Related posts

Working on @ampcode. Writing registerspill.thorstenball.com. Once authored interpreterbook.com and compilerbook.com.

• Our readingkorrents.com

  Performance-critical human contributions will remain an edge case, and in 99% of software a faster algorithm or better data structure does not matter.

  Their words

  Performance critical contributions by humans will stay what they are: an edge case. In 99% of software it does not matter that you could’ve written a faster algorithm or picked a better data structure.

  thorstenball.com

Sequoia Professor of Computer Science at Stanford, co-director of the Stanford Institute for Human-Centered AI, and co-founder and CEO of World Labs.

• Our readingkorrents.com

  AI's bottleneck was never its algorithms but how little data they were fed, which is why the field needed ImageNet rather than another model.

  Their words

  So we conjectured that the lack of data was a huge part of the reason that's the lack of progress in AI. So we took a departure from everybody else who are really focusing only on algorithm and said that we need data. we need data to drive these algorithms.

  youtube.com

Co-founder of NVIDIA and, as of 2026, its chief executive; the company designs the GPUs most large AI models are trained on.

• Our readingkorrents.com

  The business is not building a great chip; it is accelerating an algorithm domain.

  Their words

  we realized early on uh that it's not about building a great chip, it's about accelerating an algorithm domain.

  Jensen Huang: The Mindset That Built NVIDIAyoutube.com

Head of Instagram since 2018, previously head of Facebook's News Feed.

• Our readingkorrents.com

  The recommendation algorithm has no idea that you like surfing; it has a number that happens to correlate with surfing.

  Their words

  It knows I like surfing and it's like it doesn't. It just has this big ass number that happens to correlate with surfing.

  youtube.com

AI researcher at OpenAI; built the poker systems Libratus and Pluribus and the Diplomacy agent Cicero, and works on reasoning models.

• Our readingkorrents.com

  A model can make an existing algorithm a hundred times faster and still cannot invent a better one, however long you give it.

  Their words

  And then I was like, okay, can you come up with an algorithm that is better than the algorithms that I came up with or that anybody else came up with and go ahead and like look at all the published work and synthesize that and then try to come up with something novel and it it's not able to do it. And I can give it a lot of time and it's it's still not able to do it.

  youtube.com

Makes NeetCode, the structured practice list and video walkthroughs people use to prepare for coding interviews; a former Google engineer.

• Our readingkorrents.com

  The right way to hire is not four rounds of data structures and algorithms but work like the job, plus asking a candidate why they made each choice.

  Their words

  I'm not going to spend four interviews going through and asking somebody data structures and algorithm stuff. I just have them do work that might be similar to something I'd give them on the job or even just have a conversation with them. See how they think. Like can they think through trade-offs? I don't even care about what answer they give me to a problem. I care about like what's like why did they say that?

  Tech interviews with NeetCodeyoutube.com

Co-founder of NVIDIA and, as of 2026, its chief executive; the company designs the GPUs most large AI models are trained on.

From one piece Jensen Huang – Will Nvidia’s moat persist? 2 beliefs, in the piece's order there

• Our readingkorrents.com

  Moore's law now yields about 25% a year, so 10x and 100x leaps can only come from changing the algorithm and how it is computed, every single year.

  Their words

  TPUs like anything else is impacted by Moore's law. And we know that Moore's law is increasing about 25% per year. And so the only way to really get 10x leaps, 100x leaps, is to fundamentally change the algorithm and how it's computed every single year.

  Jensen Huang – Will Nvidia’s moat persist?youtube.com

• Our readingkorrents.com

  Most advances in AI have come from algorithms rather than raw hardware, which makes China's army of AI researchers its fundamental advantage.

  China

  Their words

  We have got to acknowledge that most of the advanc advances in AI came out of algorithm advances not just the raw hardware. Now if most advances came from algorithms and computer science and programming tell me that their army of AI researchers is not their fundamental advantage.

  Jensen Huang – Will Nvidia’s moat persist?youtube.com

Reinforcement-learning researcher; co-wrote the field's standard textbook, wrote The Bitter Lesson, and shared the 2024 Turing Award.

• Our readingkorrents.com

  Gradient descent finds a solution to the problems a model has seen; nothing in the algorithm makes it pick the one that generalises well.

  Their words

  Well, there's nothing in them which will cause it to generalize. Well, the gradient descent will cause them to find a solution to the problems they've seen. And if there's only one way to solve them, you know, they they'll do it. But there are many ways to solve it. Some which generalize well, some which generalize poorly. There's nothing in them in the algorithms that will cause them to generalize well.

  youtube.com

Software engineer who writes the blog Made of Bugs about performance, debugging and understanding computer systems. Previously worked at Anthropic on interpretability, at Stripe on Sorbet, and at Ksplice.

• Our readingkorrents.com

  Using a neural network is likely to become easier than designing a custom-tuned prediction algorithm for many tasks.

  Their words

  in 2025 it might end up being easier to just throw a neural net at the problem, instead of a custom-designed-and-tuned prediction algorithm!

  The ITTAGE indirect branch predictorblog.nelhage.com

Cognitive scientist and long-standing critic of deep learning's claims; writes Marcus on AI and wrote Rebooting AI.

From one piece A knockout blow for LLMs? 2 beliefs, in the piece's order there

• Our readingkorrents.com

  A language model is no substitute for a well-specified conventional algorithm, so it cannot simply be dropped into a complex problem and trusted.

  AGILLMs

  Their words

  What the Apple paper shows, most fundamentally, regardless of how you define AGI, is that LLMs are no substitute for good well-specified conventional algorithms.

  A knockout blow for LLMs?garymarcus.substack.com

• Our readingkorrents.com

  There is no route to AI alignment or safety that goes around reliability: a system that cannot reliably follow a known algorithm cannot be made safe.

  AI alignment

  Their words

  And good luck getting to “alignment” or “safety” without reliabilty.

  A knockout blow for LLMs?garymarcus.substack.com

Founder and chief executive of Superhuman, the email client; previously founder of Rapportive, which he sold to LinkedIn; a professional game designer before that.

• Our readingkorrents.com

  Product-market fit can be measured, optimised and numerically increased, and an algorithm can write the roadmap that increases it.

  Their words

  which is number one, you can measure product market fit. Number two, you can optimize product market fit. Number three, you can systematically even numerically increase product market fit. And number four, you can even have an algorithm write your road map for you.

  youtube.com

Growth advisor and operator who has led growth at Miro, Amplitude, Dropbox and SurveyMonkey.

• Our readingkorrents.com

  Acquisition built on somebody else's algorithm can be taken away at any moment, which is why it must sit below earned channels on a growth roadmap.

  Their words

  An algorithm can give it, but algorithm can also take it away at any point and you have no control because you don't own those channels.

  youtube.com

Co-founder and CEO of Perplexity, an AI answer engine; previously a research scientist at OpenAI and a PhD student at UC Berkeley.

• Our readingkorrents.com

  Vector embeddings will not solve search: a decades-old term-frequency algorithm still beats most of them at ranking.

  Their words

  There is an algorithm called BM25 precisely for this, which is a more sophisticated version of TF-IDF. TF-IDF is term frequency times inverse document frequency, a very old-school information retrieval system that just works actually really well even today. And BM25 is a more sophisticated version of that, that is still beating most embeddings on ranking.

  youtube.com

Music critic and historian; author of The History of Jazz and Music: A Subversive History. Writes the newsletter The Honest Broker on music and culture.

• Our readingkorrents.com

  Anything that might take a reader off the platform, including a news story or an outside link, is punished by the algorithm.

  Their words

  Anything that might persuade you to leave the platform—a news story, or any outside link—is brutally punished by their algorithms.

  The State of the Culture, 2024honest-broker.com

Statistician and Chief Scientist at Posit (RStudio). Author of the ggplot2, dplyr and tidyverse R packages and of books on R programming and data science.

• Recommendsrcmnd.app

  The Algorithm Design Manual

  Their words

  To me, this book is an illustration of the power of names. Today, in the era of Google, if you know the name of something, you can find out about it with a simple search. But if you don't know of what you're looking for, it suddenly becomes much harder to find it. Having in the back of your head the names of common algorithms that help you solve problems is really powerful.

  fivebooks.com

Programmer and writer on computer architecture, performance, and software reliability. He has worked on CPU design at Centaur Technology and on software at Google and Microsoft, and writes long-form technical essays at danluu.com.

• Lovedaffiliate linkrcmnd.app

  Algorithms

  Their words

  Everything about this book seems perfect to me. It breaks up algorithms into classes (e.g., divide and conquer or greedy), and teaches you how to recognize what kind of algorithm should be used to solve a particular problem.

  danluu.com

• Mixed onaffiliate linkrcmnd.app

  Algorithm Design

  Their words

  Same comments as for CLRS -- it's widely recommended as an introductory book even though it doesn't make sense as an introductory book.

  danluu.com

Assistant Professor of Computer Science at the University of Toronto; co-creator, with Diederik Kingma, of the Adam optimizer for training neural networks.

• Our readingkorrents.com

  Stochastic optimization can be done well using only adaptive estimates of the gradient's lower-order moments.

  Their words

  We introduce Adam, an algorithm for first-order gradient-based optimization of stochastic objective functions, based on adaptive estimates of lower-order moments.

  arxiv.org

Research scientist at Anthropic; co-creator of the Variational Autoencoder and, with Jimmy Ba, of the Adam optimizer, two of the most widely used tools in deep learning.

• Our readingkorrents.com

  Stochastic optimization can be done well using only adaptive estimates of the gradient's lower-order moments.

  Their words

  We introduce Adam, an algorithm for first-order gradient-based optimization of stochastic objective functions, based on adaptive estimates of lower-order moments.

  arxiv.org

Programmer and essayist who has worked at Amazon, Google, Grab and Sourcegraph. Best known for the long-form essays collected as Stevey's Drunken Blog Rants and Stevey's Blog Rants.

• Mixed onrcmnd.app

  Compilers: Principles, Techniques, and Tools

  Their words

  and it's by far the hardest book in the list so far, even more so than The Algorithm Design Manual, in no small part because it's extremely dense.

  sites.google.com