March 19, 2016

An Experiment In Morning Blues

Hello friends!

If you're curious what I get up to with my time, observe! I ran this little self-study over the past couple of weeks. This post reports it using a hodgepodge style somewhere between personal notes and a scientific paper.

If you want to encourage the norm of testing everyday ideas empirically, or just do me a solid favor, please be my mini peer-review and take a moment to comment.


When I have a bad morning, the rest of the day's productivity seems to get shot and that feels bad. I set out to do something about it.

Question: What are the key factors in having a pleasant morning?

Hypothesis: Happiness will correlate with getting dressed and eating food and getting sunlight. I expect I can manage those things more easily when I'm in a good mood.

Method: For 15 mornings, I recorded when I did one of a set of things on the whiteboard. Collected "Happy, get sunlight, get dressed, get water, eat food, noise control". Looked for visual correlations. Checked straight predictive accuracy.


sorry blind people i'm lazy

Dashes in the first image mark a wholly blank day.

The second image is some summary data of the first.
Counts: True positive, false negatives, false positives, true negatives for each column.
Below that: Accuracy, precision, sensitivity (recall) for each column


  • Noise control may help a bit. Nothing else looked like an especially accurate predictor of mood.
  • Eating in the morning may anti-predict feeling good.
  • Getting a drink might be a sensitive enough mood test to use as a Trigger Action for exploiting them.
Sanity checks:
(This is a new section in my notes, added to encourage me to notice whether my ideas could be wrong in a way that would have take embarrassingly little effort to find out. If you can facilitate one of these sanity checks, please let me know!)

  • Ask someone who can Real Science(tm) to calculate my result's p-values.
    • Common sense-wise, I had 5 happy mornings recorded to work with. All of my conclusions are based on, like, a difference of 1 data point.
  • Check correlations between the variables themselves.
  • Take a mood test with one of's tools at several different times before 11am. If I get a wide variance and/or the empirical-ish results differ from my overall self-report then my results are probably bogus.
  • Search google scholar for studies about morning routines and read the abstracts
  • Test whether I get carb crashes. Food does have a plausible mechanism for causing a negative effect.
  • Ask autistic tumblr how they tweak their morning routines to prevent overload.
Potential Follow-ups:

  • Try this again when I'm not sick
  • Trigger Action Plan: With a drink in my hand, free associate thoughts about stuff I need to do.
  • Practice making/importing data into R
  • Spend a week or two collecting a more representative sample of things I do in the morning, run another self-study with those.

March 16, 2016

Anki Deck for Systematic Program Design

I want to know what skills I can work on to write elegant code. In particular, I want to know the secret sauce to making a comprehensive project I can show off to people

Methods: I searched an edX course's materials for some causal intuitions I could reinforce. (Background understandings, like "If I get X right then Y will go a little more smoothly".) I then stuck my insights into an Anki deck to be digested at a later date.

Results: Anki Deck on Google Drive (UPDATE: 4 months later Revised Deck)

Good cards require personalization. Results may vary.

Random sample 4/49 cards
Anki card: Why bother running code that you know doesn't work yet? ... to catch little mistakes before they build upA self-referential type must have at least two cases: a reference and a base case.Anki card: when should I first try to run my program? ... as early as possibleAnki Card: Function stub means... a dummy function that returns the right type value. e.g. def add_numbers(n1, n2): return 0
An aside: taking snapshots with the intent of making them public is a really good way to catch mistakes.

Sanity Checks:
Ask [programmer friend] to look over my cards
Try the following the "recipes" on a practice programming problem
Ask people on my blog for feedback

Observe how many cards I've deleted, marked, suspended, or edited in three months.
I hit my comprehension limit partway through, maybe read it again after I've reviewed the deck.

----UPDATE July 11, 2016:
4 months later, I still have 49 cards. I've modified 31 of them. Here is the Revised Deck.

March 15, 2016

What's a good metacognitive mirror for defeating self-loathing?

Someone asked that question on a forum I frequent, and I thought I'd share my answer here.

First, what is a metacognitive mirror?

People have bias. We sometimes have ulterior motives nudging our decisions in certain ways, a kind of bias we're all aware of, and our perception of the world also develops some quirks based on the way our brains work. (See Kahneman and Tversky)

For example: we tend to question evidence that confirms our beliefs less harshly than we question evidence that disproves them. Counteracting this can be as simple as pretending you started out believing the opposite, and asking yourself how you feel about the evidence then.

Some problems like this take more effort to root out than others. Some of them actively work against our ability to notice and overcome them.

For example, feeling angry.
A lot of heated discussions go much more smoothly as soon as one or both parties take a breather to calm down. But the angrier you feel, the less you notice the ways that your anger works against you. Sometimes you can be in the middle of an argument, shouting "I'M NOT ANGRY" and barely even notice the contradiction.

Another example, getting drunk
We know alcohol causes a slower reaction time and impulsive decision making. There's good reason not to drive or make important decisions while under the influence. Unfortunately, judging whether you're too impaired to do something is also a decision, and it's negatively affected as much as any other.

What's that about self-loathing?

When you hate yourself, you can twist *anything* that happens into evidence that you are an awful person. Any mistake is unforgivable, anything good could have been better. You develop very distorted beliefs about your own abilities, what other people think about you, and what is or isn't worth doing. An astute self-hater may even notice the distortion of their thinking and proceed to beat themselves up for not having fixed it.

A litmus test on par with breathalizers for alcohol: Look yourself in the eye in a mirror and say that you're a worthwhile person.

It's stupid and corny, but it's seriously just words. If you find yourself flinching away from a simple exercise like this then your judgment is likely affected in less immediately visible ways too. Other tells: avoiding eye contact, talking faster, touching self nervously, exaggerating your words or voice or movements, nitpicking over the technicalities of your claim, making excuses not to do it another time, etc. It's like Gendlin-style Focusing.

You can combine the mentality of this with body mindfulness; notice that your body exists and spread yourself out to physically lay claim to all the space you can afford yourself.

You can especially benefit from noticing why you aren't trying both of these *right now*.

March 7, 2016

Political campaigning: Some California Ballots

TLDR: I checked Ballotpedia, and can give soundbites but do not significantly less confused about politics than I was before.

I saw a poster advertising paid political campaigning and called in on a whim.
Well, 70% whim - 20% intentionally pushing my comfort zone - 10% pleasegivemeajobIwantmoney

Motivation: If something comes of the interview they set up tomorrow and I go campaigning for spare cash, I want to know what I'm pushing. I'm not okay with making the world a worse place because I couldn't be bothered to do even rudimentary fact-checking. I think plenty of people prefer to hear an honest case, and *I* prefer to not look like an ignorant idiot whenever someone asks me an off-script question.

What initiatives did the guy on the phone bring up, and do I support them?

Google keywords mentioned (Prop 40 education tax extension, Water bond 2016,  Tobacco tax to medical)
Find out that's not specific enough to find recent things, so look up recent stuff on whatever wiki-ish site that I find other propositions on.

the notes I took during the phone call
(That was a good call, grabbing that notebook at the last minute!)

Tax Extension to Fund Education and Healthcare Initiative
An Income tax increase in 2012 (on 250k bracket) gets 12 extra years, the funds from this increase go to schools and healthcare. Guy on phone suggested this is their focus

I think this likely fails to solve underlying problems with the education system and, dependent on how that tax bracket plays out it practice, possibly shits on entrepreneurs who *would* have helped solve the issues. I wonder where the money was going before.

Water Supply Reliability and Drought Protection Initiative
$6 billion in bonds for water supply infrastructure. Cost to local gov money maybe ~3x more than it saves. ($300 vs. $100 mill annual average)

Infrastructure is important I think. I'd vote yes and hope that someone else was paying attention to the implementation details.

Healthcare, Research and Prevention Tobacco Tax Amendment
Increase cigarette tax to $2/pack, with equiv. measures on other tobacco products and electronic cigarettes. Allocate revenues to healthcare and tobacco-related stuff. If biennial audit suggests this tax is decreasing tobacco use that other anti-tobacco programs rely on, transfers tax revenues to offset them. Supposedly a net revenue increase of $1-1.5 billion in next few years, decreasing in later years.

Generally for, but suspect diminishing returns on this sort of stuff. The offset worries me, not sure how the incentives shake out. Also curious what the consumer cost of cigarettes works up to nowadays. What programs ostensibly prevent taxes like this from just penalizing addicts cuz they're acceptable targets, and do they work?

Sanity Check:
Google if bonds are usually a good idea
How much of a marginal difference would a campaigner even make to the outcome, is it worth *refraining* from?
Ask [name] or [name] to research this shit
Reverse check - What are the biggest problems in California? Do these initiatives have any effect whatsoever on those problems or is it mostly opportunity cost at stake?
Have we tried talking to someone in the $250k bracket about their lifestyles and the taxes they pay?

Talk to the interview guy
Read the whole proposal texts with a highlighter and red pen
Look up proposition 40, try to find the connection to the tax extension

March 5, 2016

Goal setting

An incomplete list of stuff I'm up to

Alumni Pair Debugging
Some in my community express feelings of extreme loneliness and ineffectiveness at times. To counteract both of these, I started a service to pair people up regularly to troubleshoot their problems together. I engineered the system to stay stable and low-effort from the start so they could rely on its continuation; now I'd like to expand its scope. I will gather data on how people engage with the system now, experiment with ways to alter the service to better serve people's needs, and bring more members on board.

Make Anki Useful Again
Spaced repetition has a lot of evidence supporting its usefulness, and I made extensive Anki decks on hoping to capture that value. In time, I learned that to use that value I had to respect its limits; efficient memorization could not replace proper *understanding* and *prioritization*. I will clear out decks not relating to current or projected objectives and alter my current reading followup system to take advantage of the incremental reading plugin.

Trends and Disagreements in the Rationalist Community

Edited from my personal notes format again.

Question: What are the trends and disagreements in the rationality community this month?

Motivation: I want to know what's up, and I want to make a difference. Avoiding the valley of bad rationality seems useful. Also the community has a distinct motte-and-bailey going on about whether rationality means "whatever wins" or the flavor of the LW sequences and diaspora.

That's a fine thing to have, in one direction: we can talk about our latest and greatest ideas, play with those flavors, invite others to join in, and when we encounter disagreement we should rightly take a step back into our core principles to ask "Wait, is this actually what wins?"

But I sometimes see it happening the other way: using the "whatever wins" banner as an enticement to get others to join in, and use our supposed intentions as a general defense for our methods. Methods we aren't centralized enough to even all agree on.

Method: Look at some major blogs. Give a quick summary of each and try to name our felt sense about a) how authoritative they are and b) whether they agree with each other.

Sources: Lesswrong Main, Lesswrong Discussion

Answer: Lesswrong main: seems to navel-gaze lately. It rehashes. It struggles to curate itself. Latest posts are Meetups, trying to distill CFAR stuff, estimating the value of people in Effective Altruism, LW karma charts, speculating about medical conditions, gamifying goals.

LW discussion: I see people chatting about LW's decline, but what do they do about it? I see continued argument about the place for politics and productivity hacks.

AI fermi paradox speculation. They make a background assumption that FAI is hard, unsurprisingly.
AI safety. Using analogies to the past, notable inability to bridge inferential gaps with people out of the OB/LW sphere

High school advice: Get GED and skip to college, or leverage private school. Get college's large wealth and power boost. Meta advice: break down skills important to making choices in
general (value awareness, prediction accuracy, agency) and check that you're stacking your character.

Media to follow... gwern's recommendations well-received

PSA about air quality, gathering what-should-be-universal statistical literacy, trying to propagate lists of high-value PSAs in general

Disagreement: is altruistic deception necessary?

I'm nowhere close to done but I'm way over time.

LW is decidedly not the zeitgeist, and it knows it's not.

People are complicated, reality is complicated, it would be surprising to have succint summarizations.

Could I do a poll of what people read, what 'tenets' people work with? What such polls already exist?

Have I spent five minutes actually imagining what a group both coordinated and flexible enough to all change their minds on a dime would actually look like?
.... Damn, it would look like the mormon church.

Featured/Promoted: try to establish what canon is well-known at the moment

Refine what it is I'm looking for.

Old LW surveys, SSC surveys, see if gwern does surveys

This tumblr post by Scott Alexander about rationalist tumblr

March 2, 2016

I imply that you imply that I imply that you suck

I made a note on a Facebook event to warn people who have issues with pets and small children that this event is likely to include pets and small children. I tried to word it in a way that would get people to contact me to deal with their concern instead of just not going, or suffering in silence. I got a pretty strong backlash, paraphrased:
"It's worded like an offer to sequester babies away from the party. I feel uncomfortable with phrasings that diminish their personhood, or imply their existence in their own home is a burden on others."
Well, I'm uncomfortable with the implications of this objection!

Don't get me wrong, I understand problems with people considering me a burden to interact with. It's awful. I have a hard time believing in my own worth. I'm continually finding old scars in places I didn't know existed. I don't want to cause that for someone else.

I did reword the note. I wanted to make sure potential issues were resolved ahead of time instead of by last-minute ultimatum, and it wouldn't accomplish my purpose if it pissed off the housemates.

I still resent the implications of this objection.

It tries to provide a protection for personhood by outlining the zone where they cannot be a burden, because, by definition, it is neutral ground to meet all their needs in that zone instead of someone else's. In the process, it runs roughshod over all my attempts to protect my own personhood and respect others'.

When I hear others talk about burdens they seem to treat it as an absolute scale instead of a relative scale. Someone is or isn't thought of as a burden, with some gradation for "how" burdensome. The label clearly matters to people. Being a burden is supposed to be terrible. For the longest time, I thought had to do everything in my power to not be one.

It makes me want to laugh and cry and scream all in one moment when people try to tug on my heartstrings by pointing out how awful it would be if there was no safe, neutral ground to interact with people from. Regardless of whether an interaction is "worth it", regardless of whether I want the interaction or actively try to disengage, it almost always costs me something to interact with people. The most obvious cost is sanity: too much stimulation with no recovery time leads to mental overload and shut down, and this cost is very much additive. The idea of a person who does not come with a cost to burden is *ludicrous* in my experience, but everyone I knew told me it was an achievable standard and failing to meet it was monstrous.

Life started going much better when I decided to to minimize the costs on me and let other people negotiate the costs to them. My neutral ground is the ground I'm standing on. I cannot interact with people as people if they're going to insist that the costs they impose don't count and can't be negotiated. Those aren't agents, those are natural disasters.