Thursday, April 30, 2015


A week ago I got to hear Roy Baumeister give an overview seminar on his research on self-regulation. Super fascinating stuff, and he's a great speaker. On a graphic of this data that I know I've seen online but I can't find for the life of me at the moment, showing how frequently people experience various types of desires, how much they try to resist them, and how often they succeed:
Not everyone who reports having a desire and does not try to resist it actually acts on the desire, of course. Sometimes the store is closed or the other person says no or you just don't have a gun.
Anyway, I don't know much about this field so I'm going to ignorantly speculate on some things I found intriguing. We hear more and more (directly due to Baumeister's research and popular book Willpower) that willpower is like a muscle - it gets depleted by using it too much in the short run, but you can also make it stronger by exercising it over time. But how perfect is this analogy? If we stop exercising our willpower for awhile, either by giving in all the time or by not encountering any tempting scenarios, does our willpower muscle atrophy?

The other relevant thing Roy mentioned is that, apparently, people with good willpower succeed by being good at avoiding temptation. This sounds very true to me. Not that I have particularly great willpower, but certainly in my experience the best way to change behavior is to manipulate the choice itself rather than to choose differently. I'm never going to be a person who goes running before work everyday, so I'm a member of a climbing gym and the campus pool and go hiking and such instead, because I inherently enjoy those things. I'm also strongly inclined to work from home, but because I know that it's good to maintain an in -person presence in the department, I schedule meetings on different days of the week so I don't have any choice but to go in. I get really good work done at night and hate both going to sleep and waking up, so instead of expending a ton of mental energy trying to make myself do something different and then feeling guilty for inevitably failing, I avoid morning commitments and take naps sometimes and sleep really late on weekends. It works out.*

So I wonder, when is Roy measuring willpower as "success in not engaging in harmful activities" or as "success in saying no to tempting activities when they are presented as options"? I interpreted him to mean that these things are correlated, but perhaps I misunderstood, or perhaps the data don't make a distinction. If the muscle analogy holds, and if the former definition of willpower is in fact often achieved through temptation avoidance, I would except the 2nd definition to be negatively correlated with the first. That is, if you get good at avoiding tempting scenarios, wouldn't your willpower muscle start to atrophy?

I briefly asked him about this after the talk and gathered that it's an open question whether the willpower muscle can atrophy with lack of use. I'd love to know the answer to that and the rest.

*Maybe these aren't good examples of what Roy is talking about though, because they don't involve routines or good habits, which are supposed to help mitigate temptation by removing choice from the equation. I'm too stubborn and FOMO-plagued and flow-exploiting to stick to routines...

Wednesday, April 29, 2015

foster kittens

I've taken on a small interdisciplinary research project on the side. Initial results:

Sprout very sadly passed away, hence the shorter timeseries for her :( The two cosmonauts are clearly thriving though.

Sunday, April 26, 2015

programming with dependent types

Now for something solidly in the "other things that you have to be a particular type of eclectic nerd to enjoy clumped under the same heading" category of this blog...

Brisbane, of all things, has a fantastic functional programming meetup group that has talks every month. Matt Brecknell gave an incredibly fun two-part lecture on programming with dependent types in Agda. And shockingly, he totally pulled off the live-coded format.

If you wanna ride the equality-is-a-type-and-other-ridiculous-notions rollercoaster, watch here:

Part 1:

Part 2:

Thursday, April 23, 2015

MOOCs + brick and mortar

As I've said before, I love MOOCs. For many reasons. I'm not on the "MOOCs will take over education and replace traditional universities!" bandwagon yet*, but I think they're a fantastic new way to learn.

Anyway, it has been interesting teaching at UQ so far because everything is online. Every room automatically records lectures when they take place, both audio and projector capture, and automatically links them to their appropriate course websites. Assignments are turned in electronically and can be marked electronically. It's fantastic.**

I cannot possibly describe how unbelievably jealous I am of undergrads' experience with this setup. I would have gotten immeasurably more out of my undergrad education if I had had lecture videos to refer to***. And I know this for sure in part due to my experience with MOOCs.

The way I watch MOOC videos is: turn the speed up to about 1.7x, depending on the speaker. Pause when there is a slide with technical content or a lot of words on it, read/process it, unpause. Go back and rewatch bits as desired/needed. Only go on to the next video when you've fully understood the current one. Come back months later when you want to use them as a reference.

Compare this to the typical lecture experience where the overall class is almost always too slow-paced, but sporadic bits are too fast, and/or sporadically you get lost in thought and tune out for a minute, and after the first such incident you're lost for a big chunk of the lecture thereafter, perhaps the whole thing. And if you have a really inconsiderate teacher who both doesn't follow a textbook and doesn't provide their own comprehensive notes (see footnote 3 again), you're just SOL. Hopefully you can reconstruct what you need based on your own imperfect transcription of whatever incomplete mess of partially-legible stuff was written on the board!

I therefore have completely sympathy for the fact that only about a third of my large lecture class attends lectures. I'm frankly shocked it's that high. I'm very jealous of this MOOC-style videos + brick and mortar hybrid system.


*Recent relevant news: At MR, Tyler agrees with me that low retention rates are a weak argument against MOOCs and cites a new article showing why: among students who sign up for a course certification, completion rates are much higher. So yes, obviously the large fraction of people who sign up for MOOCs to browse a bit and try them out won't complete them at a high rate, but that says nothing about the minority who are serious about completing them.

**Actually this has been my experience in Australia in general - everything has been moved online wonderfully seamlessly compared to the U.S. I suspect it is one of the upsides to the lack of legal privacy protections.

***And if professors hadn't been able to get away with really unbelievably annoying crap like assigning a textbook that defines mathematical concepts in one way and uses one notation to derive some results, but then lecturing with different definitions and different notation to prove the equivalence with what the textbook used as definitions. This was by far the most frustrating thing about Caltech math classes, and if a professor at UQ tried it they would be slaughtered in their teaching evaluations.****

****Well of course they would also be slaughtered in their evaluations for being brutally hard, so I'm not defending evaluations as a metric for teaching or a tool for issuing guidelines for teaching. But at least they do force professors to be organized.*****

*****You're not alone, I'm also annoyed by how many footnote rants I can't help going on

Sunday, April 5, 2015

science journalism

This is one of those topics that depresses me so much I avoid any contact with it but this quote is too great:
I spent a frustrating thirty minutes stuck in traffic listening to an--admittedly hour-long--NPR show on the higher-than-average rate of suicides among those taking antidepressants without once hearing the moderator ask, “Could this correlation be due to the fact that those on antidepressants are depressed and thus more prone to suicide?” The side benefit of this episode was that my six-year-old son in the back seat got an impromptu lecture on the difference between correlation and causation, similar to that which I received from my mother at about the same age.
More here. (Link stolen from Chris Blattman's blog.)