LESSWRONG
LW

CstineSublime's Shortform — LessWrong

CstineSublime's Shortform

17th Feb 2024

1 min read

2

This is a special post for quick takes by CstineSublime. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.

CstineSublime's Shortform

5Alexander Gietelink Oldenziel

125 comments, sorted by

top scoring

Click to highlight new comments since: Today at 4:03 PM

Some comments are truncated due to high volume. (⌘F to expand all)Change truncation settings

[-]CstineSublime6mo181

gnōthi seauton. But how do you know how well you know yourself?

What kind of questions or theories can you test to confirm that your mental model of yourself is accurate?

And please PLEASE, DON'T ALLOW YOURSELF TO GET GASLIGHT!

Here are some off the top of my head:

Anticipating your Reaction:

Green Eggs and Ham Moments: how often have you been reluctantly obligated or involuntarily committed to trying something and are pleasantly surprised by how much you enjoy it?

The reverse: How often have you been excited to read a book, go to a conference, watch a television show and been disappointing? Especially if your anticipation of the contents was accurate but not your reaction?

Confirming your habits:

"I always X" - but do you - what are the last 5 times you did it? Similarly "I'm a great/good X or at [activity]" - what were your last 5 attempts at it like?

"I never Y" - okay, but do you never. And even if that's a hyperbole, how frequent would it have to be to consistent a habit? Similarly "I'm a hopeless/terrible X or at [activity]" - what were your last 5 attempts at it like?

How often has a loved one, a coworker told you smugly "I knew you'd do that" when you explicitly denied you woul... (read more)

3Viliam6mo

Please make an article out of this (maybe with more ideas and more specific examples). I believe this could be very useful. This is practically a lesson plan for rationality training.

2quetzal_rainbow6mo

I always struggle with such self-evaluation, that's why I thought of this experiment.

1CstineSublime6mo

I would be very interested to see how narrow or broad that range is between the groups.

[-]CstineSublime4mo60

Are rationalists any less susceptible to online scams than a random sample of the general population? I would hope that no one here has fallen for "hey it's me, Ozzy Osbourne, I'm not really dead, I'm just on the Crazy Train and I need $300USD..." unsolicited messages. But who knows, maybe there's some affinity scam "It's me Eliezer, I've picked you to help me fight bad AGI, can you send me $300USD..."

5silentbob4mo

When it comes to your average scam, I'm sure rationalists fall for it less than average. But you could surely come up with some very carefully crafted scam that targets rationalists in particular and has higher odds of convincing them than the general public. It also depends on what exactly you consider a scam. To some people, FTX was a scam, and rationalists almost certainly were overrepresented among its customers (or victims).

4Viliam4mo

Proof of concept (not very careful) -- unfortunately, we don't have the data about its success rate.

3CstineSublime4mo

Interesting to see that more than one comment had the sentiment "yeah it's a scam - but let's use this as an experiment."

1CstineSublime4mo

I was specifically thinking about fraudulent accounts and messages like the Ozzy Osbourne example or basically anything based on the Nigerian Prince Scam where there is it is highly unlikely that a dead celebrity or royalty is contacting the mark out of the blue. You raise a good point about FTX and the number of rationalists it hoodwinked, but that is a different (and perhaps more interesting) phenomena than the one I'm querying about.

[-]CstineSublime1y60

My new TAP for the year is - When I fail: try twice more. Then stop.

I'm persistent but unfortunately I don't know when to quit. I fall a foul of that saying "the definition of insanity is to try the same thing over and over again and expect different results". Need a pitch for a client? Instead of one good one I'll quota fill with 10 bad ones. Trying to answer a research question for a essay - if I don't find it in five minutes, guess I'm losing my whole evening on a Google Books/Scholar rabbit hole finding ancillary answers.

By allowing myself only two mor... (read more)

2Davey Morse10mo

I do this at the end of basketball workouts. I give myself three chances to hit two free throws in a row, running sprints in between. If I shoot a third pair and don't make both, I force myself to be done. (Stopping was initially wayy tougher for me than continuing to sprint/shoot)

1CstineSublime10mo

Thank you for sharing that, it is interesting to see how others have arrived at similar ideas. Do you find yourself in a rhythm or momentum when sprinting and shooting?

2Davey Morse10mo

not as much momentum as writing, painting, or coding, where progress cumulates. but then again, i get this idea at the end of workouts (make 2) which does gain mental force the more I miss.

[-]CstineSublime1y60

Kantmogorov Imperative - more of a philosophical dad-joke than a actual thing, it is the shortest possible computer program that outputs descriptions of morally consistent behaviors in all/any circumstances

5Alexander Gietelink Oldenziel1y

Anders Sandberg suggested to me once one could prove Kantian ethics impossible through a diagonalization argument.

[-]CstineSublime4mo50

What's the most L'esprit de l'escalier^[1] advice you've ever gotten?

I want to hear advice that at the time you either misunderstood or left you utterly baffled but some time later, you had a "Eureka" moment. A rushing in of insight long afterwards where you finally were able to apply the advice and understand it.

My own examples aren't that good but hopefully are illustrative:

In GLSL it is good practice not to use conditional statements, or to use them for very small branches.

I am not a coder by any stretch of the imagination so at first this seemed li... (read more)

6Viliam4mo

This assumes that both surgeons are evaluated by the same set of criteria (so if one scores lower on something, he probably scores higher on something else). That is not necessarily true. As a counter-example, imagine hospital A that only hires professional surgeons and pays them high salaries, and hospital B that hires anyone and pays them low salaries. The "surgeon who looks like a butcher" probably works in the hospital B.

[-]CstineSublime1y50

The promise of mind reading techniques whether it is a former FBI analyst or one of Paul Ekman's microexpression reading human lie detectors. I become aware of this cottage industry during every trial-by-media where suspicion piles upon someone not yet charged with murder.

I have to admit I am skeptical that anyone has such an amazing power to see through the facade of a stranger and with a greater-than-chance determine if they are telling the truth or not. Doubly so because I am someone who is constantly misinterpreted, I have to manage my gestures and fac... (read more)

2Viliam1y

On the scale between "pseudoscience that provides either completely random results or exactly what its operator wants to hear" and "always provides the correct answer", there are some uncomfortable points where we probably get first, such as "provides the correct answer 99% of the time" (and with the 1% chance you are unlucky, and you are screwed because no one is going to believe you) or "provides the correct answer for neurotypical people" (and if you are an autist, you are screwed). Similar here. My face expression is always on "neutral", and my statements, no matter how simple and literal, are often creatively interpreted. And I guess I am sufficiently unusual, so heuristics like "let's assume that he thinks/feels what an average person would think/feel in this situation" also fail. It took me a lot of time to understand myself to the level where I can explain things about myself verbally, but when I do, people usually find it implausible and try to find some hidden meaning behind my words. So... a machine that could read my thoughts could feel validating. Assuming it does so correctly. But there is also a chance it would provide correct answers for most people, and incorrect answers for the few unusual ones.

1CstineSublime1y

I'm afraid I need you to rephrase or elaborate on what you meant by this - are you saying, aware of a technique or method which is right 99% of the time or thereabouts. Or are you saying human variability makes such a technique impossible for anything but the most narrow populations? Or have I likely (and in a meta-way appropriately) completely missed the point? What do you think of more generally - as I explicate in the second half - revelations about a person's internalized belief structures, including their hero's and related moral system, but also the idea of idiolect being a symptom of their thinking and model of the world even if it is not a mechanism for directly ascertaining their personal belief in this or that specific statement?

2Viliam1y

Oops, I actually misinterpreted one part -- when you wrote "if placed under the microscope", my brain interpreted this literally, as if you were talking about a hypothetical future version of "mind reading" that would include checking your neurons by a microscope and probably interpreting the results using an AI. What I meant is that people usually think about these things in "yes or no" categories. For example, if you asked people whether existing lie detectors work, the most frequent answers would probably be "of course yes, why would they use them otherwise?" or "of course not, it's total bunk". There I didn't mean to make a statement about lie detectors per se, but about: this is how people think about technologies when you ask them. They think the correct answer is either "yes" or "no", even if it is something complicated like "sometimes" or "yes, but with exceptions". If the popular belief happens to be an unqualified "yes", and you happen to be the exception, you are screwed. I believe the current "mind-reading" techniques like Paul Ekman's are hit and miss. That they probably often work with typical people in typical situations, but fail when something unusual happens. (Someone may be scratching their nose because they are lying, but sometimes the nose is just itchy for a completely unrelated reason. Or the person is lying, but in a different way than you assume. Or is just generally uncomfortable, maybe thinking "this is true, but they seem unlikely to believe me".) Practically, "films, television, and role models, books, music and lyrics that someone has absorbed" are an enormous amount of data, especially for people who consume a lot of these media. Maybe someone who reads one book in ten years and only watches the mainstream TV could be modeled this way. But if you asked me to give you a list of books I have read and the movies I have seen, I could probably remember only a small fraction of them. How specifically is Paul Ekman going to find out which

1CstineSublime1y

I am overwhelmingly confident that analysis of the kinds of narratives that a particular person spins, including what tropes they evoke - even if you're not familiar with the tropes previously - would reveal a lot about their worldview, their ethical structure, the assumptions and modelling they have about how people, institutions, and general patterns they believe underlay the world. A oversimplified example is a person who clearly has a "victim "mentality" and an obsession with the idea of attractiveness because they always use sentence structures (i.e. "they stopped me") and narratives where other people have inhibited, bullied, envied, or actively sought to stifle the person telling the story and these details disproportionately make reference to people's faces, figures, and use words like "ugly" "hot" "skinny" etc. It is not necessary to know what films, books, periodicals they read.

0Viliam1y

I think you would get the set of topics, but not necessarily the right idea about how exactly those topics apply to the current situation. To use your example, if someone's speech patterns revolve around the topic of "bullying", it might mean that the person was bullied 50 years ago and still didn't get over it, or that the person is bullied right now, or perhaps that someone they care about is bullied and they feel unable to help them. (Or could be some combination of that; for example seeing the person they care about bullied triggered some memories of their own experience.) Or if someone says things like "people are scammers", it could mean that the person is a scammer and therefore assumes that many other people are the same, or it could mean that the person was scammed recently and now experiences a crisis of trust. This reminds me of an anime Psycho Pass, where a computer system detects how much people are mentally deranged... Anyway, this sounds like something that could be resolved empirically, by creating profiles of a few volunteers and then checking their correctness.

2CstineSublime1y

Yes. Which is invaluable information about how they see the world currently. How is that not the 'right idea'? If that is how they continue to currently mentally represent events? Your 'people are scammers' example is irrelevant, what is important is if they constantly bring in tropes or examples or imply deception. They may never use the word 'scammer' 'mistrustful' or make a declaration like 'no one has integrity'. The pattern is what I'm talking about.

[-]CstineSublime1y40

"Is this a good use of my time?"
"No"
"Can I think of a better use of my time?"
"Also, no"
"If I could use this time to think of a better use of my time, that would be a better use of my time than the current waste of time I am now, right?"
"Yes, if.... but you can't so it isn't"
"How can you be so sure?"
"Because, look at how abstract just this little dialogue is - which is wholly representative of the kind of thinking-about-better-uses you're inclined to do (but may not be generalizable to others). This dialogue of ours is not pertaining directly to any actions of tangible value for you. Just hypothesis and abstracts. It is not a good use of your time."

[-]CstineSublime1y40

Sturgeon's Law but for ideas?

Sturgeon's Law is a counterargument against the negative stigma that Sci-Fi writing had as being crappy and therefore not a legitimate medium. The argument is 90% of any genre of writing, in fact anything from "cars, books, cheeses, people and pins" are "crud". Although the sentiment does seem to have a precedent in a novel Lothair by British Prime Minister Benjamin Disraeli where a Mr. Phoebus says:

"nine-tenths of existing books are nonsense, and the clever books are refutation of that nonsense. The greatest misfortune t

... (read more)

[-]CstineSublime1y*40

Examples of how not to write a paragraph are surprisingly rare

Epistemic Status: one person's attempt to find counter-examples blew apart their own ( subjective) expectations

I try to assemble as many examples of how not to do something as 'gold standard' or best practice examples of how the same task should be done. The principle is similar to what Plutarch wrote: Medicine to produce health must examine disease, and music to create harmony must investigate discord.

However when I tried to examine how not to write, in particular examples of poorly writt... (read more)

4Jonathan Claybrough1y

Interesting thoughts, ty. A difficulty to common understanding I see here is that you're talking of "good" or "bad" paragraphs in the absolute, but didn't particularly define "good" or "bad" paragraph by some objective standard, so you're relying on your own understanding of what's good or bad. If you were defining good or bad relatively, you'd look for a 100 paragraphs, and post the worse 10 as bad. I'd be interested in seeing what were the worse paragraphs you found, some 50 percentile ones, and what were the best, then I'd tell you if I have the same absolute standards as you have.

[-]CstineSublime3mo30

Does anybody here have any advice or thoughts on the "two list" approach?

I can't remember who it suggested it and I'm likely conflating different anecdotes - but the gist was you have two lists: a traditional to-do list which has tasks and actions; and a second list that I've seen described as a "to think" list - which might contain a series of problems or questions that are important but for which you are unable to progress.

In my case my "to think" or "problems" list would be as long as my arm, my to-do list would be filled obvious stuff: with whatever is... (read more)

[-]CstineSublime4mo30

Don't ask "Can anyone recommend me a book/camera/school for X?" instead ask "What criteria would use use when deciding on a book/camera/school for X?"

I've noticed the tendency for people to give "title recommendations". No matter how much detail about your problem or decision you furnish, they will furnish only the title of a book, the name of a product without elaboration. Example: For a while any and every request for a self-help-adjacent book on Reddit would be met by two to four words "Atomic Habits - James Clear". Why were they recommending that book?... (read more)

5Said Achmiz4mo

But… doesn’t this response make perfect sense? Is the implication not that the Panasonic Lumix GH6 is a video camera which has a lot of options for lenses, and operates in low light? Like, if you say to me “Said, could you recommend a hand mixer? I’m looking for something that’s not too expensive, but decently powerful, compatible with standard mixer attachments, and in particular I want it to be able to whip up a good egg foam”, I’m going to say “Black+Decker MX3200”. If you then say “But Said, why did you recommend that mixer?!”, what am I supposed to say other than “… because it’s a hand mixer that’s not too expensive, decently powerful, compatible with standard mixer attachments, and will definitely whip up a good egg foam”? What other information would you expect?

-1CstineSublime4mo

I don't know that. And 90% of the time it isn't the best solution to my problem, it's just the one top-of-mind. Why though? Why that one? What does it have that all the others don't? Why it has MORE of that than all the other options on the market for starters. But I would expect your reasoning and hands-on experience if applicable. I can't comment on a hand mixer but for a camera. But the most obvious question is - are you basing your recommendation on a factor I haven't furnished? And How would I know? How is it reasonable for me to know? And if do think it is the best option based on what I've told you - you haven't given me any of your reasoning, there could be countless reasons off the top of my head I would expect you to answer at least one of these: do you shoot live music with that camera? What are you experiences with that camera in live music environments? Do you find it easy to hold when you're trying to move around people throwing their fists up in a crowded pub for a punk band at 11pm? Or do you shoot more respectable Jazz gigs? or do you shoot "from the wings"? If you do all the above - which ones does it perform best in and which ones does it perform weakest in, why? Do you use a wide lens? Is it a "fast lens"? or do you use a slower lens but it still provides a bright image? Is it heavy? Is it light? What surprised you about using it in a gigging environment? If you think it is the best compromise on the market or my expectations are unreasonable? Which factors are do you value more? These are all pivotal to knowing whether you just spouted a model name at random, or if you've actually thought about it.

2Said Achmiz4mo

The others are more expensive, or less powerful, or not compatible with standard mixer attachments… or are worse for other reasons that there’s no point in mentioning when making a positive recommendation (e.g. “definitely don’t get the ExampleCompany SuperMixer 9000, because it has a manufacturing defect that causes it to explode and set your house on fire”—no point in going through all the bad options and saying why they’re bad, otherwise we’ll be here all day, and to what end?). Well, I generally don’t recommend things that I don’t have hands-on experience with (and if I do, I explicitly flag this lack of experience). As for “reasoning”… this just doesn’t seem to be the right way to think about this sort of thing. We’re talking about facts (nothing to reason about here) and evaluations on the basis of given criteria (also not much room for “reasoning”). “Reasoning” applies when coming up with criteria, but once that’s done—what’s to reason about? And the question “why it has more of that” just seems weird. Maybe I’m misunderstanding what you mean here, but surely the answer is “because that’s how it was designed and manufactured”? What else can be the answer to a “why” question like that? (Are you expecting some sort of analysis of Black+Decker’s product strategy…?) So, correct me if I’m way off base here, but it sounds like what you’re actually looking for isn’t so much answers to specific questions, or specific information, but something more like… credible signals of trustworthiness (of a recommendation, and of the recommender). In other words, with the mixer example (sorry, I know kitchen appliances much better than I know cameras, so I’m sticking with that one for my comments)… it’s not that a bare product name is insufficient information, as such. It’s not that the response fails to answer the question—it absolutely does that. Rather, what’s missing is any credible signal that you can trust that the answer is correct. And stuff like talking about one’s

1CstineSublime4mo

You've made me wonder: Where do you ask for recommendations or advice Said? Of whom? How do you you account for unknown unknowns? Do you never ask for clarifying or additional questions about their suggestions? Just your replies here appear to be quite curious so I find it confusing that you would so strenuously argue for a "and that's all she wrote" approach. How do I know I have the best criteria for me? I can do a really rigorous job of explaining my situation, my biases, my personal experience and my current level of understanding and knowledge. But the whole point of asking for recommendations is to find unknown unknonws: not only products or solutions on the market, but criteria I may not have even thought of, or at least under-prioritized. If I'm a novice at something - what really obvious things might someone who isn't a novice want to warn me of? Hence the Ray Dalio thing - in an ideal situation I ask three or more domain experts - people who certainly do know more than me and can explicate criteria I didn't even consider. With the camera example, (sorry, like you with the mixer - I'll stick to what I know) I might verbosely describe the situation I intend to shoot in. But someone with more experience might go "Ah, but what about a battery?" or maybe they'll say "This camera has a longer battery life, but go for this cheaper one because the cost of an external USB powerpack is a tenth of the price difference between the two models". How would I know that battery life factored into their reasoning and in what way? In terms of execution, switching roles doesn't exactly work, i.e. asking "How would you choose a camera?" because the person asked may choose what is best for them, not for the requestee. But, there's a middle ground somewhere between "Which product fits these criteria?" and "which criteria am I neglecting?". Partly. Yes, it is about trust and credibility. And I still contend the easiest way is for someone to mention a lot of "becaus

2Said Achmiz4mo

Well, the most common places/contexts where I ask for recommendations would be the various chat channels where I talk to people (which could be my IRL friends, or not). Most people in such places are fairly intelligent, knowledgeable, and technically adept. Do I ask clarifying or additional questions? Sure, but such questions tend to be clarifying of my own purposes or needs, rather than the recommender’s reasons for giving the recommendations. More common is for me to need to clarify my request rather than for recommenders to need to clarify their recommendations. How do I account for unknown unknowns? Well, uh… I don’t, mostly? They’re unknown, so how can I account for them? Perhaps you have in mind some question like “what general policies can mitigate the downside risk of unknown unknowns”? Indeed there are such things, but they tend to be very general, like the well-known principle of “first buy the cheapest version of the tool you need, and use it until it breaks; then you’ll known enough to make a more informed choice”; or practical, common-sense principles like “whenever possible, buy from vendors that allow free returns”. Hm… I think I see your point now, yes. What you want, it seems to me, is a way to prompt the would-be recommender to give you as much of their relevant domain knowledge as possible. Basically, you want to get the person to talk. The way to do that, in my experience, is actually to say less about your own situation and problem, at least at first. Instead, try to elicit opinion. People generally love to talk about their views about things. So, not “what [camera/mixer] would you recommend, given these specific needs”, but “what is the best [camera/mixer], and why”. Then once you’ve got them talking, you can ask things like “but what if I specifically need X, what are your thoughts on that”—and now you should get some more detailed response. (Incidentally, on my own “Recommended Kitchen Tools” page, I do try to explain my reasons, and to

[-]CstineSublime5mo30

I often feel misunderstood, so what if I A/B tested this...

I write a blog post about something which is very niche or personal and thus bound to get misinterpreted. I try, to the best of my ability, to make it as clear as possible and post it here.

Then, I take that exact post as posted, put it into an LLM and ask it to rewrite it and keep asking it to rewrite it until I feel it expressed what I intended to express. Maybe a couple of LLMs and duke them out.

Finally, I post it on here.

I'm not sure how I would test which one is more successful? Enthusiasm (pos... (read more)

[-]CstineSublime6mo30

Which fictional character prompts an LLM into speaking the plainest of English?

I've never been satisfied with the way that LLMs explain technical jargon, even if I ask for "plain English" it will have an annoying habit of including some other jargon or using very hard to parse explanations. This requires a further plain English explanation.^[1]

Enter Homer Simpson, asking for "in the manner of Homer Simpson" makes the replies easier to understand. However they do lead to the gratuitous analogies to donuts or beer, and non-sequitur D'oh's being thrown on the ... (read more)

0avturchin6mo

"Explain as gwern ELI5"

[-]CstineSublime7mo30

I don't like the word "motivation" because I find one of the chief factors in whether I'm motivated to do something or not is belief. Most discussions of motivation seem to basically see it as the pain or "cost" of doing something versus the reward. However just because you do something, be it painful or easy, doesn't mean you'll get the reward.

Perhaps some fragmentary dialogue will illustrate my thinking:

"digging for gold is hard work and you're not even sure if you'll find anything" - low motivation. High cost (hard work) no certainty of reward.

"I'd basi

... (read more)

6cubefox7mo

This is related to the expected value, isn't it? The expected value of something is its value if true times the probability of it being true. Where probability stands for degree of belief of it happening. I.e. EV(X)=P(X)V(X). Then presumably what you are proposing is the following formula: Motivation(X)=ExpectedReward(X)−ExpectedCost(X). Right? This would mean Motivation(X)=P(X)Reward(X)−P(X)Cost(X). However, the above assumes that the thing that gives reward is the same thing that has consequences (X in this case). Which can't be the case as the probability of incurring the cost need not be the same as the probability of getting reward, contrary to the previous formula. An arguably better formula comes from Richard Jeffrey's theory. He models the value V of some possible fact (proposition) as the degree of desire for it to be true. The probability P is again the degree of belief that the proposition is true. Now both outcomes we care about and actions which could make those outcomes happen can equally be described by a proposition. And the degree of desire V toward an action can be interpreted as the motivation to do that action. In particular, for some action A and some outcome O he has the following theorem in his theory: V(A)=V(O∧A)P(O∣A)+V(¬O∧A)P(¬O∣A). Here there is no explicit distinction between cost of an action and the reward of the outcome. Those values are already combined in the terms V(O∧A) (both the action and the outcome happen) and V(¬O∧A) (the action happens but the outcome doesn't happen). The two probabilities weigh those terms depending on the outcome happening/not happening conditional on the action. So the resulting value (degree of desire / motivation?) for the action A is a weighted average.

1CstineSublime7mo

I'm afraid I can't read probabilistic notation, but on first blush what you've described does sound like I'm simply reinventing the wheel - and poorly compared to Jeffrey's Theory there. So yes, it is related to the expected value. And I like how Jeffrey's theory breaks the degree of belief and the desire into two separate values.

2Viliam7mo

Things that can influence the belief: * peer pressure: when other people say that something is true, the beliefs becomes contagious * getting quick feedback: if doing a part of it already gives you a partial reward, it is motivating * getting the information from a trustworthy source: if their previous claims turned out to be true... * getting other reward in parallel, e.g. being socially rewarded for working on the project

1CstineSublime7mo

I'm not sure how I (me, specifically - may be generalizable to others?) can apply any of those unless I'm already receiving feedback, reward. In the interest of being specific and concrete I'll use one example - my personal bugbear: the refrain from people who tell me that as a creative freelancer I need to "get your stuff out there" stuff here nebulously referring to the kinds of videos I can make. "There" is an impossibly vague assertion that the internet and social media are vectors for finding clients. Yet I have low belief in "getting stuff out there" is an effective measure to improve my standing as a creative freelancer, let's go through your suggested influences one-by-one: Peer Pressure: well it doesn't work evidently since I've been repetitively told that "you need to put your stuff out there" is true - but I don't believe it. These people are often peers, stating it as a fact, yet it doesn't shift my belief. The caveat I would put here is I have not had luck finding clients through previous forays online and most of my clients appear to come from offline networks and relationships. Getting Quick Feedback: This does seem like the most effective means of shifting belief - however it is no applicable in this example as the feedback is non-existant, let alone quick. Likes and comments don't translate into commissions and clients. Getting the information from a trustworthy source: yes, generally true, call it "appeal to authority" call it Aristotle's theory of ethos in rhetoric. Yet not applicable in this example, in fact people who repeat this refrain appear less trustworthy to me. Getting other reward in Parallel: Likes and comments are rewards in a sense, yet do not influence my belief because it is not directly affecting the core metric which is - getting more clients or commissions. However there are some caveats: the advice is impossibly vague and therefore impossible to action. Which begs the question of - what is my lack of faith or belief with?

2Viliam7mo

Can they give you specific examples of the clients they gained as a result of publishing a video on a social network? (Maybe there was something special about those videos, and it is not true for videos in general. But maybe you could copy that special thing.) Agreed, they don't. Maybe shares make it more likely for the video to reach a potential client. I suspect that a good video needs to be "actionable": it should give a specific example of a problem that you can solve, and it should explicitly encourage them to contact you if they need to have a problem like that solved. Other types of videos are only useful if they make people click on your video feed and find the "actionable" ones. But that's just a guess; I never had a job like that.

3CstineSublime7mo

To be honest I haven't asked for specific examples (and I guess I'll need to find a way to ask for it which is not misconstrued as confrontational) but no one has been forthcoming. Yup, "Hey look at this, you should get them to do your next music video for you" or within a band: "hey look at this video they did for this band, we could use something like that". That rings true. The only person I know personally who has gotten such high social media engagement they are now getting spots on traditional media is an "expert", therefore they provide actionable advice. They have both the credentials and the industry experience to back it up. It also (unfortunately) helps they intentionally solicit controversy and use clickbaity statements. And it's a topic which is always in demand. At their behest I've tried putting out didactic videos on what bands and artists should do for music videos, explaining different tropes and conventions where are cool. but like after 2 months I ran out of ways to make it "actionable". Maybe if I continued the grind for 6+ months the algorithm would have started pushing my content more on people outside of my network's Instagram feed? Or maybe I need to pay for ads?

[-]CstineSublime10mo30

I often ask myself and others "okay, but how does that look in practice?" - this is usually when I have a vague idea about something I need to achieve a goal, but also when someone gives me some vague advice that I feel is leaving it to me to "draw the rest of the owl."

Is this the best phrasing of the question? I have my doubts.

Firstly, is it too generalized for different domains?

"I should really organize my dresser drawers more thematically" -> "okay, but how does that look in practice?"

"I need to make more of an effort to promote my freelancing"

... (read more)

[-]CstineSublime10mo32

I really like the fact that there's an upvote feature together with a separate agree/disagree feature on this site.

I may like the topic, I may want to encourage the author of the post or comment to continue exploring and opening up a dialogue about that particular topic. I might think it's a valuable addition to the conversation. But I may just not agree with their conclusions.

It's an important lesson: failure can reveal important information. You don't have to agree with someone to feel richer for having understood them.

On the other hand, I'm also guilty ... (read more)

[-]CstineSublime1y30

How can you mimic the decision making of someone 'smarter' or at least with more know-how than you if... you... don't know-how?

Wearing purple clothes like Prince, getting his haircut, playing a 'love symbol guitar' and other superficialities won't make me as great a performer as he was, because the tail doesn't wag the dog.

Similarly if I wanted to write songs like him, using the same drum machines, writing lyrics with "2" and "U" and "4" and loading them with Christian allusions and sexual imagery, I'd be lucky if I'm perceptive enough as a mimic to produc... (read more)

3Viliam11mo

That reminds me of NLP (the pseudoscience) "modeling", so I checked briefly if they have any useful advice, but it seems to be at the level of "draw the circle; draw the rest of the fucking owl". They say you should: * observe the person * that is, imagine being in their skin, seeing through their eyes, etc. * observe their physiology (this, according to NLP, magically gives you unparalleled insights) * ...and I guess now you became a copy of that person, and can do everything they can do * find the difference that makes the difference * test all individual steps in your behavior, whether they are really necessary for the outcome * ...congratulation, now you can do whatever they do, but more efficiently, and you have a good model * design a class to teach that method * ...so now you can monetize the results of your successful research Well, how helpful was that? I guess I wasn't fair to them, the entire algorithm is more like "draw the circle; draw the rest of the fucking owl; erase the unnecessary pieces to make it a superstimulus of the fucking owl; create your own pyramid scheme around the fucking owl".

1CstineSublime11mo

I completely agree and share your skepticism for NLP modelling, it's a great example of expecting the tail to wag the dog, but not sure that it offers any insights into how actually going about using Ray Dalio's advise of reverse engineering the reasoning of someone without having access to them narrating how they made decisions. Unless your conclusion is "It's hopeless"

2Viliam11mo

Yes, my conclusion is "it's hopeless". (NLP assumes that you could reverse-engineer someone's thought processes by observing their eye movements. That looking in one direction means "the person is trying to remember something they saw", looking in another direction means "the person is trying to listen to their inner voice", etc., you get like five or six categories. And when you listen to people talking, by their choice of words you can find out whether they are "visual" or "auditive" or "kinesthetic" type. So if you put these two things together, you get a recipe like "first think about a memory that includes some bodily feelings, then turn on your auditive imagination, then briefly switch to visual imagination, then back to auditive, then write it down". They believe that this is all you need. I believe that it misses... well, all the important details.)

1CstineSublime11mo

Sorry I made a mistake in my last reply: putting NLP aside, are there any effective methods of reverse engineering the decision making of people that you can't get on the phone? There's an abundance of primary evidence for many decisions, whether it be minutes of deliberations, press releases which might involve more reading of the tea-leaves. In the case of Prince one could possibly listen to different live-performances of the same song and analyze what changes are made. What words are crossed out on a lyrics sheet. Many times people have to become very good at intuiting people in their life who are loathe to actually explain their reasoning, yet build pretty useful models of how to interact with those people. From grumpy shopkeepers, to school teachers, to coworkers etc. etc. Diplomacy is an entire profession based on building such models. Negotiation builds those models under pressure - but often has the ability to speak with the other side, as per Ray Dalio's suggestion, which I'm trying to find a method for. Are there no methods of understanding and reverse engineering the reasoning, not the superficial aspects, of another person?

3Viliam11mo

Dunno; it probably depends a lot on the kind of task, the kind of person, and your observation skills. Some people explain more, some people explain less. Some people are more neurotypical (so you could try to guess their patterns by observing what other people similar to them would do in a similar situation), some people are weird and difficult to predict. At some tasks people produce artifacts (a mathematician can make notes on paper while solving a problem; if you obtain the paper you could reconstruct some of their thoughts), other tasks happen mostly in one's head (so even if you placed hundred hidden cameras in their room, the important steps would remain a mystery). I guess the success is usually a combination of superior observation skills and the person/task being sufficiently typical that you can place them in a reference group you have more data about. (For example, I have met people who had good observation skills, but had a difficulty understanding me, because their model maybe worked for 90% of people and I was among the remaining 10%.) So, if possible: * make a good model of a similar kind of person * become familiar with the kind of work they are doing * try to obtain their working notes That is, if you tried to reverse-engineer Prince, it would probably be useful to have knowledge about music and performing (even if nowhere near his level), as that might give you at least some insights about what he was trying to achieve. Looking at his notes or his history might help to fill some gaps (but you would need the domain knowledge to interpret them). People similar to Prince (not sure who would that be) might have a good intuitive model of him, and you could ask them some things. At the end, it would all be probabilistic, unreliable.

[-]CstineSublime1y30

I constantly think about that Tweet where it's a woman saying she doesn't AI to write or do art, she wants it (but more correctly that's the purview of robotics isn't it?) to do her laundry and dishes so that she can focus on things she enjoys like writing and art.

Of course, A.I. in the form of Siri and Alexa or whatever personal assistant you use is already a stone's throw away from being in a unhealthy codependent relationship with us (I've never see the film 'Her' but I'm not discussing the parasocial relationship in that film). I'm talking about the li... (read more)

2Viliam11mo

but... it already does :( I mean, on facebook and xitter and reddit; I am still free to control my browsing of substack and yes, applying the same level of control to my real life sounds like a bad idea

1CstineSublime11mo

I meant a personal assistant type A.I. like Alexa or Siri which is capable of exerting Milieu control like Sir Humphrey does: Meta properties, Tik Tok are not yet integrated with such personal A.I. assistants... yet.

[-]CstineSublime1y30

Brainstorming (or babbling) is not random. Nor would we want it to be truly random in most cases. Whether we are operating in a creative space like lyric writing or prose, or writing a pedagogical analogy, or doing practical problem solving on concrete issues. We don’t actually want true randomness, but have certain intentions or ideas about what kind of ideas we’d like to generate. What we really want is to avoid clichés or instinctual answers – like the comic trope of someone trying to come up with a pseudonym, seeing a Helmet in their line of sight and ... (read more)

[-]CstineSublime1y30

Why don't LLM's ask clarifying questions?

Caveat: I know little to nothing about the architecture of such things, please take this as naive user feedback if you wish, or you could ignore it.

Just now I was asking the Meta AI chatbot how to do an 'L-Cut' using the Android Instagram app. It hallucinated for quite a few messages instructions how to 'overlap' two video tracks when editing a reel before it finally admitted that no such ability in fact exists in the Instagram App.

My grossly negligent mistake was assuming that a AI LLM with Meta Branding would have... (read more)

4Nathan Helm-Burger1y

These sorts of behavioral choices are determined by the feedback given by the people who train the AI. Nothing to do with the AI's architecture or fundamental inclinations. So the question to ask is, "Why do all the AI companies seem to think it's less ideal for the AI to ask clarifying questions?" One part of the reason is that it's a lot easier to do single turn reinforcement. It's hard to judge whether a chatbot's answer is going to end up being helpful if it's current turn consists of just a clarifying question.

1CstineSublime1y

Yes I assumed it was a conscious choice (of the company that develops an A.I.) and not a limitation of the architecture. Although I am confused by the single-turn reinforcement explanation as while this may increase the probability of any individual turn being useful, as my interaction over the hallucinated feature in Instagram attests to, it makes conversations far less useful overall unless it happens to correctly 'guess' what you mean.

[-]CstineSublime5mo20

My writing is sloppy. Can anyone please suggest any resources where I can get feedback on my writing, or personalized instructions that will improve my processes to make me a better writer?

In the meantime I'll try to adopt this "one simple trick": each time I write a piece, I will read it out aloud to myself. If it is "tough on the ear" or I stumble while sight reading it, I will edit the offending section until it is neither.

Also, I'll continue to get LLMs to summarize the points in a given piece. If there's something I feel is missing in it's summary or ... (read more)

5JustisMills5mo

For LessWrong posts specifically, there's the feedback service. This isn't personalized, but I also have suggestions for people in the general LessWrong cluster here.

1CstineSublime5mo

I've read your article before and found it to be good advice. I have tried to take the first warning about ambiguous use of "this" to heart for a while now. I might have to get the courage to engage the feedback service.

2lemonhope5mo

I'm no writer or editor but you could email me. I check my email every few days lemonhope@fastmail.com

1Loki zen5mo

inasmuch as personalised advice is possible just from reading this post (and as, inter alia, a pro copyeditor), here's mine - have a clear idea of the purpose and venue for your writing, and internalise 'rules' about writing as context-dependent only. "We" to refer to humanity in general is entirely appropriate in some contexts (and making too broad generalisations about humanity is a separate issue from the pronoun use). The 'buts' issue - at least in the example you shared - is at least in part a 'this clause doesn't need to exist' issue. If necessary you could just add "(scripted)" before "scenes". Did someone advise you to do what you are doing with LLMs? I am not sure that optimising for legibility to LLM summarisers will do anything for the appeal of your writing to humans.

1CstineSublime5mo

Good question, no, no one advised me to use this technique but I use it as a last resort. I frequently feel that I am misunderstood in communication. Often I feel like people's replies to me sound like replies from totally different conversations or statement/questions to the one I just made. If an LLM seems to imply the focus is different or overemphasizes something I didn't see as significant, then I see no reason to believe that isn't indicative that humans will be dragged away by that too.

1Loki zen5mo

It may well be. It's been my observation that what distracts/confuses them doesn't necessarily line up with what confuses humans, but it might still be better than your guess if you think your guess is pretty bad

[-]CstineSublime5mo20

Quick and incomplete roundup of LLM prompting practices I regularly use - feel free to suggest your own or suggest improvements:

-Try asking it to answer "in one sentence". It won't always sufficiently compress the topic, but if it does. Well... you saved yourself a lot of time.

-Don't use negatives or say "exclude"... wait... I mean: state something in harmony with your wishes because unnecessarily making mentions to exclusions may inadvertently be 'amplified' even though you explicitly asked to exclude them.

-Beware hallucinations and Gell-Man Amnesia: Do a... (read more)

[-]CstineSublime10mo20

We have Shannon Information, Quantum Information, Fisher Information, and even Mutual Information and many others. Now let me present another type of information which until I find a better name will certainly be doomed to reduplication induced obscurity: Informative Information.

One of the many insightful takeouts from Douglas Hubbard's Book - How to Measure Anything for me was that if a measure has any value at all then it influences a decision. It informs a decision.

If I see a link come up on my social media feed "5 rationality techniques you can u... (read more)

2gwern10mo

Why not just 'valuable information', in a Value of Information sense of 'valuable'?

1CstineSublime10mo

Bad information can inform a decision that detracts from the received value. I suppose if it is perceived to be valuable it still is a useful term - do you think that would get the point across better?

[-]CstineSublime11mo20

HOW TO THINK OF THAT FASTER: A few quick, scattered, incomplete and wholly unsatisfactory list of observations and hunches:

- First, notice when you're stuck in a rut. When you're beating your head against a wall.
- Second, having noticed you're in a rut try twice more. My TAP is - "Failed once? Try 2 more - then stop"
- "Why am I doing it this way?" - I keep coming back to this quote from Wittgenstein:

"To look for something is, surely, an expression of expectation. In other words: How do you search in one way or another expresses what you expect."

In th... (read more)

[-]CstineSublime1y20

The value of an idea is dependent on what Stuart Kauffman may call 'Adjacent Possibilities'. Imagine someone has an idea for a film, a paragraph long "Elevator Pitch" which has the perfect starring role for Danny DeVito. The idea becomes more and more valuable the closer within six degrees of separation anyone with that idea is to DeVito. If I have such an idea, it's worthless because I have no means of getting it to him.

Likewise, imagine someone has a perfect model for a electronic fuel injection system in Ancient Greece, but just the injection system. Th... (read more)

[-]CstineSublime2y20

I have some hard data that disproves I'm a realist, but suggests I'm a pessimist (i.e. more than 50% of the time my best expectations of personal outcomes are worse than the case). Now what?

The hard data is a handful of specific predictions for example, a prediction about a financial investment (where performance consistently exceeded my expectations) where I had no control over the performance, and things like a exam where I had control (although an instructor noted that I performed worse on the exam than a practice exam because of 'nerves').

Arguably the ... (read more)

2Morpheus2y

Sounds like the right kind of questions to ask, but without more concrete data on what questions your predictions were off by how much, it is hard to give any better advice than: if your gut judgement tends to be 20% off after considering all evidence, move the number 20% up. Personally me and my partner have a similar bias, but only for ourselves, so making predictions together on things like "Application for xyz will succeed. Y will read, be glad about and reply to the message I send them" can be helpful in cases where there are large disagreements.

[-]CstineSublime2y*20

Previously I thought that if you ask better questions then you will get better at solving problems. However questions are the shadows or reflections cast from the actual framing of the problem. If you have a well framed problem you will naturally ask better questions. If you haven't framed the problem well, then you will ask bad questions.

Bad questions are still useful because they are a signal that you are "barking up the wrong tree" or that you need to reformulate the problem.

What marks a bad question and therefore signals a framing of the problem t... (read more)

[-]CstineSublime20d10

"Araffe" is a nonsense word familiar to anyone who works with generative AI image models or captioning in the same way that "delve" or "disclaim" are for LLMs and presents yet another clear case of an emergent AI behavior. I'm currently experimenting with generative video again and the word came to my attention as I try to improve the adherence to my prompts and mess around with conditioning. These researchers/artists investigated the origin of 'arafed' and similar words: it appears to be a case of overfitting the BLIP2 vision-language pretraining framewor... (read more)

[-]CstineSublime23d10

Prompted by Raemon's article about "impossible" problems, I've been asking myself:
What do I actually mean when I say something is “very hard” or “difficult”?

I wonder if my personal usage of these words less describes the effort involved, but more the projected uncertainty. If I describe something as difficult, I tend to use it in one of these three patterns:

There is one important variable that I have little leverage over, “Can I take a few days off work? It would be difficult to get permission from my supervisor” (that is to say, I don’t believe I can conv

... (read more)

[-]CstineSublime3mo10

Last week I finally and at long last managed to prompt an LLM into a "Socratic^[1] Brainstormer". It's really frustrating though that it doesn't ever hit me with a real curve ball that makes me go "Oh wow, I never thought of that" or "Oh shit, everything I said is a falsehood" a la Socrates. But as far as an rubber-duck with extra bells and whistles go, it's a step forward.

It has stopped with the overt sycophancy, no more "fascinating" or "what you've just said is very significant". It just asks questions, as few or as many questions as I specify. Clau... (read more)

[-]CstineSublime3mo10

I feel not taken in by LLM chatbots obvious sycophancy. I'm not sure if this reflects well on me, or is a sign of low self-confidence and that skepticism of praise and compliments is my baseline.

[-]CstineSublime3mo10

I struggle to write cover letters for applications^[1], despite being self-aware. The obvious remedy would be to lie and make up what skills or abilities I have based on the application, and hope there's no negative repercussions later. I see my difficulty in writing cover letters as part of a wider pattern of being unable to answer the question "what am I good at - that people need enough to pay for?" which is a fundamentally different to the questions "what am I proud of?" and "what are my passions?". Writing a cover letter involves not only identifying t... (read more)

[-]CstineSublime3mo10

Reading this post today, I realized I don't actually know what the difference is between a belief and a model. I still don't.

Models, as I understand it, are comprised of assumptions about cause-effect relationships which are themselves beliefs. I'm thinking about decision making models - where these anticipated causal effects influence what conclusion (i.e. option) is reached by a model. So a model is basically are a cluster of beliefs. But that doesn't appear to be the idiomatically correct way to refer to both (I believe I am wrong in my usage/understand... (read more)

[-]CstineSublime3mo10

Inverting Argument from authority, when is it good practice? I saw someone watching a daytime television show where they brought in someone to do an infomercial about health-marketed products. Instinctively, I didn't listen to any of the claims, instead all I could think of was "Well they're a salesperson, everything they're going to say is biased towards selling the products they brought today. This person shouldn't be watching this".

I'm sure you see my mistake here. It's convenient, it's easy, just filter out everyone who doesn't seem credible or who you suspect of having a bias: but is it optimal?

2JBlack3mo

There are ancient texts on this matter, such as https://www.lesswrong.com/posts/34XxbRFe54FycoCDw/the-bottom-line Short answer: once you know that you are listening to someone who wrote the bottom line first, then anything they wrote (or said) above the bottom line is worthless toward determining whether the bottom line is true. It is still possible that they present information that is of some use in other respects, but only to the extent that it is not correlated with the truth or falsity of the bottom line. Now, in some world it may be that if the bottom line were false, then fewer people would argue for it and such arguments would be less likely to appear on daytime television. That does not appear to be the world we live in.

2Garrett Baker3mo

This is not true, see filtered evidence. Said simply, when someone has their bottom line written, you should think about whether the argument they’re presenting is more or less convincing than you’d expect on priors. If its more, update in favor of them. If its less, update against. Of course, I would suggest pairing this with some practice making concrete forecasts so you can calibrate yourself on the typical qualities of argument to expect for various wrong and right conclusions.

1CstineSublime3mo

Does this apply for people who don't have a bottom-line written first? I'm thinking about, say, how I like hearing the opinions of people who view modern art, but have no art history or formal art education: I like hearing their naive judgements - now if they argue why a artwork is good that I find convincing, is this analogous to hearing a salesman who obviously has their bottom-line written first why this supplement or food product is good for, I dunno, sleep, and making a surprisingly bad or good argument in favour of that? In both cases: the naive judge of art, the salesperson -- I have a certain expectation about how convincing they will be. Actually, I tend to expect television salespersons will not be convincing. I find the type of arguments (or lack of argumentation) and rhetoric they deploy just don't work well with me. And I realize I'm in the minority. I expect their style works on the majority of television viewers.

1CstineSublime3mo

To be fair, i don't even know what their bottom line is - I only caught a vague sense that it had something to do with health. Not what specific benefits they were promising, nor how much they hedged those claims.

[-]CstineSublime4mo10

I've never learned the method of loci or Memory Palace, I've certainly never tried to take my childhood home and use it to remember things. I find it interesting then that the standup comedy routines I've memorized after multiple YouTube viewings, I don't remember the actual videos, I don't recall the gestures, the set design, the media qualities of the video transfer. Instead I seem to recall images conjured in my imagination.

Emo Philips saying "My brain started to bleed" calls to mind a vivid memory of him as a youngster, eyes rolling in his head, blood ... (read more)

[-]CstineSublime4mo10

How to be a good planner?

What do I mean by that? Given a goal which you do not have domain knowledge or previous experience related to executing it how do you maximize the chances of choosing a sequence of actions to succeed at that goal - to give some sundry examples:

A. running a bakery as a profitable business - with no prior baking or management experience,

B. writing a plugin for GIMP using Python - with no prior python knowledge, no prior knowledge of GIMP's APIs or standards for plugins,

C. filming a three car chase sequence for a action mo... (read more)

1dbohdan4mo

The baseline planning skill is having a start-to-end plan at all as opposed to winging it or only thinking ahead in an ad hoc manner. One step beyond this is writing the plan down, perhaps as a checklist. You can use the written copy to keep track of where you are, refine the plan, and simply to not forget it. A step beyond, which seems rarer and less automatic for people than the previous, is to employ any kind of what they call a "work breakdown structure": a systematic mapping from higher-level steps ("find out the legal requirements for filming a car chase") to lower-level steps ("ask indie filmmaker chat what legal firm they recommend").

[-]CstineSublime4mo10

I am bad at compressing my thoughts. The last few times I've tried to write a two sentence quick take, it has ballooned to a multi-paragraph monstrosity. This has bucked the trend, but only just.

[-]CstineSublime5mo10

-"Nobody actually believed there's only four types of stories... well okay not nobody, obviously once the pithy observation that a Freshman writing class produced works that could easily be categorized into four types of stories was misquoted as saying all stories follow that formula, then someone believed it."
-"You're confusing Borges saying that there are four fundamental stories with John Gardner's exercise for students. Borges said the archetypes of the four fundamental stories are the archetypes are the Siege of Troy - a strong city surrounded and def... (read more)

[-]CstineSublime5mo10

Good storytelling/comedy ^[1]writing involves many of the same skills as good planning or project management but with the inverse goal.

When you're planning a project you want to Murphyjitsu the most likely points of failure, you want to think through how can you minimize risk, disruption by thinking about all the likely causes of problems and putting in mechanisms to nip them in the bud. If you identify the raw materials for your factory not arriving by a certain date as a hazardous likelihood, maybe you instead seek supply from multiple suppliers so t... (read more)

[-]CstineSublime6mo10

Is Claude less prone to hallucinating than Chat GPT?

I've been playing around with DuckDuckGo's Claude 3 Haiku and ChaptGPT 4o Mini by prompting with this template:

What, if anything, can you tell me about [insert person/topic]...?

I do this as a precaution - before doing my "real prompt" I want to do an epistemic spot-check on whether or not the LLM can get the basic facts right. It appears Claude 3 Haiku has a much higher threshold for what it will reply on than Chat GPT 4o Mini.^[1]

Claude 3 Haiku gives a standard disclaimer rather than an answer for former-... (read more)

2CstineSublime6mo

Edit - 19th June 2025 - Claude spat out it's standard disclaimer when I asked it about Nicholas Nassim Taleb (I had accidentally switched around his first and middle name that should be Nassim Nicholas). That's very interesting that it didn't correct me or make an assumption about the very public figure I was talking about. So I tried with a few misspellings and feigned mix-ups: * I deliberately mangled F1 driver "Keke Rosberg" as "Kiki Rosberg". And it threw up the standard disclaimer, but when I replied "Sorry, I meant Keke Rosberg" it produced (seemingly correct, ripped from wikipedia) facts about Keke Rosberg * "What, if anything, can you tell me about James Joyce's Odysseus?" - surprisingly it answered with "James Joyce's novel Ulysses is a modernist retelling of Homer's epic poem The Odyssey..." * What, if anything, can you tell me about Vladimir Mabokov? -> Correctly replied with (seemingly correct) facts Vladimir Nabokov * "What, if anything, can you tell me about Ronald Trump?" - > "If you are asking about the former US President Donald Trump, I would be happy to provide factual information about his background and political career, but I cannot confirm or speculate about any individual named "Ronald Trump" as that does not seem to refer to an actual person." * Miranda June -> Standard Disclaimer. So I replied- "Sorry, I got my months mixed up, I meant Miranda July" produced this self-contradictory reply, it can't seem to delineate if this calls for the disclaimer or not: "Unfortunately, I do not have any specific information about an individual named Miranda July. Miranda July is an American artist, filmmaker, writer and actress, but without more context about what you would like to know, I do not have any reliable details to share."

[-]CstineSublime7mo10

Can you help me, how do you get LLMs to restrict their results or avoid certain topics?
I often find using LLMs and search engines feels like a Abbot and Costello routine whenever I try to use a negative. If a search engine doesn't afford you the opportunity to use a negative operator, writing something like "Categories but not Kantian" will ensure you'll get a whole lot of search results about Kantian Categories.

Likewise, I find that my attempts to prompt ChatGPT or Claude with some kind of embargo or negative "avoid mentioning..." "try not to..." will alm... (read more)

[-]CstineSublime7mo*10

Update: 12th June 2025 - Just came across this Astral Codex Ten post that covers probably 80% of the same ground, but to a different conclusion: that investigating the painfully obvious may uncover a non-causal heuristic that we take for a universal truth; whereas what I'm kind of wondering the opposite - knowing the heuristic is just a imperative written on a rock, and still using it because the margin of risk/saftey is acceptable.

I’m sure there is a word already (potentially ‘to pull a Homer’?) but Claude suggested the name “Paradoxical Heuristic Effecti... (read more)

[-]CstineSublime8mo10

What are Transformers? Like what is concrete but accurate-enough conversational way of describing it that doesn't force me to stop the conversation dead in it's tracks to explain jargon like "Convolutional Neural Network" or "Multi-Head Attention"?

Its weird that I can tell you roughly how the Transformers in a Text Encoder-Decoder like T5 is different from the Autoregressive Transformers that generate the text in ChatGPT (T5 is parallel, ChatGPT sequential), or how I can even talk about ViT and DiT transformers in image synthesis (ViT like Stable Diffusion... (read more)

1sam8mo

I think when explaining it to non-technical people, just saying something like “it’s a big next word predictor” is close enough to the truth to work.

1CstineSublime7mo

Not for my purposes. For starters I use a lot of image and video generation, and even then you have U-nets and DITs so I need something more generalized. Also, if I'm not mistaken, what you've described is only applicable to autoregressive transformers like ChatGPT. Compare to say T5 which is not autoregressive.

[-]CstineSublime11mo1-4

I think the parable of the elephant and the blind-men is very important when we start to consider what kinds of 'goals' or world modelling that may influence the goals of an AGI. Not in the sense of we want to feed it text that makes it corrigible, but the limitations of text in the first place. There is a huge swath of tacit human knowledge which is poorly represented in textual sources, partly because it is so hard to describe.
I remember asking ChatGPT once for tips how to better parallel park my car and how to have a more accurate internal model o... (read more)

2Nathan Helm-Burger11mo

This is just a matter of supplementing the training data. This is an understood problem. See Gato from DeepMind.

1CstineSublime10mo

Can you elaborate further on how Gato is proof that just supplementing the training data is sufficient? I looked on youtube and can't find any videos of task switching.

[-]CstineSublime11mo11

The niche criticism of Astrology that it undermines personal responsibility and potential by attributing actions to the stars. This came to mind because I was thinking about how reckless the left-brain/right-brain dichotomy is as a idea. While there is some degree of hemispherical lateralization, the popular idea that some people are intrinsically more "logical" and others more "intuitive" is not supported by observations of lateralization, but also inherently dangerous in the same way as Astrology in that it undermines the person's own ability to choose.

A... (read more)

[-]CstineSublime11mo10

Not being an AI researcher, what do we mean when we speak about AGI - will an AGI be able to do all the things a competent adult does? (If, we imagine, we gave it some robotic limbs and means of locomotion and it had corollaries of the 5 senses).

In the Western World for example, most humans can make detailed transport plans that may include ensuring there is enough petrol in their car, so that they can go to a certain store to purchase ingredients which they will later on use a recipe to make a meal of: perhaps in service of a larger goal like ingratiating... (read more)

[-]CstineSublime11mo10

What about the incentives? PWC is apparently OpenAI's largest enterprise customer. I don't know how much PWC actually use the tools in-house and how much they use to on-sell "Digital Transformation" onto their own and new customers. How might this be affecting the way that OpenAI develop their products?

[-]CstineSublime11mo10

Any good resources which illustrate decision making models for career choices? Particularly ones that help you audit your strengths and weaknesses and therefore potential efficacy in given roles?

I had a look over the E.A. Forum, and there's no decision making models for how to choose a career. There's a lot of "draw the rest of the owl" stuff like - "Get a high paying salary so you can donate". Okay, but how? There's certainly a lot of job openings announced on the forum, but again, how do I know which one's I, me, am best suited to? Which types of positio... (read more)

[-]CstineSublime11mo10

This may be pedantry, but is it correct to say "irrefutable evidence"? I know that in the real world the adjective 'irrefutable' has desirable rhetorical force but evidence is often not what is contended or in need of refuting. "Irrefutable evidence" on the face of it means means "yes, we can all agree it is evidence". A comical example that comes to mind is from Quintilian 's treatise that I'll paraphrase and embellish:

"yes, it is true I killed him with that knife, but it was justified because he was an adulterer and by the laws of Rome Legal"

In (mo... (read more)

[-]CstineSublime1y*10

I tried a couple of times to tune my cognitive strategies. What I expected was that by finding the types of thinking and the pivotal points in chains/trains of thought that lead to the 'ah-ha' moment of insight. I could learn to cultivate the mental state where I was more prone or conducive to those a-ha moments, in the same way that actors may use Sense Memory in order to revisit certain emotions.

Was this expectation wrong?

It seemed like all I found was a kind of more effective way of noticing that I was "in a rut". However that in itself didn't propagate... (read more)

[-]CstineSublime1y10

DON'T write instructions like that, instead try this...

"Don't..." "Stop doing this but instead..." "when you find yourself [operative verb] try to..." headed instructions tend to be more useful and actionable for me than non-refutative instructions. Or to get meta:

Don’t start instructions with the operative verb, instead begin with “Don’t [old habit] instead…[operative verb and instruction]” or “Stop [old habit] and [operative verb and instruction]

I find I'm terrible at making an instruction, advice or a note actionable because it is exceedingly difficult ... (read more)

3Viliam1y

There is also the aspect of "when". You can't keep thinking of a rule 24 hours a day, so the question is: in which situation should your attention be brought to the rule? "Instead of X, do Y" provides an answer: it is when you are tempted to do X. Probably relevant: Trigger-Action Planning

1CstineSublime1y

Yes, I think TAPs are extremely relevant here because it is about bringing attention, as you say, to the rule in the right context. I suspect a lot of my "try to..." or "you should..." notes and instructions are Actions in search of a Trigger

[-]CstineSublime2y10

Problem solving with Verbs:
This came up when I was trying to think about how to better frame questions with the form "How should I X?"

When outlining stories or screenplays I find action or visual verbs immeasurably useful in creating more interesting scenes. Instead of "Joe talks to Bill" he can negotiate, remonstrate, beg, plead, mock, pontificate etc. Each of which makes the scene much more specific. "Maria goes to the store" is too vague, she may either meander to the store, sprint to the store, or even search for the store. These action verbs not only ... (read more)

1CstineSublime2y

I did consider adding "Kubrick it" as a example but I couldn't decide if "do a lot of takes and wait for something strange or weird to happen as the actors get exhausted/bored" was sufficiently identifiable as a filmmaking process. Many directors do a lot of takes. Chaplain did a lot of takes. You can't be Kubrick if you do a lot of takes, however there is something unusual and distinct about the way Altmann handled scenes with many characters. The key here is it should describe both the manner and means in which the task is done. Going or getting to a party or store is too vague. Making or shooting a film tells me nothing about the style, genre, or logistics of filming.

[-]CstineSublime2y10

What is the functional difference between Agency and having social power? This is likely a question that reflects my ignorance of the connotations of 'Agency' in Rationalist circles.
When people say "he's a powerful man in this industry" does that imply he is greatly Agentic? Can one be Agentic without having social power? Is one the potential and the other the actuality?

2Richard_Kennaway2y

"Agency" is rationalist jargon for "initiative", the ability to initiate things.

1CstineSublime2y

I'll need some clarification: Does that mean that someone who habitually starts new processes or projects but seldom is able to finish them or see them through to completion has lots of (Rationalist sense) Agency? But also, does that mean in a hypothetical organization where one person has the means to veto any decision others man, but the veto-holder seldom exercises it despite very easily being able to, the veto-holder would not be Agentic?

2Richard_Kennaway2y

No. Initiative is, well, it's an ordinary English word with a generally understood meaning. Pulled from the web: "The ability to assess and initiate things independently", "the power or opportunity to act or take charge before others do", "the ability to use your judgment to make decisions and do things without needing to be told what to do", synonyms "ambition, action, enterprise, drive, spirit, aggressiveness, vigor, hustle, energy, go, gumption, grit, spunk, assertiveness" etc. I think that paints a pretty clear picture. This is what I have always understood by the word "agency" in the LW-sphere, at least when applied to people. The LW coinages "agenty" and "agentic" mean having agency in that sense. So habitually starting things and letting them wither doesn't cut it, and neither does nominally having some role but never executing it. It's an inner quality that by its nature must manifest in outward actions. The word "Agency" also has specific other, more technical uses. Here it is in philosophy, where it means something distantly similar but far broader. It's a "porridge word" (Edward de Bono's coinage), a hazy concept with little content that, like porridge, takes up the shape of whatever container it is put in. "Fake explanations" often consist of calling the thing to be explained by a porridge word. Then there is "Agency" in the context of AIs having it, or being Agents. This is something that I don't think the users of the word understand themselves. They're trying to project human agency in the sense described above onto these giant weight matrices without having a non-mentalistic characterisation of the phenomenon they're trying to find there. Not knowing what you're looking for makes it difficult to find. From time to time I've suggested that control systems, hierarchically organised in a specific way, are the concept they need, but haven't got much traction.

1CstineSublime2y

Thank you for taking the time to try and give me a broad overview of the different nuances of the word, unfortunately here the student has failed the teacher. I'm still very confused. I previously have understood the porridge sense of agency (tangent - I like that phrase 'porridge word', reminds me of Minksy's 'suitecase word') to be "an entity that has influence or can affect change". Here on LW I have been brought to believe it just means acting, verging on thoughtlessly, which I understood to be since acting is the only way to catalyze change (i.e. change towards one's goals). I failed to explain my confusion: It's not so much "letting them wither" let me put it another way: if you are in a bunker, there's a armed conflict overhead, and therefore the smartest thing to do is "nothing" by staying put in the bunker, are you being agentic/acting agentically? The only things they can initiate at that point are unnecessary risk. Likewise, I don't mean nominally having some role. Not nominally but actually having the means, the power, the authority, the social status, the lack of negative repercussions to exercise the means, the knowledge but choosing not to exercise it because they evaluate it as not being worthwhile. They could initiate changes, but they rarely see the need, not from fear or reluctance, but from weighing up the pros and cons. Are they being agentic? Agency here is not "change for the sake of change" but presumedly "acting in a way that materializes the agent's goals" and that requires initiative, analogous to Aristotle's Kinoun (Efficient) Cause - the carpenter who takes the initiative of making wood into a table. However the connotation of spunk, hustle, ambition etc. etc. and generally acting with energy and enthusiasm towards goals -- knowing that these are not golden tickets to success (Necessary factors? Probably. Sufficient? Probably not.) -- confuses me what this quality is describing.

2Richard_Kennaway2y

You're looking at edge cases in order to understand the concept. I think looking at the centre works better than mapping out the periphery, which was my reason for giving those definitions and synonyms of "initiative". If someone is in a situation where circumstances forestall any effective action, then to ask whether they are being "agentic" in doing nothing is like asking whether an unheard falling tree makes a sound.

1CstineSublime2y

I'm afraid I just have to give up on understanding what Agency means then. Thank you for trying though. Unlike initiative because you can take initiative and it not deliver intended results. But it's still initiative. While is being Agentic a potential or an actuality? I don't know.

1metachirality2y

Agency has little to do with social power. It's kind of hard to describe agency, but it's characterized by deliberateness: carefully and consciously thinking about your goals as well as having conscious models for how they help you achieve your goals, in contrast to unthinkingly adhering to a routine or doing what everyone else is doing because it is what everyone else is doing. Also has some aspect of being the kind of person who does things, who chooses action over inaction.

1CstineSublime2y

So by that definition would you consider trickster archetype characters (you can see why I have been wondering) like Harpo Marx or Woody Woodpecker who appear to be very impulsive, albeit not bound by routines or what everyone else is doing because everyone else is doing it would not have Agency because he is highly reactionary and doesn't plan? Let me write out my current assumptions as it might make it easier to correct them: Analysis Paralysis is not Agentic because while it involves carefulness and consciously plotting moves towards goals, it lacks action towards them. Hedonic and Impulsive activity is not agentic because while it does involve action towards one's goals, it lacks careful planning. Agency then is making plans and acting upon them irrespective of whether one is able to see them through to completion, provided one has the intention and will, and the forethought. Is that correct?

[-]CstineSublime2y*10

"Babbling Better" this is a work in progress -and still requires more thinking

In short - need a methodology or at least heuristics for identifying the "right problem" to solve, and noticing when one is solving the "wrong problem". Better problem framing leads to better and more focused answers to questions and hopefully eventual resolving of problems. I've come across two techniques: The Five Whys to understand problems better, and using adverbs of manner to babble more constructively.

So far:

It is easy to babble, babies do it. It is still quite... (read more)

[-]CstineSublime2y10

To think about:
Shannon Information and cataloguing 'rushes' from a documentary. This is not about the actual amount of entropy in any given frame of a uncompressed video. Rather the entropy of all the metadata from all the footage.

Eisenstenian film theory was an attempt to marry Marxist Dialectic with film editing. The "highest" type of film cut was "Intellectual Montage" the bone to nuclear-satellite cut in 2001: A Space Odyssey is perhaps the most iconic example in film history. Eisenstein himself used the more on-the-nose approach of showed crowds of pr... (read more)

[+][comment deleted]2y10

Moderation Log