News for hackers. Stuff that matters.
Re:The Politics of Pockets by brenschluss on Monday September 26, 2016 @08:01PM (#12586640)
> in that plan is not Macy's management, but rather women not wanting to buy those pants.

And why do you think that is? Diagnosing the symptom is far easier than understanding the cause. You're doing the former.

History allows us to actually understand how historical/societal trends have changed over time. To ignore that is to be overly simplistic.

Like saying "Q: Why is the '<a>' tag used to make links in HTML?" "A: Because browsers interpret <a> as a link!"

Re:The Politics of Pockets by dlss on Monday September 26, 2016 @09:16PM (#12586926)
> And why do you think that is? Diagnosing the symptom is far easier than understanding the cause. You're doing the former.

When you're talking about emergent behavior of extremely complex systems, "why" is incredibly hard to answer accurately and incredibly easy to answer inaccurately. I honestly have no clue why women en mass (apparently) don't care for pockets. All I can say is that would seem to be their aggregate preference. Can you honestly tell me that anyone can say more?

For example, I also can't tell you why I exist. I'm aware that there are many groups who claim to know why (god), but in my opinion their explanations are wrong. I'm satisfied to simply #include evolution, and note that my female ancestors and male ancestors probably liked each other, at least on average.

I also can't really tell you why I checked a bug into the project I'm working on yesterday. I can tell you what the bug was, and what the fix is... but why? I'm not sure. Maybe human brains just have a non-zero failure rate? I really didn't want to code it wrong, but it happened anyway. All I'm really sure about with respect to my bug is that the answer probably isn't because there's a historical trend toward bugs.

Circling back, there are many areas in modern life where women have deviated strongly from their historical conditions. You might even say that deviation from historical trends is itself a historical trend. But you can't say that without giving up on historical trends as a useful explanation :p

Re:The Politics of Pockets by brenschluss on Tuesday September 27, 2016 @01:18PM (#12591237)
> "why" is incredibly hard to answer accurately and incredibly easy to answer inaccurately.

Of course. Yet we should try, shouldn't we?

Your life, and mine, are made out of millions of choices that have been made by people before you. The shape of the building you're in right now, the size and proportion of your chair, the computer, the width of the roads you drove on, or the form of the subway you were on. The fact that you're typing these letters on a keyboard in a linear sequence, from left to right. There are countless historical reasons why you

To say "I don't know why any of these happened; I'm just going to look at the present" is fine. But it makes it impossible to evaluate any action in context. Your profile says you're a statistician - surely you understand the value of understanding things in context?

So, here's another example:

Bob says: "I honestly have no clue why black males en mass (apparently) commit the most crimes. All I can say is that would seem to be their aggregate behavior. We should stop and frisk and detain black males based on these statistics. Can you honestly tell me that this is wrong?"

Would you agree? Why examine pesky historical trends, and the history of slavery, the civil war, structural racism? After all, 'why' is incredibly hard to answer accurately. The facts are clear, after all. Isn't it okay to enact policies based on the present that don't care at all about the past? /devilsadvocate

Re:The Politics of Pockets by dlss on Wednesday September 28, 2016 @09:51AM (#12597588)
>> "why" is incredibly hard to answer accurately and incredibly easy to answer inaccurately.

> Of course. Yet we should try, shouldn't we?

I mean, maybe? If you see a way to do it that you think would actually work? What I was saying above is that I'm not aware of any methods that would work, and that I've spent quite a while looking for one. At best it's a pure waste of time, at worst you accidentally trick yourself into believing you've found something.

Same example: It's very unlikely that even if there is a reason "why" you and I exist, that we will ever learn the answer. To spend time on this problem looks like it's at best wasteful, and at worst results in insanity or cult formation.

> Bob says: "I honestly have no clue why black males en mass (apparently) commit the most crimes. All I can say is that would seem to be their aggregate behavior. We should stop and frisk and detain black males based on these statistics. Can you honestly tell me that this is wrong?"

This is a very very different statement than the one above. Although the sentences have similar structure, mine is about the limits of knowledge, and yours is someone being overconfident about a policy recommendation (since there are presumably many other possible approaches to policing that would need to be considered before a policy could be deemed best).

If you're asking how I'd respond to someone saying that today? I dunno, the hypothetical speaker sounds dumb. I guess I'd ask for references and do some reading before coming to a conclusion. That's what I did for the pockets argument, and how I came upon the above economic rebuttal.

I also really don't see what stop and frisk has to do with black crime statistics mentioned above... are you saying the statute literally mentioned race? If so that seems dumb / unjustifiably specific. If not, well, you just failed an ideological turning test[1].

My understanding as someone who doesn't follow politics is that some parts of the country are sufficiently high crime that people want to designate those places as essentially airports (so you'll need to be screened if you'd like to enter). That doesn't sound insane to me, but then again I haven't really read anything on the subject.

[1] https://en.wikipedia.org/wiki/Ideological_Turing_Test

Re:The Politics of Pockets by brenschluss on Wednesday September 28, 2016 @01:13PM (#12599420)
> What I was saying above is that I'm not aware of any methods that would work, and that I've spent quite a while looking for one.

> Same example: It's very unlikely that even if there is a reason "why" you and I exist, that we will ever learn the answer. To spend time on this problem looks like it's at best wasteful, and at worst results in insanity or cult formation.

History! Research! That's the best we can do, but there are entire disciplines devoted to trying to understand.

I would rather that humankind as a species is enlightened enough to try to understand the universe, past and future and present, and to consider that 'why' and 'how' is a more important question then, I don't know, appeasing some shareholders.

It sounds like you're interested in highly deterministic ideas of why, and any form of inaccuracy is unacceptable, so to find highly accurate 'proofs' for 'why' something happened in such a highly complex system as "the world" is nearly impossible. Better not to try.

It's difficult, yes, because there are no clear answers. But part of many things is about making decisions without clear answers, or clear reasons. Does it sound like madness? Sure, from the point of view of statistics or pure models. But in the field, in situ, improvisation and heuristics and 'best-guess' progress is currently what we have.

> I also really don't see what stop and frisk has to do with black crime statistics mentioned above...

Stop and frisk is racial profiling. Racial profiling is based upon prejudices or statistics about crime based on race. Please do a little bit of research. There's a reason why it was ruled unconstitutional in New York City.[1]

--

At the end of the day, I think that approaches like yours are dangerous because the choices you make will unwittingly map onto the choices that others have made, previously. What seems like an independent decision with independent analysis actually is an amplifier of others' existing decisions. I am suggesting that asking "why" is the only way we can have independent and clear thinking.

Take for example 'turning high crime areas into required screening areas' that "doesn't sound insane" to you.

On the surface, it might make sense. Dangerous areas should be segregated. That way we can lower the danger. Diagnose the symptom and stop it.

However, if you understood the impact of ghettos, the impact of prisons (which is a form of segregation through spatial/societal exile), and the history of racial segregation through financial policies like redlining[2], you'd understand that you'd be setting a policy that was the exact same reason why these high crime areas happened in the first place. You'd be amplifying a feedback loop.

The only way to understand is to do some research and understand what the major factors for crime are. Is it a lack of jobs? A drug-trade fueled underground economy? A lack of education? A lack of strong familial/social structures? What improvement would have the biggest impact in the area?

Are these analyses trying to understand 'why' going to be rock-solid 100% verifiable? No. Is the symptom 100% verifiable? Yes. Despite that, it makes sense to at least TRY to understand the cause.

[1] https://en.wikipedia.org/wiki/Stop-and-frisk_in_New_York_Cit... [2] http://www.theatlantic.com/business/archive/2014/05/the-raci...

Re:The Politics of Pockets by dlss on Wednesday September 28, 2016 @01:56PM (#12599771)
The point I was trying to make was not that good judgement under uncertainty is impossible. As you mentioned, I'm a statistician... as such, I'm 100% certain I would never say that :p

What I was trying to say was that the methods commonly used in social sciences departments, and the watered down versions used in popular social science writing (perhaps typified by OP's essay) are basically a waste of time if your goal is understanding rather than entertainment.

That's not to say I think that real world action (including ethical action) is impossible, or that it's not worth doing. That's to say I have a strong, somewhat professional, opinion on how it should be done.

I am not alone in this. There are groups of statistically minded people who are making headway avoiding the traps of "common sense". http://www.givewell.org might be a good introduction, perhaps reading about how the Bill & Melinda Gates Foundation allocates funding would also be helpful if you'd like details. They do around ~500 times more measurable good per dollar donated than the average charity (source: http://www.givewell.org/international/technical/programs/ins... etc). That this level of improvement was possible is a harsh critique of the sort of watered down social sciences methods we're discussing.

> I am suggesting that asking "why" is the only way we can have independent and clear thinking.

I mean, I get that you're saying that... but oh man is a citation needed there. "What"/"How" is normally the question that leads to insight and paradigm change in my experience. I don't usually cite Kuhn, but https://en.wikipedia.org/wiki/The_Structure_of_Scientific_Re... might be worth a read for you.

Why almost invariably leads to https://en.wikipedia.org/wiki/Attribution_bias and/or https://en.wikipedia.org/wiki/Fundamental_attribution_error , both in my experience and according to the literature.

> However, if you understood the impact of ghettos, the impact of prisons (which is a form of segregation through spatial/societal exile), and the history of racial segregation through financial policies like redlining[2], you'd understand that you'd be setting a policy that was the exact same reason why these high crime areas happened in the first place. You'd be amplifying a feedback loop.

> The only way to understand is to do some research and understand what the major factors for crime are. Is it a lack of jobs? A drug-trade fueled underground economy? A lack of education? A lack of strong familial/social structures? What improvement would have the biggest impact in the area?

I'm so confused. You think the reason people have proposed stop and frisk is because they haven't considered what the major factors correlating with future crime are? (ie what makes a high risk area, how to city plan around that, etc) I would be very surprised if that were the case. If it's a stupid policy (and I'm more than willing to believe that it is) I would have thought it would be an example of shoddy social sciences thinking (ie people who shouldn't have asked "why")

How do you think government officials decide which policies to enact? They went to some of our nations best schools (yes, that's meant to be a critique of those schools). It's not like doing what social science has been doing, only harder, will somehow change the results we're seeing.

Or at least that's my best guess. As we start talking about academia being dumb I have read more :p

----

If you haven't read it "Politics is the Mind-Killer" (http://lesswrong.com/lw/gw/politics_is_the_mindkiller/) is short and rather amazing as well. (Not wanting to go funny headed is the main reason I've avoided reading about stop and frisk).

Just for fun here's a quiz letting you guess which interventions do the most good (as you might imagine "common sense" isn't going to be much help) https://80000hours.org/articles/can-you-guess/

Re:The Politics of Pockets by brenschluss on Wednesday September 28, 2016 @06:00PM (#12601693)
> the methods commonly used in social sciences departments

What methods? Could you clarify? There's an immense diversity within 'social sciences' departments; are you including all forms of history/theory, including the history of technology as well?

> There are groups of statistically minded people who are making headway avoiding the traps of "common sense". http://www.givewell.org might be a good introduction

I'm familiar with Givewell. As a non-profit, Givewell does good work, but they perform a form of meta analysis of other non-profits, and do not do direct work. As a result, Givewell isn't a good example of "real world action",

I'd love to see another example of 'real world action' that you think makes good judgment under uncertainty that isn't a form of meta-analysis.

As a software developer AND architect (the bricks and mortar kind, not the software kind), I find that my logic-oriented education is often less-than helpful on the construction site. What's more important is a holistic (or cybernetics[1]) way to understand a series of entangled questions - why people use space, how they interact with a building, how a structure is constructed, what technologies are used, construction time and cost, risk, relationship to building codes, building inspectors et cetera.

So to me, a form of thought that thinks asking "why" is "insanity or cult formation" is very well-meaning but ill-informed in terms of much forms of action. It works on the screen, but it doesn't work when it interfaces with complexity.

Or take it this way. Logic-based thinking works in a highly controlled environment, where the logical axioms are pure and easy. Heuristic-based action works in a highly complex environment, where there are no axioms, rules are always bent, and things are always interconnected in highly unpredictable ways.

> "What"/"How" is normally the question that leads to insight and paradigm change in my experience.

Sure, I've read SSR. So to clarify, your argument is that spending time on 'why' is "at best wasteful", and thus "what"/"how" is better?

I think we agree here, but I'd say that it seems like you're in disagreement with your previous answer.

> I'm so confused. You think the reason people have proposed stop and frisk is because they haven't considered what the major factors correlating with future crime are?

> How do you think government officials decide which policies to enact? They went to some of our nations best schools (yes, that's meant to be a critique of those schools). It's not like doing what social science has been doing, only harder, will somehow change the results we're seeing.

First of all, yes, I do! History has proven that people have been consistently wrong in thinking about policy. At some point, the country was against miscegenation as a policy, for goodness's sake. Why wouldn't they be now?

You assume a neat synchronous relationship between knowledge and policies, so lack of knowledge is the reason why good policies are not being enacted, and thus if bad policies are being enacted, then the current forms of knowledge practice (what you call "social science") are inadequate.

I don't think you've considered the results of politics, of democracy, of messy decision-making practices between groups of people who disagree. This often manifests as a temporal delay.

Take climate change, for example. Every sane scientist in the field agrees that climate change is a real thing. Yet there are not enough policies actively addressing the issue.

Couldn't I say: "You think the reason people haven't addressed climate change is because they haven't considered what the major factors correlating with future climate change are? They must have since they went to good schools, so if there aren't policies, then it must be that our current science about climate change is bunk."

[1] https://en.wikipedia.org/wiki/Cybernetics

---

> "Politics is the Mind-Killer"

Politics is endlessly fascinating to me; if you want a merger between politics-thinking and statistics or modeling, The Parable of the Polygons (http://ncase.me/polygons/) (which is a model of the Schelling Segregation Model) is an interesting project.

Re:The Politics of Pockets by dlss on Wednesday September 28, 2016 @08:44PM (#12602527)
> I'm familiar with Givewell. As a non-profit, Givewell does good work, but they perform a form of meta analysis of other non-profits, and do not do direct work. As a result, Givewell isn't a good example of "real world action",

While I do agree that decisions under time pressure take on a different character than more deliberative decisions, I disagree with the implication that they are somehow fundamentally different than performing day-to-day construction coordination tasks. The complexity level that the givewell guys grapple with when formulating their models is very high (ie finding a simple model that's not missing the point is a lot harder than using such a model after the fact).

> What's more important is a holistic (or cybernetics[1]) way to understand a series of entangled questions - why people use space, how they interact with a building, how a structure is constructed, what technologies are used, construction time and cost, risk, relationship to building codes, building inspectors et cetera.

(Other than the non building specific parts) these are exactly the kinds of questions givewell grapples with -- time and cost, risk: shifting disease patterns, unstable governments in the third world, estimated future efficacy, capital saturation, etc.

> Or take it this way. Logic-based thinking works in a highly controlled environment, where the logical axioms are pure and easy. Heuristic-based action works in a highly complex environment, where there are no axioms, rules are always bent, and things are always interconnected in highly unpredictable ways.

Right, statistics (well, probability theory) is an extension of logic to the sorts of highly complex environments you're describing.

>So to me, a form of thought that thinks asking "why" is "insanity or cult formation" is very well-meaning but ill-informed in terms of much forms of action. It works on the screen, but it doesn't work when it interfaces with complexity.

Perhaps you were meaning something different than I was with regard to why questions? I was referring to what OP does in her essay, and what you evoked when you asked for with "> women not wanting to buy those pants. And why do you think that is?"

Ie a non-tested (and often non-testable <- a much worse sin) speculation not backed by evidence with the feeling of truth. A pet theory in an area I know relatively little about for the purpose of seeming smart or interesting in conversation. A unnecessary heuristic to cloud my future self's vision. An unwarranted and unnecessary generalization about a group of people. A new religion to join.

To speculate accurately about the reason for one person's actions may sometimes be possible. I'll say again that I don't believe this can be done with any accuracy in the area of societal level women's fashion choices. There are simply too many women, buying pants for too many reasons. We can make aggregate predictions about that group much like an actuary would price a life insurance contract, but that deals with the what of a situation, not the why. You can even correctly infer the structure of the what (ie "cigarette smokers should be charged more for life insurance"), however you must be very careful when doing this, as this is extremely difficult to do without experimentation.

> So to clarify, your argument is that spending time on 'why' is "at best wasteful", and thus "what"/"how" is better?

> I think we agree here, but I'd say that it seems like you're in disagreement with your previous answer.

Well good! Glad to meet a sane person such as yourself. I have not read SSR myself (it's social science nonsense / overgeneralization in my opinion), but I'm glad to hear that my mentioning it seems to have been helpful. I've gotten the cliffs notes by spending time on the internet.

I think the root of our disagreement was I thought my replies where being taken in the context of the article (ie "I think pockets are political because history"), whereas I think perhaps you've merely been responding to my replies (which I wouldn't blame you for -- it's an awful article). If you could be more specific with a question re: consistency I'd be happy if I discovered something I've missed, and perhaps you'd be happy to discover something you'd missed.

> You assume a neat synchronous relationship between knowledge and policies, so lack of knowledge is the reason why good policies are not being enacted, and thus if bad policies are being enacted, then the current forms of knowledge practice (what you call "social science") are inadequate.

I did no such thing. My model is that (1) misguided speculation by social scientists leads to (2) good sounding by ultimately wrong intervention ideas that are (3) turned into a religious doctrine by way of party politics and then (4) implemented to the great suffering of all who are governed by them. My belief is that this is the default for most new social policies. The data on unintended problems caused by policy decisions seems to back this model up.

In my previous reply you are referencing above, I was evoking this in the hopes of getting you to stop spending your time doing (1), on the grounds that it doesn't seem to help anyone and at worst hurts them. Being less ambitious in areas of great epistemological risk has a better base rate than the current methods. Perhaps I was reading too much into your comments though. I do know that OP was doing roughly #3 above.

> Take climate change, for example. Every sane scientist in the field agrees that climate change is a real thing. Yet there are not enough policies actively addressing the issue.

Yeah, I have no opinion on climate policy other than an inclination toward the status quo. Forecasting societal economic effects is very hard, smart people like Thiel have taken the con side, and environmental regulation looks like it's a death sentence for lots of people in the third world (though economic hardship. I point this out because a decision that involves killing actual living humans to save hypothetical future humans should require more evidence). AGI timelines also seem important here, as it's likely to either fix our climate issue or kill us all before the climate becomes a problem. (Note I do think the science is largely accurate, it's policy questions I'm referring to here).

>Politics is endlessly fascinating to me; if you want a merger between politics-thinking and statistics or modeling, The Parable of the Polygons (http://ncase.me/polygons/) (which is a model of the Schelling Segregation Model) is an interesting project.

I usually strongly dislike Vi's work, but my love of Schelling was enough to get me over the hump -- fun :)

I don't think the Mind-Killer interpretation is incompatible with "endlessly fascinating". The tribal influence on thought is too light to be felt until it's too heavy to be broken.

---

PS. Do you happen to be in the market for a good book club, and live near SF?

Re:The Politics of Pockets by Chris2048 on Tuesday September 27, 2016 @06:39AM (#12588522)
> I'm aware that there are many groups who claim to know why (god)

This is a complicated example. The fact that people make these claims is itself a phenomenon with complicated, physiological explanations. Hence the process of investigation interferes such as to confound the investigation.

Re:The Politics of Pockets by dlss on Tuesday September 27, 2016 @07:14AM (#12588669)
While I do agree it's a complicated example, I also feel like the phenomena of people claiming to understand why women in aggregate behave a certain way similarly confounds OP's investigation. So a complicated example for sure, but for fair reasons.
Re:The Politics of Pockets by Chris2048 on Tuesday September 27, 2016 @06:18AM (#12588452)
> History allows us to actually understand

I though this form of analysis (favored by Marx) was debunked?