#diaogue prompts | Explore Tumblr Posts and Blogs

incoherentham · 7 years

Text

Updates - Sep 2017

I have a deck of flashcards where I put all of the factoids or bits of advice I’ve changed my mind about. Every so often, I go through the deck and write a summary of what I've changed my mind about. This is that summary.

Nonviolent Communication (NVC)

"Rude", "Waste", "Deserving", "Responsibility", and "Goodness" are best categorized as Evaluation/Judgements.

"Stress", "Calm", "Confusion", "Hurt", and "Want" are best categorized as Emotions

I don't really endorse NVC in the same way I used to. How to Talk so Kids Will Listen and Listen so Kids Will Talk gives a more succinct and accurate summation of the kind of insights that NVC should give you. (I have also heard good things about Crucial Conversations.) These cards in particular exemplify my worst takeaway from NVC: Evaluating how meticulously my monologue separates information from implication instead of opening up and keeping up a dialogue.

Power struggles, hurt feelings, and wasted time are signs that you need to do better

Vague prompt with a vague passive-aggressive answer. In truth, not everything is about you.

Replace "I have to" with "I choose to"

Abusers can use this easily by creating false dichotomies. It's genuinely useful in a narrow window of circumstance, when the speaker is self-actualizing enough to create their own options but has rigid thinking around them that they aren't communicating clearly.

Red Flag: you might have trouble making your argument specific or concrete because judgments and evaluations just don't gloss to observations

Holding this belief in my head left me feeling guilty about everything, because when I get stressed I just want bad things to stop happening and don't have the spare mental space to carefully pull this impulse into its constituent pieces. A commitment to using NVC favors those who can consider their wording at leisure over those experiencing an active harm they need stopped. I consider this a major failing of the framework.

Negatively phrased requests offer no understanding of the right thing to do, and form ironic processes instead.

I feel very confused about what I need, I mostly request an ongoing conversation to figure it out because I don't have an active recommendation to give. Aiming to create a coherent plea before asking for help makes it so much harder; I call reserving the right to say "don't, at least for now" having a safe word.

If you find yourself approaching with the objective to change people and their behavior, pause. Go back to your needs, and ask for empathy and ideas to satisfy them.

There's some value in using this as a framework to draw out why others are trying to change me or my behavior. Trying to analyze myself in this way mostly made me avoid thinking about other people's behaviors as changeable.

A flat statement after feelings have been expressed leaves it unclear to others what we want them to do. Ask after underlying needs.

Sometimes making a flat statement is the only way I can get words out. It requires almost tricking myself into believing that what I say just reads into an obscure log file that no one ever looks at. Log files don't have opinions on what the admin should do, they only report what they detected. Expecting that any particular statement will invite a diaogue about root causes can throw me into enough panic that my mouth refuses to produce anything.

(I'm gonna stop adding NVC things now, even though that's not all of them. Suffice to say it was an excellent tool to beat myself up with.)

Other communication frames

Priming is the phenomenon in which exposure to one concept makes related concepts more easily accessible

Priming studies didn't replicate

When you expose people to info that changes their attitudes in a way they don't approve of, they may consciously cancel out the unwanted influence in the moment and shift beliefs later when their guard is down. (Sleeper effect)

I am more suspicious of this claim since the backfire effect failed to replicate. I think it points to a real pattern that exists, where people will stick to their rhetorical position while an argument is going and reexamine their beliefs based on evidence presented later; I doubt that it is basic or pervasive enough to be called out as a human bias.

Embodied cognition: increase persuasion with hand gestures, happiness with a smile, empathy by mimicry

I am slightly more suspicious of these kind of claims since the power poses failed to replicate. They check out intuitively - I use smiles as positive reification in meditation, hand gestures are very communicative, and incorporating emotion mirroring has had THE BIGGEST EFFECT SIZE on my casual social interaction by far - but on their own these ideas do not quite prescribe action.

A memorized Hook, Line, Sinker sequence to leave interesting first impressions on people

didn't drop into an introduction naturally, lacked confidence to force it out

Persuasive levers: Comfort, Acceptance, Greed, Ego, Drama

After careful consideration, I realized that coming at social skills from the angle of persuasian would backfire horrendously. People notice when you make decisions on more complicated considerations than "what would a good person do".

Spaced Repetition Technique

If I can give a partially right answer to a card, split it up into multiple cards.

My inability to tag consistently makes it hard to make connected groups of cards. It takes too much overhead to keep track of multiple cards for shaky concepts (i.e. ones I may want to change or correct later). Given that the majority of my cards are now secretly Implementation Intentions I find it more useful to change a card to be concrete & specific than to split it up.

Note: Upon reflection, I moved this card into my 'advice' deck. It's still a useful suggestion for people using spaced repetition like traditional flash cards.

If I think I made a bunch of cards with the same style before, pause and look them up.

It's useful to keep question phrasing consistent. If similar questions are phrased differently, you can start guessing the answer based on quirks of phrasing which are irrelevant to the question itself. But as I said above, it's too much overhead to track card by card. Instead, use card templates.

When designing the card answer, I need to understand what I need to understand what I need to know for this card.

So vague I have to explain the explanation, so it's useless for its original purpose. For my current SRS use case, it's flat out wrong — I have a couple decks that are designed to be a process involving spaced re-exposure, over which I expect my understanding to grow and change.

Elm and Elixir syntax

Programming syntax cards were useful to make: I had slow down and think about each piece of information. However, most of them were not very useful to review.

I should have cleared them out sooner. As long as I kept them, though, I could imagine myself as a Virtuous DoGooder who offered free basic syntax decks to any aspiring programmer who needed them.

That was way off. The correct thing to do as a beginner is not to solicit a premade deck, but to read the freaking docs and make your own reference cards. I won't send anyone my small hodge-podge of "stuff I don't want to look up on stackoverflow again" until they've done that first.

Habits

Habit knowledge is encoded in procedural memory

What is procedural memory exactly? The kind of memory that habit knowledge is encoded in, I guess. This is an isolated fact. I can say the words but I don't know anything significant about the world.

According to Wood and Neal's Habit Change manuscript, health interventions based on knowledge/intentions do not work to change long-term behavior. They don't consider how to supplant habits.

The framing of it rubs me the wrong way.

1) Aiming to change behavior: your body operates by a mostly-self-correcting system that shouldn't be trivially override-able every time you are convinced of a brilliant scheme to muck with it. If all it takes is a persuasive argument for you to overturn your lifestyle, then I have a church membership to sell to you.

2) I think there do exist people who will reliably change their long-term behavior based on knowledge/intentions (and who are not gullible yuppies). At least if you know what interventions work, you can maybe intervene on yourself. I think there's a deeper skill involved, though, based on combining your abstract reasoning and other systems cleanly into a greater system. The pieces are all there, but they're often just a little too broken to work. You need a trustworthy filter for good ideas, reliable translation from abstract to instinctual terms, a functional reinforcement system at all, a way to reason well about blackbox algorithms, explicit data collection AND lots of up-close exposure to the real world. we need more of it.

People with high self-control traits attain goals by forming habits that allow them to achieve goals without experiencing unwanted temptations (Wendy Wood's Psychology of Habit)

This is a very incomplete model. You cannot engineer your environment to do all your agency for you, though I certainly tried. It will not work for people who are just missing, or maybe misapplying, a certain kind of forward drive.

This is a fundamental issue I have with CFAR. It really helps some people to be told that their brain has internal mechanics that can be altered in such and such ways if they're giving you problems. Being told how to tinker does not teach me how to build a working system from spare, half-functioning parts. They're assuming a crucial centerpiece of the system without teaching it, so it's confusing when all the pieces check out but the engine still doesn't run.

Stimulus reinforcement a good way to control our behavior because automatic systems run effortlessly (unlike willpower, which breaks down)

Man I don't even remember what specific thing 'stimulus reinforcement' means here. As for my best guess, I think any attempt at naive reinforcement on yourself will have either be too narrow or have lots of weird side effects. This is because most people are way smarter than their model of themselves. You will hit goodhart's law sooner than you think.

Trainees put on a continuous reinforcement schedule for along time are likely to quit upon their first failure, so training trials should contain many successes mixed with occasional failures.

This is probably a good first approximation of what works to tell someone, such that if they try to implement it they will probably get much closer to the ideal than they otherwise would have. I don't think it's strictly true though... You get the most out of training at the edge of the learner's ability, and one of the things you expect to see at the correct level is that the learner neither succeeds effortlessly nor fails all the time. I suspect there's a thing about the kinds of distributions of reward in the real world that makes quitting at the first failure in a long range of successes a fairly rational tactic. You may actually not want to train people out of this tendency if you could.

Training trials should contain many successes mixed with occasional failures to increase resistance to extinction. (trainees on continuous reinforcement schedules are likely to quit upon the first failure.)

I think intentionally mixing in failures is wrong in principle, but "occasional failures" as typically measured is a proxy for something that matters. It's important to train variations, multiple contexts; if you are setting thing up precisely enough to prevent all failures then you may be training something much narrower than you think you are, and it will not continue outside of the narrow context. You stay more motivated as well when you train at the edge of your ability, which appears from the outside as having many successes mixed with occasional failures.

You halt a positive feedback loop by matching its most common/powerful triggers to actions that cut off its source material. You halt a negative feedback loop by matching the most common/powerful triggers to actions with neutral or positive feedback.

A masterfully crafted technically correct answer. All the confusing concrete details have been abstracted away, and as such I have no idea how to apply it to real world situations. How do I recognize what kind of loop some phenomena is? How I determine its common or powerful triggers? How do I identify its source material, and what kind of feedback a certain action gives? lol i dunno

Yearly/Weekly task list to check off

Weeks don't map to anything in my brain, so actions never become habitual. I find it hard not to find weekly lists arbitrary and resent the imposition from my past self. Some tasks sincerely need my attention at regular intervals to accomplish, and I'm trying to find other ways to create consistency on those. Yearly only listed prescriptive measures while not actually having a comprehensive enough spread to plan my year around.

Personal Habits

Habit: When I am bursting with irritation, locate the nearest person to complain at them.

It seemed an important to do something to acknowledge that screaming silently hurts me and prove to myself that I am worth defending. I hoped that triggering this action with the initial frustration instead of the later despair would catch problems before they snowballed. However, trust in others needs to be built incrementally. By the time I'm irritated enough to have a rant I know I'm angry enough to do real damage in suddenly letting it out at someone.

Frames that have worked better for me lately include complaining into a list (so I can deal with one piece at a time) and seeking reassurance.

You can cement appropriate response protocols in people's heads by having them come up with scenarios individually, solve the scenarios in pairs, and talk about the solutions as a group.

I decided this entirely on the apparent success of it being used on a group of EA Global 2016 volunteers. That's not nearly enough data, I know. I think this strategy approximates correct training technique better than a lecture does, but is probably not ideal. It certainly depends too much on the appropriate thing to do being transparently recognizable as such.

Morning routine: get dressed, take pills, eat cereal

My actual routine tends to be more adaptable than the cards I make about it.

Also right now I'm doing a thing where I don't do routines, I do what I want.

"every time i've gone for a walk in the sun on a quiet street, i've felt a little better, even when i was depressed, and i would bet money on that happening again if i did it now despite System 1's insistence otherwise."

Many things have slight improvements on mood while depressed, but I get sick very quickly of trying to fill the grand canyon one penny at a time.

if my empathy is failing me take 5-htp

need a more well-rounded theory of drugs, as I don't trust throwing contextless TAPs at problems to form effective solutions when there's interactions and tolerance in the picture.

Use one or more hard trigger for positive thinking: context switch, access to food, make more tools in-reach, be heard out by people, exercise, sleep, sunlight

Need to want to feel better or I will avoid these things anyways. Conflicts with several other cards attempting to enumerate exactly the same things.

Track in phone when I use happy songs, loving-kindness meditation, 5-htp, or naps to deal with feeling awful

Writing down in phone marks these as deliberate interventions, which requires me to, before doing things that might help, a) notice that I'm feeling awful, b) admit to myself that I'm feeling awful, c) admit I don't want to be feeling awful, and d) believe it (I'm?) worth the effort to try.

Keep count of movement and happies vs. flinches to break helpless spirals

tends to make me think too short-term to solve systemic problems

Keep count of noticing detail vs. shutting down, or positive bids vs. disengagement while socializing

I kind of need my attention spare when socializing. Also too short-term to help me solve underlying causes.

Keep count of sensory details vs. noticed confusions vs. hypotheses while debugging or PCK-seeking

Definitely had to write them down when considering people's complex problems and worried it made me look distracted or judgmental.

Keep count of intuitions vs. outcomes when finding The Most Important Thing

Continually trying to figure out the Most Important Thing to do tired me out quickly.

Computers

Your data type design is not done until it has a function template giving the core structure needed to interact with it.

I have not created an ordering to how I do data type design, meaning there's not really anywhere in the process I can stick an "after this, write a function template". I'm sorry, but you really can't make effective 'before' trigger action plans. At best you can do 'when I think about starting , run through this list of prerequisites', and the pain of extra overhead often causes me to avoid thinking about .

Cmd [ or ] to move between tabs

Doesn't work in any Mac program I've tried.

Good Machine Learning Scaling: Get every feature in approximately a -1 to 1 range

For what kind of machine learning is an important question, before I'm pretty sure some kinds actually require 0 to 1 scaling. This is also the most vanilla(1) kind of standardization I can think of; it's got nothing on batch norm. -(1) actually no, vanilla is a fine, upstanding flavor. this is more like, sparkling water flavored soda or cream-flavored ice cream.

Why did I bother, seriously why?

Thucydides said 'It is a habit of mankind to use sovereign reason to thrust aside what they do not fancy'

Why did I think I would ever need to know this quote. Even if it sounds nice to put in a paper or article, I should probably stick to quotes I know the context of.

Starbucks customer service method is to Listen, Acknowledge the complaint, Take action to solve it, Thank them, and Explain why it occured (LATTE)

This is a cute acronym for customer service protocol. There are TONS of cutesy acronyms for customer service protocols; this does not seem particularly notable.

cryptomnesia: when a memory returns and it appears to be a new thought

Obscure vocabulary for something I rarely observe. I guess it is kinda cool to know it's a thing in case it ever happens and someone's like "is this normal"? Except I don't actually know if it's common or a symptom of anything.

Memory encoding is linking new info to info structures already memory

which this card utterly failed to do

E.T. Jaynes says randomness is not a property of the world

Another out of context quote! I really do not know enough about Jayne's general philosophy that I should be confident quoting him as an authority.

For a CRDT to be safe, updates must be Commutative, Associative, and Idempotent.

Isolated factoid. I need a better idea of what a CRDT is, what it's used for, and/or when it would come up. Would also like a better grip on how to determine that updates satisfy the three properties. (I bet it's easy as pie to verify in Haskell.)

Grueling Level Grinding Attempts I Quickly Became Averse To

tensorflow and cuda everything

reading the code for my partner's startup

haskell everything

Kotlin

Other

The first step to rooting a phone is to install TWRP

Useless without the other steps, such that I'm not even sure I don't need to look up stuff first.

Is (X) safe for dogs?

I started this because I keep winding up in houses with dogs, and it seemed prudent to know some common food rules offhand. It wasn't very efficient to memorize foods when the dog was too well trained to steal food, I was too apathetic to feed it scraps, and the list was always too adhoc to generalize. I'll just google it if I'm concerned in the future.

Do Core Transformation technique all the time

For some reason I really resist diving into the depths of my deepest insecurities on a regular basis.

Maybe I should try core transformation on that resistance... Oh that's right, personal growth is useless work in that it doesn't give the tribe a reason to not kick me out.

Everyday math is a crucial skill, a la Eliezer's post "The Simple Math of Everything"

Trying to force math concepts and fast computation tricks into my head in the hopes that I caught something that happened to be applicable was a recipe for exhaustion. The very basics are critical for later learning efficacy, but how you get eager learners is by creating opportunities to see what math is useful and why, and what prerequisite concepts you need to grasp in order to use the math to its full extent. How you get mastery is use, and with only the flimsiest of toy examples to suggest where to apply skills of course I'm not going to use them.

Using the rote practice in a limited context way as the Default Way to learn math skills is ridiculous--you do targeted practice to address deficiencies that are holding you back. Maaybe you do multiple, varied step-throughs of the process in order to assess your ability. Ugh.

According to Haidt, we might be blind to our own selfishness because this helps us project a positive self-image

This is not a straight forward fact about humans. This is one of those corrections that gets taken most seriously by exactly the people who don't need it or need the exact opposite correction. It needs to include a concrete way to estimate its applicability.

When I notice I'm missing important information, that is NOT the time to push to extract feedback. My current mechanisms won't often find what they missed just by looking harder.

I've this unfortunate habit of panicking at the messenger, whenever I notice big holes in my understanding. This was an attempt to prevent making things worse in that way. Now, instead of supressing the instinct to "interrogate everyone"/"solve everything now", I've moved to integrating an understanding that the quick solution IS making a small update to longer term strategies.

Identify your Signature Strengths and take steps to use them more often

My self-image contradicts itself in ways I don't know how to resolve. I feel scared and confused when I try to sort out which beliefs reflect my real abilities, which ideas manifested out of depression or internalized verbal abuse, and which characterized myself in specific circumstances that don't apply to me anymore (or rarely).

Keep major decisions in LADR format

Not lightweight enough for me to use consistently, need a way of classifying which decisions merit recording this way. I might reconsider using them in my strategic reviews.

Lines of a general 101 introduction speech to memorize

I imagined I would give a lot of 101 talks which this might have been helpful to have pinned down. I have little reason to expect it works, having never tested it. It feels silly to fiddle around optimizing a hook when poor vocal projection and scattered thinking put much tighter bottlenecks on how good I get at giving talks.

Software development: For the next time I have a problem, I try to think of a few words that will quickly focus my thinking.

Vague

Someone operating on a delicate live system must know if there are any "points of no return" in the procedure, and stop to re-check the system.

Many things are delicate live systems, having this in my head encouraged me to obsess over decisions until deadlines even more.

As evidenced by people predicting music competition winners better with no sound, only visuals, style is more important than content.

Could it be that being voted a pop-star is better predicted by style than content? No reason to expect this generalizes to every part of life.

The difference between sensitivity and accuracy: sensitivity is % correct of positive examples, accuracy is % correct of all predictions.

The difference between sensitivity and precision: sensitivity is out of all real positives, precision is out of predicted positives

terms used too loosely or interchangeable in every day examples to be worth attaching precise data wrangling definitions

Anki checklist:

☐ I know why I want to make tis card. ☐ I understand what I need to know for this card ☐ I know how to recognize when/where I need this ☐ There are no almost or partially right answers to this card ☐ There are no technically correct answers which comically miss the point ☐ There is no way to cheat to the right answer ☐ I cannot get marked wrong for being smarter than the question writer

How I'm using anki cards changes often, and have subtler common qualities than this checklist

#update #anki

1 note · View note