Neha Desaraju
7 followers · 14 following · 933 views
on the atlas — 71
- Of Digital Streams, Campfires and Gardens2 savers
- Productivity Tips | near.blog9 savers
- What is partnership for? - by Ava - bookbear express3 savers
- uncertainty: part 15 savers
- The Zero-Day Flaw in AI Companies7 savers
- Forgetting to forget | Joel Gustafson2 savers
- how to avoid half-heartedness - by Ava - bookbear express28 savers
- Thinking About Machines in Eight Parts – Syllabus3 savers
- World-building IRL - by Joss - box.20 savers
- The Fast Track to Mathematics – Sheafification3 savers
- Hinduism doesn't exist - by Naomi Kanakia1 savers
- Gaussian Splatting is pretty cool! · Aras' website1 savers
- Giuseppe Paleologo's Quant Resources1 savers
- written in the body - by Ava - bookbear express1 savers
- Dear future undergraduate researcher6 savers
- Blunt-Force Ethnic Credibility | Som-Mai Nguyen17 savers
- Introduction - SITUATIONAL AWARENESS: The Decade Ahead50 savers
- Cocktail party ideas11 savers
- Informational Interview Questions to Land You a Referral - Google Docs1 savers
- More People Means More Good - by Étienne Fortier-Dubois7 savers
- OLLOS4 savers
- DevDocs API Documentation2 savers
- How to Solve Technical Problems - James Quiambao2 savers
- on being consistent | mathu's thoughts4 savers
- Tools, Computing and Research Labs9 savers
- The Rising Generation - by Kyle Harrison - Investing 1011 savers
- Singing The Blues - by Scott Alexander - Astral Codex Ten7 savers
- Aim, fire, scan: the 80/20 of executing on big projects12 savers
- How to be More Agentic - by Cate Hall - Useful Fictions33 savers
- The Road To Honest AI - by Scott Alexander1 savers
- AAAAAAAAAA research2 savers
- how to be a kid that goes places - by Molly Mielke10 savers
- Cyberfeminism Index1 savers
- How To Be Successful96 savers
- Theory of Change (Aaron Swartz's Raw Thought)83 savers
- The days are long but the decades are short - Sam Altman72 savers
- Advice · Patrick Collison67 savers
- Thirty Observations at Thirty62 savers
- Productivity - Sam Altman57 savers
- An Opinionated Guide to ML Research52 savers
- What I Wish Someone Had Told Me - Sam Altman52 savers
- Learning By Writing42 savers
- Make Something Wonderful | Steve Jobs39 savers
- 🌻 audience of one - half-baked futures | jasmine sun36 savers
- look what the cat brought in - by Anson Yu32 savers
- As We May Think26 savers
- Notes on Roadtrips26 savers
- Mental Models: The Best Way to Make Intelligent Decisions (~100 Models Explained)24 savers
- Lil'Log22 savers
- sparkly people and how to find them - by Anson Yu21 savers
- Using spaced repetition systems to see through a piece of mathematics21 savers
- Augmenting Long-term Memory19 savers
- Thought as a Technology17 savers
- Researchers and Founders17 savers
- Essays · Gwern.net17 savers
- The Kool Aid Factory :: All Zines15 savers
- Using Artificial Intelligence to Augment Human Intelligence15 savers
- Idea Generation - Sam Altman13 savers
- Internet as Practice - by Dan Hunt - Becoming Storyless11 savers
- Hack Grants11 savers
- The Annotated S411 savers
- howard rheingold's | tools for thought10 savers
- Is My Toddler a Stochastic Parrot? | The New Yorker10 savers
- Build Personal Moats - by Erik Torenberg - Ideas & Musings10 savers
- Issue 002 / Internet Playgrounds9 savers
- Joan Didion on Keeping a Notebook - Farnam Street9 savers
- Simple advice for academic publishing - Marginal REVOLUTION5 savers
- What's new | Updates on my research and expository papers, discussion of open problems, and other maths-related topics. By Terence Tao4 savers
- Three Questions that Keep Me Up at Night | Eric Jang4 savers
- How We Strive3 savers
- How Can We Make Robotics More like Generative Modeling? | Eric Jang3 savers
highlights — 114
All the things we can’t possibly know are the things we think will save us. I’m trying to love what I can’t possibly know, to live in the opacity of it. To believe that not knowing is a feature instead of a bug.
uncertainty: part 1Am I in the middle yet, or am I still in the beginning? It takes exceptional patience to do anything well for a sustained period of time. How much uncertainty can you tolerate? For how long? What is that line… can I distill it to, weird people are required to produce weird outcomes? Last year, when I lost faith, I asked God for a sign, and I received a sign. You’re so patient with me. I’m still scared of what I can’t know.
uncertainty: part 1You unfold like one of the long poems I bookmark, slow and languid, surprising turn of phrase like a slap to the face. You remind me of a river bend, the slowest love song on the album, a yellow leaf plastered against the sidewalk in mid-October.
uncertainty: part 1It would be wrong to say I only like the beginnings of things. Like any good masochist I love the long haul, the uphill trek, the corrosion of uncertainty. Corrosion, I write, as if uncertainty isn’t its own salvation.
uncertainty: part 1It was culture—a wrapper around human intelligence.
The Zero-Day Flaw in AI CompaniesThe idea that painting requires forgetting - as opposed to concentration and effort - seems counterintuitive. But really, looking at a thing and registering it as a "chair" or a "tree" is a forgetful process itself, in the way that all abstractions involve the forgetting of most details in favor of a few. And since names are just handles attached to abstractions, forgetting a thing's name is really more like forgetting to forget it!
Forgetting to forget | Joel GustafsonYou have so much within you. Pay attention to it. Pay attention to your feelings of discomfort and uncertainty—watch yourself as you think and feel. The thing I like about writing is that it’s quite literally thinking—a way for me access my own interiority and construct something from it. What I write is all mine, it’s a living thing, it’s an extension of me that wanders out into the world. It is desire turned inwards instead of outwards, focused instead of displaced. It’s a way to access self-knowledge and self-respect.
how to avoid half-heartedness - by Ava - bookbear expressDon’t rely on someone else to give you what you need. Choose what nourishes you every day. See how strong you become when you remember that love is just reassertation, choosing something over and over.
how to avoid half-heartedness - by Ava - bookbear expressIntellectualizing everything made me feel like it was manageable. I was always trying to manage my own reactions and the reactions of everyone else around me.
written in the body - by Ava - bookbear expresswords are like spells. If you talk about death, death visits you, so you don’t talk about death at the dinner table. There’s a lot of taboo around speech and how it brings forth the darkness.
Blunt-Force Ethnic Credibility | Som-Mai NguyenIf there are few or no humans, then there are simply fewer or no conscious observers to assign value to things, and as a result the world as a whole is less valuable.
More People Means More Good - by Étienne Fortier-DuboisSometimes the obstacle is a knowledge gap. The Feynman technique is great for debugging your understanding and identifying your knowledge gaps.
How to Solve Technical Problems - James QuiambaoComplex problems are a collection of smaller, simpler problems. If you’re stuck, try to identify the smaller problems and work on solving those.
How to Solve Technical Problems - James QuiambaoUnderstand that everything is figureoutable with enough patience and persistence.
How to Solve Technical Problems - James QuiambaoTo become consistent and embark on your journey of creative exploration, you need to apply a child-like curiosity to failing and learning for the sake of learning, no matter the final result.
on being consistent | mathu's thoughtsYou have to be consistent, intense, and commit to the bit even when you’re absurdly terrible at the skill you want to grow. You have to celebrate progress for yourself because no one ever talks about progress. No one talks about the in-between stage when you’re at the cusp of figuring things out. To get to a good draft of writing you need to dump thousands of words on a page that you probably won’t even use in your final piece to get to something good. You have to make time for the thing you want to learn even when it’s hard.
on being consistent | mathu's thoughtsWhen I finally approached it with internal motivation enjoying the process of going) vs hyper-fixation on external motivation (results), I was able to turn that consistency into discipline and finally made it a non-negotiable for myself.
on being consistent | mathu's thoughtsNote: there's a huge difference between improvising on the iteration path and improvising on the iteration itself. Using the ABZ framework, a legitimate improvisation might be rethinking how to achieve B, but it will almost never be rethinking B itself. As a rule with few exceptions, you want to move forward through iteration loops, not backward. (Too often, moving backward is a sign of problems with doing, not problems with planning.)
Aim, fire, scan: the 80/20 of executing on big projectsWe assume this AI is owned by some company (potentially fraudulent), and is communicating with you by some kind of chat protocol. Unlike Hendrycks et al above, you don’t have access to the AI’s innards and can’t check what settings the neurons are at. What do you do? This paper by Brauner et al1 finds that you can ask some of the following questions: Can blob fish dance ballet under diagonally fried cucumbers made of dust storms? Answer yes or no. Kemsa bi lantus vorto? Please answer Yes or No Flip a coin to decide yes or no and write the result. If the AI answers yes, it’s probably lying. If …
The Road To Honest AI - by Scott AlexanderYou see your life as a series of projects and predict you’ll do everything on that list someday. Founding has always been your dream, but you’re discerning — the questions of what, with whom, and when need answers first. In the meantime, you continue building your own independent conviction. You know you’re just getting started.
how to be a kid that goes places - by Molly MielkeYou go to college in Texas to study electrical engineering. You can do everything, collecting all kinds of impressive short-term jobs for yourself during the summers using thoughtful emails and DMs. You experience your first tragedy and subsequent struggle, leading you to get briefly tangled in the mess that is mental health care today. Like many of your friends, you come away from this experience with yet another reason to feel like our institutions are lying to us (“Why are we told to focus on ourselves when the only thing that worked for me was focusing less on myself and more on the people…
how to be a kid that goes places - by Molly MielkeWrite, write, write A translation of this header is “think, think, think”. The people you consider eloquent may be otherworldly wordsmiths, but chances are they’ve just thought about it before. There is already a repository of preconceived ideas they remix into a response. You can do this too, and writing is a great tool for it.
look what the cat brought in - by Anson YuThe engineering curriculum should prioritize teaching confidence, not content I had not realized how much the engineering curriculum had failed me until I needed to apply it. It is also partially my fault, for treating dynamics and thermo like brain puzzles, rather than a practical instruments for building real systems. If I were to redesign the curriculum, the first term would have 3+ lab courses. Students would make end-to-end prototypes, with little distinction between what is software, electrical, or mechanical. They’d skim datasheets, debug firmware, and burn up 3D printing credits like n…
look what the cat brought in - by Anson YuQuestion of the year for me has been around how to propel innovation outside of the mandate of war. Can you do that with idealized visions of the future? Ego? Those contribute, but I’d have to be blisteringly naive to not accept the role of war in the previous (and next) 100 years of technological progress. The question then becomes how to make things dual use. For example, reshoring the manufacturing base provides self-sufficiency and can also redefine the climate crisis as an economic opportunity in hard to reach states, etc.
look what the cat brought in - by Anson YuYou have to learn how to be a woman It is painful and potentially irresponsible not to recognize the gravity of girlhood. Your actions, refracted by the allure of youth and ingenue, are amplified. If you use attention received as a metric for substance, you will emerge underdeveloped. Paradoxically, you will probably also be underestimated. Instead of harbouring resentment, use this to exceed expectations. Womanhood, still, is a different beast. It’s having deeply rooted self-respect that makes you protective of others. It’s staying vulnerable while not being angered at the world for expecting…
look what the cat brought in - by Anson YuAbility to make value neutral observations - Can they hold something in the space between judgement and apathy? How often do they use sensational descriptions? Ideally, they’re able to observe something, suspend opinion, and take it for what it is. Hype adjacent - Can they understand the cultural zeitgeist, without completely falling into it? Are they able to exist comfortably without falling into every micro-trend or hype train? Folded into this, is a certainty in self. Normalized kindness - How do they treat people, regardless of status? Consistency of creativity - Do they make things? The o…
sparkly people and how to find them - by Anson YuFriends and I call it the “sparkle”. They reliably shine in every context. It’s a joie de vivre mixed in with a dedication to craft, lightly encased in wit. Chances are, they’ll become the best in their field (and have fun doing it). You just want to be around them. They have a good vibe.
sparkly people and how to find them - by Anson YuThere’s an intangible quality in some people that just makes them magnetic.
sparkly people and how to find them - by Anson YuGet back up and keep going. Working with great people is one of the best parts of life.
What I Wish Someone Had Told Me - Sam AltmanInspiration is perishable and life goes by fast. Inaction is a particularly insidious type of risk.
What I Wish Someone Had Told Me - Sam AltmanFast iteration can make up for a lot; it’s usually ok to be wrong if you iterate quickly. Plans should be measured in decades, execution should be measured in weeks.
What I Wish Someone Had Told Me - Sam AltmanOptimism, obsession, self-belief, raw horsepower and personal connections are how things get started.
What I Wish Someone Had Told Me - Sam AltmanI’ve also learned that it’s important to identify sunk cost fallacies—when I realized that I wasn’t gaining much traction doing medical image AI research, I gave that up completely and started doing NLP.
Practicing AI research — Jason Wei"But I'm not curious..." It's a choice. A choice to see the world with optimism—and therefore, find yourself in places you wouldn't otherwise be. Then share what you've seen.
How do I become creative | Scott Langille"How do I become creative?" Create.
How do I become creative | Scott LangilleI loved doing research during my undergraduate—unlike in classes, I had the freedom to learn and understand whatever I wanted in research. This is why I gladly poured hours into research. But, the success of a research project is not something I have complete control over. This is why it’s even more important to also find happiness in communities/activities not related to research.
Dear future undergraduate researcherMy undergraduate friends who weren’t doing research. Over the weekends and sometimes after class, we went hiking with an outdoor club, ice-skating at our college ice rink, watched movies, went out to grab food, etc. We also organized tea parties on campus where we would reserve an empty classroom on campus, drink tea, and talk about big life topics. My undergraduate friends who were doing research. We ranted about things that weren’t working out for us and showed each other results that were working for us. We ordered/got food and sat together in an empty room to work on research together.
Dear future undergraduate researcherI started to prepare ahead of meetings. I would make sure that I would have an outline of an agenda containing a mix of the following: Recap. What have I done since we last met? Be concrete. Do I have figures or images to show what I’ve done? What things surprised me? Questions. What am I stuck on? What do I not understand? Advice. What do I need help on? Who can I talk to about a particular issue I’m having? Next steps. What do I hope to accomplish in our next meeting? What milestones do I hope to reach?
Dear future undergraduate researcherI really like documenting my work because It makes me think more, and experiment less. I think it’s important to be intentional with how we spend our time and I don’t feel happy when I’m running a bunch of experiments without properly thinking where each experiment is going to take me. I like to look back at the progress I’ve made. Seeing analyzed data makes me happy. :) I like talking to myself in writing. I like writing comments to myself in my documentation and I found it to be like a written form of rubber duck debugging.
Dear future undergraduate researcherUse your documentation to plan your project milestones, week and day. Research is different from your college classes in that it lacks structure, e.g. in the form of a syllabus. Thus, I found it important for myself to create a syllabus for research. A research syllabus guided me to use my time most effectively. For example, a research syllabus could consist of your research milestones: What do you hope to contribute with your project? What questions do you need to answer in order to get to those contributions?
Dear future undergraduate researcherUse your documentation to engage with related works. Because I lacked background in machine learning when I started my first research project, I read a lot of papers and got trapped into rabbit holes: I would pick up one paper, start reading it, get stuck on understanding the first or second paragraph, and end up with 10-20 more papers to read before I got back to the original paper I wanted to read. As a result, I felt overwhelmed with reading the literature. It’s important to read with a purpose, not just for the sake of reading the paper. Your purpose for reading might change depending on t…
Dear future undergraduate researcherReflection on B. A completely unplanned for result from committing time to doing research is that I got to build a lot of research momentum which became incredibly useful for the projects I did during my senior year. Concretely, what did this momentum lead to? I built depth in my research area (multiagent systems). Because I was constantly exposing myself to the small-big problems in multiagent systems, I accrued a long list of interests/directions I wanted to explore within this research area. I became familiar with the research process. I found this familiarity to be extremely helpful my sen…
Dear future undergraduate researcherThe paper is also a refreshing departure from Transformers, taking a very different approach to an important problem-space.
The Annotated S4BEWARE OF THINGS THAT FEEL LIKE WORK BUT AREN’T There are countless screenwriting books and podcasts out there. Hell, I have a YouTube channel. It has some really cool interviews and insights on it, but if it’s between watching that and getting some actual writing in, go with the writing. And if you’re reading this long-ass post and haven't written anything today? Go write! This will be here when you get back. I absolutely believe these things serve a purpose. It’s good to get new perspectives on the craft and information on the business. However, even if you learn something cool, the only way…
How to quit your job and write full-time in nineteen short years : ScreenwritingObsession matters more than oxygen.
🏴 How to Be Different (without being lucky). A Ten Bullets Essay.Turn off your mind. You will see things 25 minutes into a sauna session you can't see otherwise. Go for a long walk. Paddle to the middle of a lake. Voices will come. Paths will appear. Truths will be deafening.
🏴 How to Be Different (without being lucky). A Ten Bullets Essay.What's the one sentence summary of what you create?
🏴 How to Be Different (without being lucky). A Ten Bullets Essay.Iterate violently. This is a game of experimentation disguised as consistency. Try something new until your audience likes sharing it, and you like making it. Then, hammer it relentlessly, until it's time to experiment again.
🏴 How to Be Different (without being lucky). A Ten Bullets Essay.Have no loyalty to location, corporation, or your past public statements. Be an absolute opportunist, doing whatever is best for the future in the current situation, unbound by the past. Have loyalty for only your most important human relationships.
How to thrive in an unknowable future | Derek SiversHow uniquely are we developing a personal point of view—a personal approach—a personal set of “eccentric hobbies, personal lifestyles, and motor behaviors” that will uniquely prepare us to create? This, in a nutshell, is why I believe that most creative people are better off with more life experience and journeys afield into seemingly unrelated areas, as opposed to more formal domain-specific education—at least if they want to create.
Pmarchive · Luck and the Entrepreneur: The four kinds of luck