, something has gone wrong somewhere!

> Google hosts a diversity class, the class asks for feedback, Damore shares his thoughts on diversity

For what its worth, this is a pretty serious misrepresentation of things, and if that were the extent of things Damore wouldn't have been fired. As I understand it, he gave the feedback to HR and nothing happened. He proceeded to post the document in larger and larger discussion groups, until he eventually posted it into a really big group, where it was linked to an even bigger group. That's also a big part of what moves this from "feedback" to "manifesto", its disingenuous to call it "feedback on a diversity class" when you're emailing it to a thousand people, nearly none of whom ever took the class. That's pretty far into thesis nailed to the church door territory.

The other thing that puts it into "screed" territory is that it doesn't have a coherent thesis. Is it about gender diversity? Well a lot of it is, but it also has a significant subtheme about political diversity (and trying to psychoanalyze Google as a "left-biased" company), and has a number of potshots at "the left". Why?

Re: Kolmogorov Complicity and the Parable of Lightning (2017)

#74

Post by throwaway81523 »

civilized wrote: post #57
Lambent_Cactus wrote: post #51

Earlier quoted context omitted.

Most of his articles could be about a third as long with no loss in clarity. The meandering windups can be _rhetorically_ effective, especially when they appeal to the biases of the reader, as they appeal to mine. But they're not truth-directed, and often obscure weak or blindered reasoinng, as I think they do here.

Isn't this a bit cliche? "It's too long" without engagement with the argument or even with how exactly it is too long? I've just seen this a few too many times in regards to Scott and would like to see someone actually substantiate it.

If you can make the same points with 1/10th as many words, the original was too long. The guy writes extremely well, but could use some practice at editing for brevity. A wikipedia link like https://en.wikipedia.org/wiki/Pluralistic_ignorance gets partway there.

Re: Kolmogorov Complicity and the Parable of Lightning (2017)

#75

Post by nextaccountic »

exolymph wrote: post #6

This was published in response to a lament about James Damore, in case anyone is wondering about the immediate context. The proximate issue was that women and men, on average / in aggregate, are different. But the general principles apply to other issues as well.

Do you have a link that discusses this?

Re: Kolmogorov Complicity and the Parable of Lightning (2017)

#76

Post by ALittleLight »

joshuamorton wrote: post #73
ALittleLight wrote: post #55

Earlier quoted context omitted.

I read your comment opposed to Damore as something like "Damore says this distribution of traits exists in the general population, but it doesn't necessarily exist in the subset of Google." Please correct me if I'm wrong. I think that point is true and I think it is the kind of thing where, if Google were considering taking Damore's feedback, they should investigate. Is Google's population of employees noticeably dif…

> I think that the assumption that Google, with 150k (or whatever it was at the time) employees plus candidate employees, roughly approximates population trends is a reasonable starting place for a single author writing up some thoughts. If this were an academic paper we would probably expect the author to consider this at some point. If it's some thoughts the author put together on diversity? Seems like a crazy stan…

I don't know why you are saying "everyone involved agrees that Google doesn't match broad population trends". Are the men at Google taller than the women at Google? Google employees are probably unusual in variables that relate to their work with computers, but it's not at all obvious that they would be unusual on other traits, like height, for example, or assertiveness, cooperativeness, etc.

Regarding Damore sharing his memo - he shared it in places where the topic was being discussed. He wasn't mailing this to all workers, but posting it in groups that were discussing diversity and asking for feedback. If Damore's memo had reached different conclusions - like say that Google had fewer women because of sexism in education and implicit biases among hiring committees, and he had shared it in the same places, would he have been fired? I think we both know the answer to that.

Re: Kolmogorov Complicity and the Parable of Lightning (2017)

#77

Post by lordgrenville »

throwaway81523 wrote: post #74
civilized wrote: post #57

Earlier quoted context omitted.

Isn't this a bit cliche? "It's too long" without engagement with the argument or even with how exactly it is too long? I've just seen this a few too many times in regards to Scott and would like to see someone actually substantiate it.

If you can make the same points with 1/10th as many words, the original was too long. The guy writes extremely well, but could use some practice at editing for brevity. A wikipedia link like https://en.wikipedia.org/wiki/Pluralistic_ignorance gets partway there.

Distilling it to 10% would get the facts across, but the long build-up of argument makes it much more convincing. Rhetoric is different from, say, technical writing: it optimises not (just) for efficiency but for effectiveness.

Re: Kolmogorov Complicity and the Parable of Lightning (2017)

#78

Post by ciphol »

throwaway81523 wrote: post #74
civilized wrote: post #57

Earlier quoted context omitted.

Isn't this a bit cliche? "It's too long" without engagement with the argument or even with how exactly it is too long? I've just seen this a few too many times in regards to Scott and would like to see someone actually substantiate it.

If you can make the same points with 1/10th as many words, the original was too long. The guy writes extremely well, but could use some practice at editing for brevity. A wikipedia link like https://en.wikipedia.org/wiki/Pluralistic_ignorance gets partway there.

The SSC post is MUCH better written than that Wikipedia article.

Re: Kolmogorov Complicity and the Parable of Lightning (2017)

#79

Post by joshuamorton »

ALittleLight wrote: post #76
joshuamorton wrote: post #73

Earlier quoted context omitted.

> I think that the assumption that Google, with 150k (or whatever it was at the time) employees plus candidate employees, roughly approximates population trends is a reasonable starting place for a single author writing up some thoughts. If this were an academic paper we would probably expect the author to consider this at some point. If it's some thoughts the author put together on diversity? Seems like a crazy stan…

I don't know why you are saying "everyone involved agrees that Google doesn't match broad population trends". Are the men at Google taller than the women at Google? Google employees are probably unusual in variables that relate to their work with computers, but it's not at all obvious that they would be unusual on other traits, like height, for example, or assertiveness, cooperativeness, etc. Regarding Damore sharing…

> I don't know why you are saying "everyone involved agrees that Google doesn't match broad population trends".

The gender balance at Google is not representative of the broad population. Presumably there are axes on which Google's demographics match population demographics, but we're talking about one axis where everyone involved agrees that they don't.

> Regarding Damore sharing his memo - he shared it in places where the topic was being discussed. He wasn't mailing this to all workers, but posting it in groups that were discussing diversity and asking for feedback.

Right, and it got basically no response, until he finally posted it in a very large group that wasn't related to diversity. As I understand it, there were months between when he initially wrote and shared it with D&I people and and when he posted it in the big group where it went viral.

> If Damore's memo had reached different conclusions - like say that Google had fewer women because of sexism in education and implicit biases among hiring committees

I'm not clear what your point is. You seem to be arguing that "because one view is acceptable to express in the workplace, we must be allowed to express the opposite view in the workplace" or something. But that doesn't make a whole lot of sense.

Re: Kolmogorov Complicity and the Parable of Lightning (2017)

#80

Post by f137 »

RangerScience wrote: post #19
civilized wrote: post #11

Earlier quoted context omitted.

Ah yes, Mr. Damore. Author of a document the entire prestige media called an "anti-diversity screed" in perfect lockstep, even though anyone could read it for themselves and see it was nothing of the sort. 2017 sure was a year for the history books.

I found that JD's document was a moderately anti-diversity screed, tho. I did a deep read (the document, everything the document linked to, most of the things that those linked to) back when it all went down. IMO, while James raised a valid question ("how do we know any of this is working, or even can work?"), some of his mistakes were in attempting to answer it (instead of solidly beginning a conversation); attempti…

So, you basically say he was wrong to state his opinion openly.

> I really, really wonder what would have happened if he'd sought out diverse co-authors.

I wonder too what would have happened if Jesus sought out diverse co-authors for his teachings.

Post reply on HN