jeromebaek

Pi is computable. What is computable, can be represented in finite information (the algorithm used to compute the number). Almost all real numbers are not computable.

Now tie this with computability theory. Almost all real numbers are uncomputable. It's obvious that they don't "exist" in any sense if by "exist" we mean "computable". Now consider that entropy, is information, is the measure of randomness, or uncomputability, is time. What is uncomputable, coming into existence, is the passage of time.

So Airbnb has tried to unsuccessfully pivot to a payments infra company for several years now. Notable acquisitions for this effort include acqui-hiring ChangeTip, a bitcoin micropayments platform, and Tilt, a social payments platform. But Square and CashApp are way ahead of the curve. I suspect there's internal political problems.

Sure. But most senior devs are not Tim Patterson.

Explains why there's little outbreak in Africa.

Philosophy of Computation at Berkeley: http://po.cab

relevant: falsehoods programmers believe about time https://news.ycombinator.com/item?id=4128208

As a LGBTQ person this article is so creepy

Y'all are gonna call me a crackpot, but consider P, PSPACE and IP. PSPACE is (probably) much more powerful than P. Also PSPACE=IP (interactive proofs). Learn interactively. Treat your compiler as the prover, you as the verifier. Don't try to understand everything by simply reading the manual -- that's only utilizing P, not PSPACE.

this is hilarious. sorry for laughing

gosh, this is unsettling. my brain is literally getting stuck on an infinite loop trying to read this and coherently put them together

No. peak market share of Windows XP was 80%. market share of chrome is hovering around 65%. search engine share of google is like 90%.

please stop spreading the myth that ChromeOS is somehow immune to malware. nothing stops a dev from uploading malware to the Chrome web store, and there are many such cases that happened. the only meaningful metric of how safe a piece of software is, is the size and investment in the team responsible for the security of the product. And MS is leading in this aspect.

the best way to ramp up on a codebase is to simply follow the critical path and draw an execution diagram of the entire critical path. tools like these are gimmicks.

ML is hierarchical. the upstream layers categorise more abstractly and downstream layers categorise more specifically.

this reeks of internal political problems. this isn't good for google nor for google's partners. the only reason they would autocannibalize is if there are too many engineers sitting around doing almost nothing so they start inventing problems to solve.

the economy of literature, marc shell. it gives you a semiotics of money, a way of understanding money qualitatively. it has been 100x more valuable than any economics textbook.

Think about it. Soft AI that assists the driver (automatic turning, cruise control, automatic brakes) is now mainstream. Why is it that full self-driving cars are categorically different? Because it requires full faith in the machine. When assistive technology goes wrong, the driver can correct it and -- get this -- the reason the driver can correct it fast enough is precisely because the driver doesn't fully trust the technology. the driver is on alert, always, because they know the technology is not meant to be trusted fully. now a full self-driving car asks the driver to trust it fully. so the "driver" can take a nap, read a book, whatever. if the driver needs to be on alert, it's by definition not a fully self-driving car. and i dont think we will get there, ever.

it's entirely possible that the Berkeley CS professors, who are obviously geniuses but also kind of dolts, knew nothing about these bureaucratic intricacies and took the most obvious way out.

i was an undergraduate 8-hour TA at UC Berkeley for six semesters between 2015 and 2018. i was also part of the union. i'm surprised few of the comments are mentioning the main issue: union contract stipulates that 10-hour or more TAs receive tuition fee remission, but even 9.9-hour TAs receive zero remission. (there was at least one 9.9 hour TA, though not in the EECS department.) in 2015 when the 8-hour TA position was implemented there were initially not very many of them, and there were more 20-hour TAs, mostly graduate students. but between then and now Berkeley's CS program grew exponentially. so more and more 8-hour TAs were needed. what started as a clever hack was beginning harm other departments' workers by setting a precedent (ex, the 9.9 hour TA in chemistry). so the union fought and won.

Berkeley's undergraduate CS education is a unique self-sustaining juggernaut. there are not only TAs officially hired and paid by the department, but also numerous volunteers tutoring, lab assisting and otherwise helping each other out. it would be a tragedy if the bureaucracy destroys this beautiful self-sustaining machine through this decision. but i trust Berkeley will find a way as always to teach the most with the least resources.

it makes me sad this sort of unnuanced, oh-so-rational take unengaged by any sort of philosophy is still getting upvotes in HN. posting here feels like a lost cause.

Wow! I’m impressed. He sounds like an adult now. He actually has nuanced opinions. What a world we live in.

nobody likes "artificial intelligence". laypeople are scared of it. the media blames it for all evils. it brings connotations of frankenstein, arrogant atheism, etc. it's reasonable to want to hide behind a different term.

This is all true, but it’s a far cry from OP’s statement

Scaruffi is an interesting guy. I've been following him since 2014. One of the most prolific polymaths of our time. This is a good book: https://www.scaruffi.com/nature/purchase.html

Azure doesn't run O365, Exchange, Bing, or Xbox Live. Please stop spreading misinformation. (I work at MS.)

The AI/ML hype drives away talented students from engineering and systems research, as well. Because of the hype engineering somehow is less prestigious than “data science”. Wonder how long until this trend will blow up.

“This software may not be sold or published in printed form without written permission from the copyright holders.“ This is almost certainly because parts of the source code appear in the textbook, and publishing laws apply.

Great post! Informative and interesting

A little-discussed issue is that big tech cos tend to do better on diversity and inclusion. Obviously it varies by startup, but when you’re in a survival mentality it is difficult to be open to people of different sexualities, races and experiences. You tend to get clumps of similar overachievers. Paul Graham said this was an “advantage” but for people like me in the LGBTQ community it certainly isn’t.