Inclusivity is a Trade Off

[Content Note: talking about talking about sexual assault][not a typo]

A few years ago I was an extremely active member of a martial arts studio.  Martial arts has risks, and this school chose to take more than the bare minimum- sparring involved head shots and take downs.  I was willing to accept the risks of this with most people at the dojo, because I knew they were acting with my safety in mind and the risks were worth the benefits, but there was one guy, Snotlout, who did not pass that test.

An artist's rendering of Snotlout
Snotlout is a psuedonym

Where most people will aim to hit near your face, so a mistake means you get a tap, Snotlout aimed to hit you, and the failure mode was hitting your face really hard.  He took blind kicks at full power and blamed you for not getting out of the way.  He once flipped a child flat on their back and his first concern was letting us know how lightly he had touched them.

The school wouldn’t kick him out, wouldn’t even really place restrictions on him.  When I complained to the de facto leadership it was always redirected to what I could do to take care of myself, but when I did so (e.g. insisting on slow motion sparring), I got push back from other de facto leadership.  No one would kick him out or place the necessarily level of restrictions on him, apparently out of fear those restrictions might drive him to leave.  If I really pushed, the people who would talk to me about it would say that they or someone they loved was that dangerous when they came in, and they wanted to give that guy the same chance.  Which is a beautiful thought, except that I know at least three people for whom he was a contributing factor in leaving the school.  Where was the inclusivity for us?*

It's not only morally wrong that Astrid has to keep fighting off Snotlout, it's bad for the tribe's survival. Between the two of them, she's clearly the one to bet on.
Where would Berk be if Snotlout had driven out Astrid?

Last year Seattle Effective Altruists had a member who brought up sexual assault a lot, in ways that made it clear it was personally relevant to her.  This made me really uncomfortable, but I was aware of how often rape victims are silenced and how damaging that is, so I didn’t say anything.  What occurred to me much later was that statistically there was at least one other victim of sexual assault in room, probably more, and they might find also find it uncomfortable.**  The choice wasn’t “do I silence this rape victim or not?”, it was “who do I make/let be uncomfortable?”, even if I didn’t know who the other person was.  Obviously a trauma victim discussing work with personal meaning to them is in no way equivalent to a jackass endangering people’s safety in order to prove how awesome he is, but that is part of my point: even actions with very good motivations have costs.

Back to martial arts.  Notice that I said de facto leadership?  The problem wasn’t that someone calculated Snotlout vs. [me, the two people I know about, and unknown number of others he drove away] and chose him.  It was that no one did the calculation and no one was responsible for making sure it was done.  There wasn’t even anyone I could negotiate with to ensure my personal safety; a plan I worked out with one senior student would be publicly ridiculed by another.*** No one had ownership of student safety so there was no one to turn a pile of individual complaints into “wow, that guy is dangerous and we should do something”.  To this day I’m not sure why “we have to be welcoming” meant “you have to let him hit you in the face over and over”, and everyone I asked described it as a decision made by someone else.  I don’t even think this worked out particularly well for that guy, because while no one was willing to restrict him, a lot of people would have been happier if he just left, and it showed up in petty things like him never quite getting added to the parking mailing list.  Eventually, after driving out who knows how many people, he screwed up so badly the school had to put severe restrictions on him.  He never came back.

In the real world bullies rarely improve their behavior without seeing its consequences.
An unanticipated struggle to find parking rarely inspires the kind of self reflection that leads to redemption or dragon riding.

Back to EA.  My eventual solution to the “how to talk about rape” problem was to simultaneously ask the woman to tone down the sexual assault talk in meetings where it wasn’t relevant and host her own meeting on the topic.  Unfortunately she left for other reasons before I could implement this.  But if I’d had the chance, it only would have worked because I was empowered as an organizer to do both of those things.  If I’d approached her as a peer, the request to limit talk about sexual assault would have had less heft, and a dedicated meeting would have been a suggestion, not an offer.  But it probably wouldn’t even have gone that far, because it wouldn’t have felt like my place to do it.  That leaves the hypothetical assault victim that didn’t want to constantly hear about rape to defend themselves by approaching her directly, and possibly requiring they disclose their history to see results, which they shouldn’t have to do.  In order to be truly welcoming to them**** , someone had to proactively make the space safe.

There have been other, less fraught trade offs.  One person’s friendly debate is another’s attack, and a third person’s derail.   I think one of my major contributions to the group has been not the decisions we made on these (although those were awesome), but that we made decisions at all, and worked out how to implement them.

If you are an organizer, for EA or something else, these are my recommendations:

  • Have a small, identifiable group with whom the buck ultimately stops.  Individual meetings in Seattle are run mostly on a who-is-excited-about-this system, but there are three people explicitly in charge of the administrative stuff, including disputes.
  • Make explicit decisions about your norms, share them, and enforce them.
  • Explicit is not the same as fixed.  I’m extremely excited about our plans to experiment with different norms at specific meetings, even if some of the norms would make me miserable as a participant.  Not every meeting needs to be for every person.
  • There’s a fine line between overpreparing and sticking your head in the sand until something blows up.  Some of our best decisions are “let X keep going unless Y happens, and then figure out a plan.”

*See also: Geek Social Fallacy #1.

**Much later still I would learn I was right.

***10 minutes before the same guy started quoting The Gift of Fear on listening to your instincts, and specifically leaving situations where you felt afraid.  I walked out.

****Or people who were uncomfortable talking about sexual assault for other reasons, or people who just wanted to talk about the planned topic.

Links 6/5/15

The best argument I’ve read for friendly AI research.  The best counter argument I’ve heard is that my one friend who does AI research says there ain’t no way step 3 (superintelligent machines) actually happen.  But the math/CS/philosophy problems this is motivating seem like they’ll bear useful fruit even if that’s true, and given what we do spend money on, this doesn’t seem like a ridiculous use of a few million dollars.

Included entirely for the phrase “Deepak Chopra of skepticism” (fixed)

When your volunteers are racist.

When your volunteer supervisor wants to heal your inner child.

The failed economics of code reuse.

Heart Rate Variability

Heart Rate Variability is one of those things that has such an obvious meaning I feel dumb asking follow up questions, but is consistently used in ways that confuse me.  It sounds a lot like arrhythmia, which is bad, but The Willpower Instinct consistently refers to it as good.  Plus it refers to it changing in ways that must be measured instantaneously, but changes in variability have to be measured over time, right?

Here is what I have figured out: we (I) think of the heart as beating to A Rhythm, which is your BPM.  The rhythm can speed up or slow down, but it’s still a rhythm.  A deviation from that is an arrhythmia, which is Bad.  We (I) think this because the wikipedia article on sinus rhythm basically says it, and because the article on HRV implies it’s measured in five minute increments over 24 hours, which means it’s basically a measure of range.  But at least some of the time HRV refers to beat-to-beat variation, and it’s being measured in response to an immediate stimulus (although, maddeningly, no one specifies the time period).

HRV_300
four heart beats, separated by .859, .793, and .726 seconds.

Your parasympathetic (relaxed/restorative) nervous system sends signals to your heart to decrease your heart rate.  Your sympathetic (fight/flight/freeze) system sends signals to increase it.  High parasympathetic activity also seems to be associated with high variability, at least to a point.  My personal guess is that high variability indicates both systems are operating and interacting, while low variability indicates one has taken over, and that your body is somewhat biased towards a higher rate, so it takes a bigger push from the parasympathetic to get the rate down.  Having one system dominate is not always bad: when you are running from a tiger, your heart should beat as fast as possible and redirect blood from digestion and immune system to muscles.  And when you are truly safe, the parasympathetic tells your body it can safely pay off it’s technical debt.  But often having both, and being able to switch between the two, is useful.

There’s  a lot of data showing high heart rate variability is increased by known Good Things (meditation, exercise), and low HRV is associated with bad things (alcoholism, PTSD), but I don’t see hard enough data on causality that I’m confident of the direction.

Links 5/29/15

What happens when you wring out a wet cloth without gravity? (h/t: Sydney)

space program: justified.

Weaponized data: How the obsession with data has been hurting marginalized communities.  I really disliked this on my first read through, because what are you going to do, not base your decision on data? But then I realized that I was treating “data”” and “information” as synonymous, and the author was not.  I’m not clear on exactly how she’s defining “data” here, but I agree with most of her examples of bad data or data being misused, and I like most of her proposed solutions.  I also liked her post on trickle-down community engagement.

Worms maintain memories after being cryogenically frozen and revived.  This isn’t my area of expertise, but I noticed a few things:

  • The type of memory they are talking about is imprinting, which is probably more durable than human experiential memory.
  • They used two different freezing procedures, and one had an 80% death rate.  The other one had a 0% death rate.

But for a first study, this is pretty exciting.

Say it wrong first.

Things I’ve learned about academic work from playing Age of Empires.  This is interesting, because it’s all about combat, and what I learned from Age of Empires is that fighting is boring and base building is fun.   Actually now that I think about it that’s  also what I learned from Hardcore History’s series on World War I.  That’s why I play Colonization: there’s a nice long period of supply chain building before you have to do the stupid war, and you can just quit then.

The world of estranged parents forums.  Narcissists kind enough to congregate and share their thought patterns.

GiveWell Summer Fellowship.

Rehighjacking the dopamine system

One of the contentions of The Willpower Instinct is that a lot of things (hyperpalatable foods, lottery tickets, facebook…) are hijacking the system that incentivizes you to be productive*, making you think feel that if you do this one more thing, you will achieve an important goal (nutrition, money, social connection). You can try to stare it down, but what you are really fighting is not the specific action, but the goal, and everything in you believes that goal is good, so it’s an expensive battle you will ultimately lose.

Recently I’ve been experimenting with hijacking back.  Let’s use food as an example.  I noticed that I would feel a very strong drive to eat candy, but eating it alleviated the urge for less than a second.  There was no amount of candy that would move me to a not-wanting-candy state.  Once I had that perspective, I could use higher brain powers to figure out what would  actually scratch that itch.  Moreover, I could use the urgency generated by the proximity of candy to power making real food with actual nutritional value.   And the dopamine system couldn’t really do anything about it, because it had sold candy to my body as a means to an end.

It works for video games too.  Sometimes I play video games because they are fun.  Sometimes I play them because I need the feeling of Doing a Thing, and doing actual things will involve an unpleasant intermediate steps.  Of course video games can’t actually make me feel productive, but the reward system tells me that means I need to play harder.

onemoreturn

This is a difficult urge to fight.  But if I pull back and notice doing the thing is not making the urge to do the thing lessen, it becomes obvious I need to do the unpleasant intermediate steps in order to get what I want.

*This is identified only as “dopamine”, but dopamine is a neurotransmitter, so she must mean dopamine in a particular part of the brain.  For simplicity I’ll keep referring to it as the dopamine system.

Dopamine: the tension builder

This is Huey.  He has not met you yet, but he loves you, and he wants you to love him too.  Almost all dogs are human-oriented, but Huey takes it to the next level.  He has been known to hyperventilate with happiness when a friend arrives- and that’s after he’s been at home all day with his work-from-home owner, so it’s not loneliness or separation anxiety, he is just that happy to see you.

IMG_20150522_222451
Pictured: a dog that would really like to be told he is good.

Sometimes Huey’s desire for affection puts him at odds with himself.  For example if you tell him to stay and walk away, he wants to do what you said, but he also wants to be near you.  He will literally vibrate with the effort it takes to stay in place.  But once you tell him good puppy, all the strain goes out of him.  He rushes over for his well deserved affection, and then he’s done.  He might even go play with a chew for a bit.

If you’d like a more human example, consider the orgasm.  Most humans on the edge of orgasm will do anything to finish it.  Immediately post orgasm, they will do absolutely nothing.  Even people who have multiple orgasms will usually have a final one after which they really don’t care.

According to The Willpower Instinct, dopamine release corresponds to the vibrating-with-effort phase, not the good puppy phase.  Dopamine doesn’t mark pleasure, because pleasure doesn’t need to be marked.  Dopamine tells you pleasure is just around the corner, so keep trying.  Note that this is slightly different than Peter Redgrave’s timestamp hypothesis I talked about before, but either would account for a lot of available data.  For example, people with ADHD (who are energetic and yet somehow unable to get themselves to do the thing to get the results they want) are on average low in dopamine, and almost all ADHD treatments increase dopamine availability*.  Dopamine levels rise copiously during sex until orgasm, at which point they plummet.  Dopamine is also heavily involved in addiction.

When asked for comment on his dopamine levels during training, Huey rolled over and requested tummy rubs.
When asked about his dopamine levels, Huey asked which answer would make me love him the most.

This makes slightly more sense to me than the timestamp hypothesis, but is less interesting.  Normally interesting is a bad sign for a hypothesis (almost anything is more interesting than chance, and chance explains a lot), but I’m not sure that applies in this case.  The timestamp hypothesis is a lot more specific and thus testable, the priming hypothesis seems sort of vague in comparison.  And of course both could be true- dopamine simultaneously says “reward coming soon” and “reward is because of this”.    Those actually make sense to go together, as opposed to some things our body has combined.

I’m never going to be able to conclusively prove one or the other sitting at my computer, but let’s talk about how they differ.  Timestamp hypothesis suggests people with low dopamine will be attracted to things that make it is easy to distinguish what caused a reward, like video games and obsessive facebooking.  Based on my friends and reddit, those are heavily associated with both depression and ADHD.  My literature search turns up some support for this**, but nothing with a design I consider rigorous.

Timestamp covers the H in ADHD better (a low baseline makes a small increase more distracting than it should be), but priming seems more apt for ADHD- inattentive type (ADHD minus H) and depression, where nothing is attractive enough to pull the person out of their focal activity.

So the answer is “some of both plus probably some other stuff”.  But variations in which of these dominate might explain why a given person reacts to low dopamine the way they do.

*In addition to changing other neurotransmitters and hormones.  The brain is complicated and no one understands it.

**Most interestingly the effect of bupropion in treating video game addiction.

Links 5/22/15

Effective Social Justice Interventions: this is a great example of using EA as a technique to address areas the EA-as-philosophy sphere hasn’t touched.

The Last Day of Her Life:  a psychology researcher’s decision to and process of ending her life as her Alzheimer’s progresses.   Fun fact: state-sanctioned euthanasia requires you be mentally competent and have less than six months to live.  Alzheimer’s patients are mentally incompetent years before they die of the disease.

The (crime-related) Broken Window Theory states that low level visible crime (graffiti, litter) leads to more crime, of all varieties. It is most famous for being Rudy Guilani’s method for reducing crime in New York City.  My understanding was that that had been debunked, and NYC’s drop is crime was caused mostly by demographic trends.  But some researchers did some fairly rigorous tests of it and it held up.  Caveat: they tested visible crime’s evidence on other crimes of similar magnitude, not escalations like theft.

This week’s “beautiful theory killed by an ugly gang of facts” award goes to the meditation chapter of The Willpower Instinct, which promises fantastic benefits from the very beginning.  In fact it says that meditating badly is in some ways better for you than meditating well, because it is the practice of refocusing yourself after you become distracted that is so beneficial.  Unfortunately none of the studies cited show that exact things, and what they do show is a small effect on a noisy variable, in a small sample.

[I don’t want to be too hard on The Willpower Instinct.  It encourages you to do your own experiments and stick with what works, I found some of it helpful, and it’s good for getting yourself into a willpower mindset.  It’s just scientifically weaker than it would have you believe.]

Sine Rider: if xkcd was a video game.

Soylent + Blender Bottle = Awesome

Soylent should be more convenient than real food, but the world is set up for real food, so sometimes it is not.  You’re not supposed to carry it warm for hours, but it’s not water-soluble enough to just mix by shaking, or even with a spoon.  You get clumps and those protective cysts of dry powder on the sides of the glass.  Enter Blender Bottle, a normal water bottle with a whisk ball inside it.  As long as you don’t overfill it it mixes extremely well.  Not quite immersion blender well, but it’s pretty homogeneous.  And you can do it anywhere.

Technically you could accomplish this with any water bottle + the whisk ball.  The advantage of the Blender Bottle is 1.  not every water bottle can stand up to that amount of shaking and 2.  the Pro Stak System.  You can buy little ~tupperware containers that interlock with the water bottle.  Unfortunately they don’t hold a useful amount of soylent, I use a ziploc bag for those.  But they are big enough to carry around salt.  I always drink water with salt.  Nothing was stopping me from carrying a salt shaker with me before, but I didn’t, so I didn’t drink enough water out of the house.  Apparently I will carry an attachment to my water bottle filled with salt.  So the blender bottle has solved two problems for me.

Credit to Brian for recommending a version of the Blender Bottle to me.

Bupropion finds its real family

When it comes to anti-depressants, there’s SSRIs, tricyclics, MAOIs, and… bupropion*.  I always wondered why it was that every other anti-depressant** came in a variety of forms, but there was only one bupropion.

Bupropion is metabolized into several different compounds in your body, the most important being hydroxybupropion, which is a norepinephrine-selective reuptake inhibitor***, meaning it causes the neurotransmitter norepiniephrine  to hang around longer so your nerves experience more of it.  “Selective” means it doesn’t affect all receptors equally, but is otherwise spectacularly uninformative.  Bupropion itself and several lesser metabolites also increase dopamine availability, but it’s not clear if that happens to any measurable degree in humans in the doses we take.  So we assume the effects on norepinephrine is the major reason bupropion works, but no one really has idea what it is going on in the brain so we can’t be sure.

Turns out there is another NSRI available, and it is actually quite well marketed. Just not for depression.  Atomoxetine is better known as Strattera, one of very few non-stimulant treatments for ADHD.  Its effects are not identical to bupropion, but they’re pretty similar, and there are studies showing atomoxetine is useful for treating depression.  Why then does no one market atomoexetine as an antidepressant, or bupropion as an anti-ADHD treatment (even though it’s shown promise)?  I don’t want to jump to the worst possible explanation, but the FDA requires new trials for every usage of a drug, and crazymeds.us is not wrong when it notes there is more money in treating ADHD and competing with very heavily controlled drugs that work very differently than there is in treating depression and competing with many off-patent drugs, one of which works in a very similar way.  And now that’s its been approved for something, any doctor who wants can still prescribe it for depression.  Meanwhile bupropion went off patent before ADHD was really a thing, so no one has any incentive to pay for additional testing now.

There are a number of other NRIs and NSRIs, few of which made it to the US.  But I hope bupropion takes comfort in knowing it does have a family, even if they spell their name differently.

*Brand name Wellbutrin when marketed as an anti-depressant, Zyban when marketed as an anti-smoking aid.

**Until you get to the really weird stuff they use for treatment-resistant depression.  Every atypical antipsychotic is its own little snowflake.

***The more common name for these is selective norepinephrine reuptake inhibitor, but that has the same acronym as serotonin norepinephrine reuptake inhibitor, a different kind of drug, so I use NSRI to avoid confusion.

Review: Mindset (Carol Dweck)

I went into this pretty skeptical, based on Scott Alexander’s analysis of the science.  But the reality was worse than I imagined.  First, she never even defines terms like talent or ability.  I would use ability to mean “current level of performance” and talent to mean something like “innate propensity to excel at task, as manifested in initial ability, ease of learning, or ceiling on ability.”  She… maybe uses ability to mean both those things?  She’ll talk about initial ability or talent and then increased ability or talent after practice, but that doesn’t mean the same amount of effort will get everyone to the same place, or all places are reachable by all people.  For that matter, she never defines mindset.  She talks about it like a fairly fixed trait (meaning it stays constant from one situation to another), but her own studies show it being changed by a four second speech.

Second, you can’t just make a list of good things and a list of bad things and wrap all the good things under your label and bad things under it’s opposite.  Here is a list of statements I believe will be uncontroversial:

  • A person who treats failure as a learning opportunity will learn more and be happier than a person who treats it as a mandate to curl into a ball and cry.
  • Ditto for viewing feedback as a source of information, rather than a referendum on you as a person.
  • Sometimes people start out bad at a task, practice, and then get really good at it.
  • This is more likely to happen if the person believes practice can improve their skill.
  • Children (and probably all people) tend to do better when their successes are ascribed to something they can control than to forces outside their control.

These things don’t necessarily go together.  For example, it is entirely possible to believe almost anything is learnable, and then beat yourself up for failure because you should have learned it already.   I’ve seen me do it.  “I’ll do better next time” can just as easily become a mantra to avoid mindfulness as to encourage it.

Third, I can’t even with the chapter on corporations.  Jack Welch brought stack ranking aka “rank and yank” to the masses, and she uses him as an example of not only having growth mindset, but fostering it throughout his company.


[An artist’s rendering of working at GE]

After this I refused to trust her anecdotes, and Scott already took down the studies.   You might think that left the book with nothing, but surprisingly it didn’t.  Her descriptions of the individual facets of growth or fixed mindset and how they affect people were useful and informative, even if I don’t think they have anything to do with each other.  And I think growth vs. fixed mindset might actually be a useful schema for institutions.  It certainly captures a lot of what’s wrong with American schools.

And as inspirational reading, it’s pretty good.  I would love to live in a world where one determined teacher takes 40 students from illiterate to Shakespeare, and stereotype threat is countered with a short speech.  In a world that overvalues innate talent, a push too far in the other direction may still leave us better off.  But that doesn’t make it correct.