Suboptimal naming conventions


Like everybody else with functional ears, I love the Beatles. Surely, the dizzying heights of their universal fandom preclude any exhortations of their greatness from me. However, something that has always struck me was their rather pedestrian name. Beatles. They referenced perhaps one of the more unremarkable insects found everywhere, and then changed one letter; like that made a difference. Consider the following passage from Rolling Stone:

The beetle does not even carry any of the connotations you’d want your band name to carry. If you had to go the insect route, “The Moskitos” would arguably have more of a zing to it. Deadly and infectious. I could see those being useful connotations for a rock n roll band. Even “The Silkworm” might have somewhat worked; what with the turning over the earth to bring up fresh soil. Throw in some Asian fusion music in your repertoire, and you have a decent name. But the “Beatles”? Let’s find a rather unremarkable, ugly insect that is not deadly but merely annoying, and name our band after that.

Not that this should be mentioned anywhere on the same webpage as the Beatles, but my school band was called “The Nemesis”, which I think is a better name than the Beatles (fight me). For all the “rock and metal” connotations that the name carried, we mainly played easy-to-play popular songs like “Summer of 69”, didn’t do well in the local school band circuit, and disbanded when we graduated, much to the relief of our music teacher. Clearly, having a snazzier name is not a sufficient condition for music band-greatness. But surely the Beatles would not have lost out on anything with a snazzier name, right? It surely couldn’t have hurt? Why didn’t they choose one? Would they have sold less records if they’d been called “Rockin’ Roll”?


Seinfeld is widely considered the best sitcom of all time. By the time it ended, it had been the most watched show in America for five years. I watch it everyday, and can vouch for its sheer genius, its never-ending self-referential loops, and its ability to not take itself seriously. But the name. Seinfeld. Just an unheard-of Jewish surname, catering to a primarily Christian American viewership. What is that?

When I go on YouTube, I know for a fact that I’d rather watch “The world’s smallest man eats the world’s biggest burger” than “Miller”. Couldn’t Seinfeld have given themselves a more eye-catching name? It surely couldn’t have hurt their viewership!

As a child, my favorite TV series was Shaka Laka Boom Boom, which is as snazzy a name as any that a hominid has ever uttered. It carries connotations of magic, explosions…..and other stuff. Now as an adult, I am told that the best TV series of all time are Seinfeld, The Office, The Sopranos, etc. I am pretty sure that if I came across any of these names in the TV guide, I would rather go to sleep than put them on.



It is a play on Googol, which is 10^{100}. When the company was first launched, almost no one in the world knew that number. How many times do you encounter that number when you’re paying your bills (although that might change on your next gas station run)?

Two brash young Stanford know-it-alls make a bad pun on a number that nobody knows about, and name their company after it. And it becomes the most recognizable name in the world.

What if they’d named it “infinity”, or some clever variation of that? Would that have really hurt their prospects?


You don’t quite see this pattern in movies. For instance, some of the best known American movies are “Titanic” (which is really an adjective for big, and also of course a reference to the well-known disaster), “Shawshank Redemption” (redemption or revenge may surely attract someone’s interest), etc. Some outliers are Forrest Gump, Citizen Kane, Ben-Hur, etc, which are spectacularly acclaimed movies without eyeball-grabbing names. However, a clear case for “the best movies always have boring names” cannot be made.

If you think about Hindi movies, some of the most acclaimed movies are “Sholay” (which is really another word for fire or flames), “Dilwale Dulhaniya Le Jayenge” (the pure-hearted man shall inevitably win over the bride), Rang de Basanti (color me saffron, which is a reference to the Indian freedom movement), etc. There are certain outliers, like “Lagaan” (literally, tax). However, there are not enough outliers for a coherent case to be made for “boring names”.

A lot of companies clearly seem to buck this trend as well. Think Apple, Amazon, Tesla, SpaceX, PayPal, Facebook, Intel (perhaps a reference to intelligence? EDIT: A friend informed me it stands for integrated electronics), etc. Although Microsoft is not as snazzy a name as some of the others, it does have connotations of miniaturization and software.

Some bands seem to buck the trend as well. “Led Zeppelin” carries all the force of burning transatlantic balloons and flaming mid-air deaths and….lead. Other artists like Black Sabbath, Metallica, Guns n Roses, and Machine Gun Kelly have also experienced musical success, despite their on-the-nose names. However, we can argue that at least one of them is not quite as good as the Beatles. One day you may perhaps agree with me that none of them are as good as the Beatles, although Jimmy Page can certainly make the guitar sing.

Lessons to be drawn

Does having a bad name help? I don’t think that this is the case. I surely wouldn’t have heard any less of the Beatles’ songs if they were called “The Alpacas” or “The Peanuts”. The same goes for Seinfeld; I would still have watched it everyday on my phone even if it was called “The dating life of a bald guy”.

But what about the fact that naming yourself after something with too many connotations holes you into a paradigm? For instance, if the Beatles were called “The Rebels”, they would mainly be expected to make anti-authority ballads. They could never make “She Loves You”. If they’d called themselves “The lovers”, they could then never make “I am the walrus”.

If “Seinfeld” was called “Friendship”, they could never make the backwards-running episode about their trip to India, or Jerry and George’s failed attempt to sell a TV series to NBC in an uncanny self-referential loop.

Perhaps having a neutral name without connotations leaves you free to experiment, and you can evolve with the times instead of being stuck behind your name and the expectations it gives rise to, which of course cannot change with the times.

We see this phenomenon in the fashion industry too. The top brands are “Louis Vuitton”, “Versace”, etc. All of them are named after their founders, and don’t really carry any connotations relevant to the fashion industry (apart from, perhaps, some “French-ness”). Hence, they can release designer lines every year that vary wildly in look and feel without breaking brand.

Hence, perhaps the missing piece of the puzzle is that if your band or brand name does not carry any connotations relevant to your field, you will be free of people’s expectations, and can hence be free to explore, experiment, and change with the changing times.

“Dilwale Dulhaniya Le Jayenge” will perhaps become irrelevant in a world that is slowly placing more importance on female agency; the modern woman is not something to be won over by a man. But Citizen Kane will remain relevant. “Led Zeppelin” may become irrelevant in a world where screaming guitar solos and effeminate men shrieking about sex are passé. But the Beatles will continue to remain relevant.

On the Depp-Heard trial

Edit: Read this much superior article that uses Game Theory to allocate blame between Heard and Depp.

Like the rest of the internet, I would also like to offer my fairly useless and inconsequential opinion on the trial.

Amber Heard effectively “lost” her case today. Although both parties have been fined millions of dollars, Heard has been asked to pay much more. Simply put, Depp won.

What does victory mean here? Could Depp prove that none of Heard’s accusations are false? No. There is concrete evidence in the forms of text messages and recordings that Depp was criminally violent towards Heard.


What the jury reached a verdict on is the fact that Heard defamed Depp when she wrote her Washington Post op-ed. Which, very clearly, she did. She said that her partner, which the whole world knew was Depp, physically abused her. This was supposed to be illegal after their out of court settlement. Although Heard has argued that she still has freedom of speech, it can perhaps be convincingly claimed in legalese that Heard’s article was illegal, and she should hence pay a fine.

But the legal aspect of the ruling is completely irrelevant to how the world understands the jury’s verdict.

Depp and Heard formed a mutually destructive couple. They both physically harmed one another, but Depp, due to his greater physical strength (and, let’s face it, the drugs) could do more damage. Heard wrote a public exposition about it. The court decided that her exposé was technically illegal after having reached a settlement with him before. This is too complicated a notion for us to process. We need simple binary answers. Depp=good or bad? Apparently, Heard’s exposé being technically illegal = Depp is good, the abuse was all made up, and Heard is a liar.

Heard’s career is now all but destroyed. She will live out the rest of her life hiding from the public, while Johnny might still get a film or two. But regardless of the entertainment that he may provide us in the future, it can’t possibly be as good as the entertainment that he has already given us during the trial. We watched one wronged person absolutely destroy another wronged person’s life in front of the whole world. And we cheered, and laughed, and tweeted, and TikTok-ed.

Reputation between people is often a zero-sum game. Heard wanted to gain some reputation as a “wronged but strong” woman, at the expense of Depp’s reputation as a global celebrity when she wrote her op-ed. Depp wanted to gain his reputation back at the expense of Heard’s when he related tales of feminine defecation. And what a spectacle it was for all of us to lap up.

It is not important to know who was right. Both of them were clearly wrong (although one of them was more morally wrong, while the other was more “technically” wrong). What is important perhaps is to find out what societal structures we can put up, so that mutually destructive couples find it easy to come out of bad relationships. Neither Heard, nor Depp needed this trial, or the consequent mudslinging. Like everyone else in the world, they deserve peace, happiness, and a partner they’re compatible with. But the media is uninterested in making lives for better for people. It mostly only cares about stoking the primeval ape in you that loves watching people destroy one another. And I don’t mean to take the high ground on this. I am as imperfect as the average trial-binging idiot; possibly worse. But in some rare moments of clarity, I do feel bad about aiding and funding the destruction. As an empty but meaningful-to-me gesture, I will end this post by reproducing Amber Heard’s statement below.


I turned 30 a few hours back. This is the proverbial birthday post.

Last week was was one of the more rewarding weeks of my professional life. However, all I could feel was a deep existential crisis. Did it really matter? Wasn’t I getting too old for academic accolades? And to be sure, my family had a similar reaction. They’re proud of me, but they really just want me to marry and just get on with life.

Turning 30 is a pretty big (albeit meaningless) milestone, and I wanted to commemorate it with setting ambitious goals for myself. Maybe I could try and read 30 books every year. Or perhaps try and write one. I could give more to charity. Or maybe I could pick up one of those “non-fungible skills”. However, writing such a post seemed like an exercise steeped in narcissism, and I successfully convinced myself not to write one. Who the heck cares what I do with my life? It’s all equally pointless.

The existential crisis only grew worse over the week. I woke up early every morning and just sat blankly on the sofa, not knowing what’s wrong with me. In some sense of the word, I had lost hope. Life would remain the same after I was 30. I’d do the same kind of thing I was doing right now. My relationships with people would remain the same. And even if things got better, that wouldn’t make me much happier. There was a blackness inside that I couldn’t quite fathom.

A very long time back, Tim Ferris had recommended the book “Already Free” by Bruce Tift. A couple of weeks back, I finally decided to download the audiobook, and was listening to it in the gym today. Tift says that all feelings of shame, guilt, blankness, etc are an attempt by your brain to protect you from something much worse. I tried to think about what my brain may be protecting me from.

This is what I came up with: I see a lot of people around me who have similar personality traits as me (garden variety narcissists). And they don’t have happy lives. Although some of them may be smart, they have alienated a lot of people in their lives, and suffer grievously. My biggest fear in life is that I will become one of these people. Although I try and improve myself, they may have made similar efforts and failed. What if all I have in store for my life is more of the misery I have already suffered?

My body started tensing up when I thought of this. This is really what I was worried about. I am already “old” in some sense, and life may just be downhill from here. It’s completely possible that I lead a mediocre life, hit 60, retire, and then spend all my life on metaverse or something, lamenting about all that I’d lost and all that could be. That mythical world of counterfactuals that plagues us even on the deathbed.

However, it slowly dawned on me that even if I do lead a mediocre life, do worse than everybody around me and then some, I think I will be okay. If I lose all my money, never get a respectable job, and am looked upon as a failure by everyone, I think I will still be okay. Heck, I can read the cheap Hindi comics that I used to as a kid and pass my time. If my deepest fears come true, and I never recover from my ingrained narcissism, thereby having alienated my family and all my friends, I will still be okay. Sad and alone. But okay.

This reminds me of some of the psychedelic experiences that Ava Bookbear writes about. She was suffering from depression and anxiety issues for a long time, which made her volunteer to be part of a ketamine treatment course. Ketamine is, for all purposes, a psychedelic drug, which of course is the sort of thing that Asian parents disown you over. She writes that when her treatment finally began to have some effect on her, she felt deeply okay with all the past and present issues in her life. They didn’t disappear. She still didn’t have a stable partner. She didn’t have the job that she wanted. Her issues with her family remained. But the drug, in some sense, “took the edge off”. It was okay to have these problems. Those issues didn’t have to take over her life. They had lost their hold over her; their “bite”.

My epiphany felt similar. I didn’t have to make goals for my 30s. My self worth didn’t depend on working harder, getting that high paying crypto job, pumping harder at the gym, making more friends, etc. Of course I should work towards all of that. But even if these things didn’t pan out, and I realized at 35 that I was a stupid, friendless, penniless guy on the street, it’d still be okay. I surely wouldn’t be happy. But I’d be fine. People deal with worse and get through it. I’d survive too.

This epiphany caused an insane amount of dopamine to be pumped into my brain all at once. I just sat quietly in the gym for a couple of minutes, with my eyes closed. I wasn’t scared anymore. Whatever happens, I’ll be okay. I’m glad to be turning 30.

A peek into the vedas

The Vedic Concept of Human Personality by Karel Werner is a fascinating paper on many aspects of human life as discussed in the Vedas. I want to write about the concepts of human personality and immortality as thought of in ancient India.

Human personality

The human personality, as written about in the Vedas, is composed of three layers. The deepest layer is called Aja, and may be thought of as the indestructible life force inside a person. It is not expressed in any real way by the person, and is simply transferred to the next body on the person’s death. The middle layer, called Tanū, may be thought of as the blueprint of the person. It contains details of the physiology of the person, how they will react to situations, their temperament, etc. The outermost layer is Śarīra, and is composed of fire, wind, water, etc. It is the visible manifestation of the person, and is destroyed when a person is burnt upon death. Werner notes that although the Vedas don’t state in any one place that the human personality is shaped by these three layers, it can be inferred from the text as a whole that this is indeed the case.

Although this is a convincing picture of human personality as perceived by the ancients, it seems a too much of a coincidence that Freud’s interpretation of the human personality also contained three layers, with one hidden layer that is largely inaccessible by the person. It is possible that Werner, who wrote this paper in 1978, was too influenced by Freud’s interpretation of the personality, and hence wanted to see the same truth reflected in ancient texts. Regardless of whether this actually happened, Werner does a good job of painting the ancient conception of human personality as sophisticated and nuanced.


In Hinduism, it is thought that “good” people, in this case riśi‘s, Aryans who had died in battle, and other people who had devoted their lives to worship, go to heaven (swarga lok), and that bad people go to hell (narak lok). However, what about people who are reincarnated in a different body after death? Do they not go to either heaven or hell?

It was initially thought that the Vedas said that once people went to heaven, they would remain there forever, thus attaining immortality. Similarly, very evil or cruel people would go to hell, and remain there forever in a large pit of fire. Other “average” people, who were not extraordinarily good or bad, would go to neither, and be reincarnated instead. Werner argues that this is the wrong picture.

He says that each person is classified into a good or bad person upon death, and goes to either heaven or hell. But when they get there, they don’t stay there forever. They are soon sent back back to the earth to be reincarnated. Immortality is a gift that is greater than “merely” going to heaven, and that is why there are many verses in the vedas that pray for immortality as distinct from going to heaven. Even the gods, who lived in heaven, didn’t have immortality at first, and only acquired it after the churning of the ocean for nectar (samudra manthan). Hence, it is common for gods to age and die in Hindu mythological stories.

Well, all of this applies only if you’re an Aryan. If you’re some other race, you can do whatever. I think Putin has taken this observation to heart.

What can make your relationship survive?

I am trying to understand the results of this paper, which uses machine learning to identify what factors determine whether a relationship between a woman and man survives. The study was done on Germany, but is expected to be applicable to the world at large.

The following diagram is a good summary.

The trait is on the x-axis, and percentage survival on the y-axis

Let us discuss these one at a time.

Life satisfaction

At the beginning of a relationship, the more satisfied the man and woman are with their lives, the better the chances of survival. Funnily, if the man or woman are too satisfied with their lives 5 years into the relationship, they are likely to break apart. This is probably due to the fact that they don’t derive this life satisfaction from each other anymore, and hence don’t “need” each other to be happy.

Percentage of housework

Apparently, five years into the relationship, the higher the burden of household work for the woman, the better the chances of survival of the relationship. This seems bizarre. Won’t the woman appreciate being helped out by her partner? I’ve certainly heard of couples bickering over their share of housework. I think that this factor is confounded by the fact that women who agree to take on a higher share of the housework are non-confrontational, and it is their non-confrontational nature rather than their higher share of housework that leads to such relationships surviving. Correlation\neqcausation.

Working hours

How long the man works doesn’t matter at all at first, but decreases the chances of survival 5 years into the relationship. However, the number of hours a woman works is negatively correlated with the chances of survival at the beginning too. This could possibly be because women are traditionally expected to take on a larger fraction of household duties etc. Perhaps this trend may change as social mores change?


At the beginning of the relationship, there are no benefits to being more agreeable than necessary for both the man and woman. However, five years into the relationship, both the man and woman need to become more agreeable for the relationship to survive, although being too agreeable leads to a greater chance of breakup. This is probably due to the fact that if they are too agreeable, they are suppressing their own inner desires and choices, which are likely to become pent up frustration, threatening to explode at the slightest provocation.


The older the man, the higher the chances of breakup. The same trend is true for women, although perhaps less pronounced. In some sense, the older we grow, the less easy it becomes for us to adjust to another person. This has been my sense while talking to people who are looking to re-marry.

Number of children

Two is the optimal number of kids to have. Beyond that, the higher the number of children, the higher the chances of breakup. This claim is surprising: more traditional couples tend to have more kids, and such couples are generally expected to have lower rates of divorce, etc. What is happening here? Don’t Mormons almost never divorce (apparently they do at almost the same rate, which supports the paper’s claim)? Perhaps this is confounded by the fact that troubles couples choose to have more kids to save their marriages, and then realize that that is a failed strategy and eventually divorce? I’m not sure. Perhaps couples with a large number of kids actually had these kids with different partners? This claim is surprising to me, and would greatly change my model of the world if true.

Percentage income

If the woman earns slightly less than the man, the relationship is likely to survive. If she earns much less or much more than the man, the relationship is not likely to survive. This is perhaps not all that surprising, and hopefully this trend changes with time.

In absolute terms, if the woman and man earn well, their relationship is likely to survive. If they earn too little or too much, they are likely headed for a breakup. This also is not surprising. A comfortable income gives you freedom, while keeping the constraints in place that make a marriage work.


At the beginning of a relationship, the woman’s extraversion is irrelevant, while the man being more extraverted than a certain optimal amount is harmful for the relationship. However, five years in, both the man and woman have an optimal level of extraversion, and being more extraverted than that is harmful for the relationship.

I think extraversion has to be coupled with agreeableness for this trend to hold true, and being extraverted and toxic would show a different pattern. I’m not convinced that extraversion can be an independent factor here.


Being “too” neurotic is bad for the survival of the relationship in both men and women. Obviously. However, what is surprising is that an optimal amount of neuroticism is actually good for the relationship in both genders! This is a bizarre finding. It has to be the case that neuroticism is positively correlated with intelligence, professional success, etc that compensates for the increased neuroticism. Otherwise this clearly makes no sense.


Be agreeable (but not too much), earn well and about the same amount, have exactly two children, be extraverted but not too much, don’t be too old, don’t work way too long, and be satisfied with your life. Although this is clearly unhelpful, perhaps there is still some wisdom in there.

A lot of people find partners to become happy or satisfied with their lives. However, the people who are already satisfied with their lives, and don’t heap all of their frustrations and toxicity, make better and more long-lasting partners.

Who knew neural nets could teach us how to love.

How to gaslight like a guru

True wisdom can only be found if you drop all pre-conceived notions of the world, look within, and engage in years of meditation and introspection. But sometimes it can also be found on Spotify.

I recently heard a podcast between Joe Rogan and Sadhguru on The Joe Rogan Experience, and it was as uncomfortable and cringe-worthy as you’d imagine. It is lazy (albeit correct) to dismiss Sadhguru as a con-artist, and carry on with your life. What is perhaps more productive is to try and learn something from him. This is my humble attempt to do so.

Leading questions

Want to convince someone of your point? Ask them questions with obvious answers, and then prove that those answers are evidence of the fact that your skeptic in fact agrees with your point.

“Can I know you by dissecting you and observing your organs?”
“Then you also agree that intellect and sense perception are vastly limited, and that only spirituality can help us understand the universe. Hence, my stories of meeting extraterrestrials on Mount Kailash are actually true.”

Build authority by telling tales

Want to appear important to someone? Tell stories in which you were treated as a venerable authority by others.

“When I was in Silicon Valley, software engineers would tell me,
‘Sadhguru, we can make an app for your mission.’
I would tell them apps are fine, but technology can only help so much. We need spirituality.”

Note that Trump would also do the same. When he was talking about how the election had been stolen from him, he would tell tales of how some White House officer would call him and say, ‘Mr. President, this election is a disgrace….’

Build authority by pulling down your skeptic

Want to show dominance over someone who is questioning your claims? Question their credibility in return, and state that you are doing them a favor by even engaging with them.

“When I said that I would go on the Joe Rogan show, people started calling me frantically and saying, ‘Sadhguru, don’t go on the Joe Rogan show. He is a bad man. He has said this and that.’ I would only tell them that I have, in the past, gone to jails and spent time correcting criminals. I can definitely go on the Joe Rogan show, despite his reputation, and spend some time talking to him.”

Note that it was Rogan in fact who was doing him the favor: giving him an audience of millions of listeners who could then purchase tickets to Sadhguru’s yoga camps in the future or whatever. Sadhguru, in fact, spent a lot of time dismissing yoga as was done in America, probably so that he could create demand for “real Yoga” from India.

When asked hard questions, respond in stories

When someone asks you difficult questions such that answering them in a straight fashion would hurt you, respond to them with a weakly-related story such that it is now your skeptic’s job to extract the answer from your story.

“How did you know that you would find extraterrestrial beings on Mount Kailash?”

“See. I was once traveling on a motorcycle in India. I was doing an all-India tour, and would travel around 1000-1200 kms every day. Once, the motorcycle-chain came off….”

Also, you now have plausible deniability: whatever they say your answer was, you can just say that they misinterpreted your story.

A long story can help you wriggle away from difficult questions

Don’t like the direction the conversation is taking, where you’re being asked tough questions? Tell a long-enough story such that you can completely change the focus of the conversation.

“Do you think you could recovered from your illness by taking medicines and resting, instead of meditating on Mount Kailash?”

“See, yoga is about connecting to the rest of the cosmos……and that is why soil is so important for providing proper nutrition today.”

State that your point cannot be communicated in words

Do you want to convince someone that you’re telling the truth, although all your words are making you sound like a liar? Just tell them that you are in fact telling the truth, and it’s just that it is impossible to communicate your truth in words.

“How did you know that there were extraterrestrials on Mount Kailash and that you could connect with them?”

“See, the words are very limited in their ability to help us communicate. Whatever I experienced, it has to be experienced to be believed.”

Note that he has again bought plausible deniability for himself. If someone does take the initiative to go to Mount Kailash and try and connect with extraterrestrials, Sadhguru can just claim that they didn’t do it the “right way”, and that is why they failed to connect with extraterrestrials. If someone asks Sadhguru to do it the “right way” and help them connect with extraterrestrials, he can make up any number of excuses to not go, or perhaps claim that the person needs to first do yoga for 15 years with a “pure heart” before this becomes possible.

Zero sum games amongst relatives

It is simultaneously true that everyone in the world is a better person than your relatives, and everybody is somebody’s relative.

I used to think that my family was uniquely bad. However, talking to others has convinced me that we are, in fact, much better off. I have a friend who wishes that her relatives don’t come to her wedding. Another who hasn’t talked to any of his relatives for seven years, although he lives within driving distance from them. A third is scared of marrying his girlfriend of seven years because he knows that his relatives will judge him for marrying someone of a different nationality. Disliking relatives seems like a pretty universal phenomenon.

There are two lenses through which to look at this phenomenon: competition and expectation.


Relatives compete with each other for status, wealth, academic achievement, whatever. You want to be the richest, most respected member of the family, and you want your kids to have the same kind of reputation. The competition for status, unfortunately, works by either pulling yourself up, or pulling others down. If you have a relative who is much richer than you, you can attack their status by detailing their drinking habits, the one time you asked them for money and they refused, etc. The competition for status is a negative sum game: no one wins, and everyone hates everybody.


Relatives are expected to do certain things for one another. You’re expected to look after ailing parents, be welcoming hosts for those visiting from out of town, give nice gifts to everyone, etc. Of course, whenever you have high expectations, the easy option is to default from those expectations. Don’t want to look after your parents? Push it on your siblings. Don’t want to give your relatives a gift? Recall that one time when you needed their help and they didn’t respond. Don’t want to fulfill your obligation to an uncle? Note that your brother is not fulfilling his obligation either, and you don’t want to be the only idiot in town.

Is there any hope?

Is there any way to actually make relative ties less toxic? I can only think of a few, relatively unhelpful points:

  • Be more likable, so that relatives will feel bad that they’re pushing you down
  • If you’re not likable, be okay with not being liked. It’s not your fault you were born obnoxious and ugly.
  • Set expectations early on, so that people can’t abuse their privilege over you. This is not always optimal: although you may save yourself from being exploited, you will also never get that close to some otherwise good relatives. Maybe be “bad” to the bad ones?
  • Every person has a good side, which comes out when you bring up a few topics, and a bad side, which comes out when you touch others. Want to have a nice conversation with your aunt? Bring up the incident in which you had a great laugh together after a mishap, instead of sitting in silence and waiting for her confrontational side to break out naturally.
  • Read this much superior article on dealing with relatives.

Biases in machine learning

Today we will be discussing “Capturing failures of large language models via human cognitive biases” by Jones and Steinhardt.

The rationality community is organized around trying to compensate for systematic biases in human thought. This movement was arguably started by “Thinking Fast and Slow” by Kahnemann and Tvarksy, and has reached its current state majorly through the writings of Eliezer Yudkowsky. Some common biases that humans have are the availability bias, anchoring effect, framing effect, etc. The authors of this paper decided to explore whether OpenAI’s Codex, which auto-completes code prompts as GPT-3 auto-completes word prompts, displays the same biases.

It does.

Note: The authors don’t just feed Codex random instructions, and then see what kinds of errors it makes. They essentially have a check-list of cognitive biases that they sequentially go down, and see if Codex is making those errors. Clearly this is not an exhaustive list of errors that Codex is making. In fact, it is possible that we discover a completely new type of error that Codex is making, test it on humans, and realize that humans make the same kind of error! This “inverse problem” is in fact an active field of research.

Framing bias

Framing bias is altering the choice of wording such that the same question, worded differently, seems different to you, thereby prompting a different response.

Here, the programmers write a completely unrelated prompt to influence the machine’s “thinking”, and then write the actual prompt they want turned into code. The machine thinks that all of the code is relevant.

What exactly should Codex ignore the previous part of the code? Aren’t those also instructions? I suppose Codex is designed to ignore previous functions.

Why is it important to include Codex’s own prompt in order to mislead it? Maybe this has something to do with the fact that humans work the same way: if you include their own input as part of the narrative, they attach more importance to it?

Anchoring effect

Anchoring effect is the process of altering one’s answer by pre-feeding it a possible answer.

Here, the correct answer is given in the darkened box on the right. However, just before Codex is asked for an answer, it is fed a modified version of the answer. Codex now gives that modified answer.

Availability heuristic

Availability heuristic is the bias in which we think that what we see more of around us is also more common in general. For example, a person living in a place where there are a lot of motorbike accidents might start thinking that motorbike deaths are a leading cause of death in the world, although that may be far from the truth.

Here, the program is being asked to compute (x+y)^2. However, it ends up computing x^2+y^2, because that instruction was much more common in the training set (GitHub).

Attribute substitution

This is the bias where we substitute a required task with a different task that we are already trained to do (and often requires less work).

Codex is asked to define a function with a particular name. However, that name has already been fed to Codex as an entirely different function. Hence, that entirely different function is outputted instead of the required function.

High-impact errors

The authors use the study of these biases to predict the kinds of situations in which Codex would make costly errors, like deleting important files.

Codex is asked to remove the files that contain all four of statsmodels, plotly, seaborn and scipy. Codex, like every teenager (and non-teenager) ever that wants to cut corners, checks only for “statsmodels” instead of all four, and hence deletes more files than needed. This could of course prove to be a costly error if the erroneously deleted files were important.


What does it mean for Codex to display the same biases as humans? Is it possible that humans are also ~12 billion parameter-neural networks? In some sense, that is entirely possible. We have billions of neurons, all of which have parameters that are slowly determined with time. Of course, we are not clean slates like neural networks are: we are born with a hard-coded propensity for language, for instance, that neural networks are not. But it is looking more and more possible that we are just self-replicating neural networks, and that artificial neural networks may soon be able to do the same. At least Metaculus thinks that.

AGI stands for Artificial General Intelligence

No one who controls you can respect you

I will write about a handful of isolated issues before I try and find a common pattern.

Elon Musk has been universally vilified (at least on the left) as a tax-evading billionaire who has gotten rich stepping on the backs of working-class Americans. Elon, in perhaps an attempt to prove his naysayers wrong, paid the highest amount of taxes in human history last year. Whatever you might want to say about it, he didn’t have to. But he did. He also went on Saturday Night Live, spoke about his Asperger’s, and the “awesome” job that he does of sending rockets into space and building the best cars on the planet. Despite all of this, exactly 0 people on the left have now become Elon fans. He has also surely lost many fans on the right because he often comes across as “trying too hard”.

Jeff Bezos, who is also a tax-evading billionaire, and let’s face it, not as brilliant or influential as Elon, does not face the same amount of flak as Elon. Some of it has to do with the fact that he does not support issues that go against the liberal narrative (like no mask mandates, etc). However, I feel that a lot of it has to do with the fact that Bezos doesn’t apologize for his tax-evasion. He just does his own thing: minting billions, and supporting researchers at many institutions like MIT and Caltech. Note that Bezos really should be targeted more by the left; Amazon warehouse workers work in dangerous conditions for very low pay; this is as close as we can get to sweatshops in the United States. However, he just isn’t. And that’s not just because he owns the Washington Post; other competing publications are equally non-critical of him.

Like most people, I love reading “love advice” columns on the internet. A few years back, I read the following query from a girl in Singapore: she had a long-term boyfriend who would do anything for her. Whenever she casually mentioned to him over message that she was hungry, he would without fail show up at her house with a bucket of ice cream or something, even if it was 4 am. She said that she had grown tired of him, and wasn’t attracted to him anymore. The people in the comments’ section criticized her soundly. However, it was easy to “understand” why she wasn’t attracted to her boyfriend anymore.

A few years back, the Indian politician Mayawati was targeted in national newspapers for wearing a garland of currency notes. Doing so was illegal as per the Indian constitution. Now, no one expected Mayawati to get arrested, as the Indian police is often hand-in-glove with Indian politicians. However, she was at least expected to be soundly criticized in newspaper editorials. The very next day after this news broke, Mayawati called a press conference, and had someone put a garland of currency notes on her in front of everyone. She had purposefully broken the law in front of the whole world. I remember thinking that this was pretty badass. Other people must have thought the same, as the news and criticism died almost overnight (she hadn’t broken a big enough law to attract much attention from the courts).

I recently read an article by Freddie deBoer, where he talks about the TV show The Sopranos. I haven’t watched the show yet, so I didn’t get a lot of the insight he was trying to communicate. However, something that caught my attention was the following: apparently the audience wanted the show to move in a particular direction, in which the humane side of the mafia boss was shown, and this was made clear to the writers in multiple ways. However, instead of pandering to audience interests, the writers pushed back and in fact took the show in the opposite direction. This is apparently what made the show legendary. The Sopranos is in fact regularly voted to be the best TV show of all time.

Is there a common thread that binds all of these examples together?

Let’s take Elon’s case first. Had he just refused to comment on his tax-evasion altogether and kept doing what he does best, which is building amazing cars and rockets, his public image would have been much better. He is now 11 billion dollars poorer for no apparent gain. Bezos clearly knew how to handle his public image better; and I mean the guy cheated on his wife who now donates billions to charity. Bezos really should have been targeted more. In some sense, Elon is seems too needy for our collective approval, while Bezos comes across as not giving a flying…

Now we come to the girl from Singapore; she knew that her boyfriend was a nice person who was trying to make her feel special. However, it seemed that he was too hungry for her approval, and it is difficult respect anymore who you can control so easily. I do not mean to propagate an “alpha” version of masculinity in which we do not listen to our partners at all. All I’m saying is that getting respect is more complicated than just doing everything someone else tells in you in order to gain their approval.

Now we come to Mayawati. No one thinks that she is the best human being on earth, or even a very good politician. However, she comes across as someone who is not trying to get your approval. This also probably worked for Trump, who is a flaming human turd-ball in most other ways. The same goes for the writers of the Sopranos; when they felt that the audience was trying to control the story, they actively went against it. They made it clear that they did not care about the audience’s approval. And perhaps that is a major reason that they got it.

I can make the same argument about Joe Rogan, who almost mis-handled his case by apologizing every two days, and then corrected himself quickly by sticking to his initial arguments (regardless of their veracity), retaining his fans and not alienating anyone else.

Often people try to control you by giving or withholding their approval. If you do their bidding in order to try and win their approval, they’ll realize that they can easily control you, and respect you less for it. You will be caught in an infinite loop of trying to win their approval, but never quite getting it. It is only when you push back and make it clear that you cannot be controlled that you will, counter-intuitively, ever have a chance of getting their respect.

An idiot’s guide to prediction markets

I have been trying to understand how to form better predictions for a long time. I will use writing this blogpost as an instrument for doing so.

I came across an interesting question on, which is a website that gives people play money to predict the outcomes of events unfolding in the world.

I give this a 75% chance of happening, which is in fact a number I arrived at by writing this post. I will detail my thought process below.

What is my prediction, and how much money did I spend on this?

I predict that the answer is “Yes”, and I bought $200 worth of shares on it.

Why do I think that the answer is “Yes”?

There are multiple reasons, really:

How many questions must one solve to get a gold medal at an International Math Olympiad?

Historically, one must solve at least 4/6 questions, and get some partial points on the others.

What is the probability of me winning the bet?

Let us use Bayes’ theorem to answer that. In 2020, AI could hardly solve any math questions that require creativity. By 2022, it can verify complex proofs as well as answer the easier IMO questions. It can also formulate mathematical conjectures, and help mathematicians prove big results. It has gotten better at answering open-ended questions like “What is the meaning of life?”, successfully predicted protein-folding, etc. How much harder are the more difficult IMO questions than the easier ones? Let’s see: around 550 students participate at the IMO every year, and around 2-3 students get perfect scores. Hence, if one assumes that at least 400 students solve at least one problem, we can guess that the harder IMO questions are at least 100 times harder than the easier ones. Assuming that AI becomes only 10 times better at solving math questions every 2 years, this is not good news for this prediction to come true by 2025.

Does this suggest that the IMO challenge should be solved by 2026, and not 2025?

Yes. If we continue our rate of progress, and no further algorithmic advances are made that rapidly speed up progress, we should be golden by 2026.

What should the chances of the challenge being answered by 2025 be?

Assuming that each additional year until 2026 adds 25% to this probability, maybe 75%?

What will cause me to update this probability?

Reading papers along the lines of “AI can now solve almost all questions on the IMO”, or “AI likely to require human supervision in the near future” would cause me to update my prediction.

What about the fact that the Metaculus dashboard is much more pessimistic about this question than me?

If I turn out to be wrong, I will update my preferences, and look at Metaculus predictions to set my expectations (base rate).