I asked him more than 10 years ago if he would be interested in a formalisation of the proof, and he politely declined. I guess he was right to decline, my proposal would not have been viable then anyway.
If you've found web development frustrating over the past 5-10 years, here's something that worked great for me: give yourself permission to avoid any form of frontend build system (so no npm/React/TypeScript/JSX/Babel/etc) and code in HTML and JavaScript like it's 2009.
The joy came flooding back to me! It turns out browser APIs are really good now.
You don't even need jQuery to paper over the gaps any more - use document.querySelectorAll() and fetch() directly and see how much value you can build with a few dozen lines of code.
This is part of why I designed Tarsnap to keep data as secure as possible, even from me. If someone stores their crypto keys -- or world domination^W optimization plans -- on Tarsnap, I don't want to get kidnapped and tortured by anyone trying to steal that data.
1. Learn basic NNs at a simple level, build from scratch (no frameworks) a feed forward neural network with back propagation to train against MNIST or something as simple. Understand every part of it. Just use your favorite programming language.
2. Learn (without having to implement with the code, or to understand the finer parts of the implementations) how the NN architectures work and why they work. What is an encoder-decoder? Why the first part produces an embedding? How a transformer works? What are the logits in the output of an LLM, and how sampling works? Why is attention of quadratic? What is Reinforcement Learning, Resnets, how do they work? Basically: you need a solid qualitative understanding of all that.
3. Learn the higher level layer, both from the POV of the open source models, so how to interface to llama.cpp / ollama / ..., how to set the context window, what is quantization and how it will affect performances/quality of output, and also, how to use popular provider APIs like DeepSeek, OpenAI, Anthropic, ... and what model is good for what.
4. Learn prompt engineering techniques that influence the qualtily of the output when using LLMs programmatically (as a bag of algorithms). This takes patience and practice.
5. Learn how to use AI effectively for coding. This is absolutely non-trivial, and a lot of good programmers are terrible LLMs users (and end believing LLMs are not useful for coding).
6. Don't get trapped into the idea that the news of the day (RAG, MCP, ...) is what you should spend all your energy. This is just some useful technology surrounded by a lot of hype of all the people that want to get rich with AI and understand they can't compete with the LLMs themselves. So they pump the part that can be kinda "productized". Never forget that the product is the neural network itself, for the most part.
There are places I've found the topological perspective useful, but after a decade of grappling with trying to understand what goes on inside neural networks, I just haven't gotten that much traction out of it.
I've had a lot more success with:
* The linear representation hypothesis - The idea that "concepts" (features) correspond to directions in neural networks.
* The idea of circuits - networks of such connected concepts.
I had a phone call with him in about 2000, because he was then publishing a lot of material about DRM (and attacks on DRM), and I was also into anti-DRM stuff and was thinking of going to an industry meeting related to it. I wanted to know if he would publish whatever documents I might obtain there.
I remember that he said I could make a business card (!) saying that I was a special representative or special agent or journalist or whatever I wanted for Cryptome.
I said something like "wait, really?" and he said something like "well, who I am to say who does or doesn't work for Cryptome?" or "why should anyone believe you when you say you do or don't work for Cryptome? people should never believe each other!" or something like that.
He also warned me to watch out for people messing with my laptop in the hotel.
I didn't end up making the business card (I thought it would make people more suspicious of me rather than less, which was probably right), but I think I did send him a couple of documents, in retrospect probably very boring ones.
I met him briefly in person once, ironically at the announcement lecture for Wikileaks at HOPE in New York. I remember being confused because I assumed he would get along well with the Wikileaks people, but he was already kind of skeptical or cynical somehow.
He was also famous for posting extremely cynical takes to mailing lists.
John seemingly felt that power had already corrupted everyone or was always on the verge of corrupting everyone, and that one should be extremely reluctant to believe in anyone's stated motives for anything. I don't know if he thought there was some way out of that scenario or that that was just human nature. He always reminded me of the epigraph of Illuminatus!, attributed to Ishmael Reed: "The history of the world is the history of the warfare between secret societies."
I definitely admired his courage and independence.
Founder Home Assistant here. Want to chime in that I always love to see write ups like these to see the great things what people achieve with Home Assistant.
Not everyone might know, but last year we started the Open Home Foundation as a non-profit in Switzerland and I donated Home Assistant to it[1]. It's fully funded by users. There are no investors involved.
We are fully committed to building out a smart home that focuses on local control and privacy. Yes there are rough edges, but we're actively working on it in the open, with progress being released every month.
Congrats on the launch! I'm one of the authors of that paper you cited, glad it was useful and inspiring to building this :) Let me know if we can support in any way!
On October 25, 1988, I gave Steve Jobs a demo of pie menus, NeWS, UniPress Emacs and HyperTIES at the Educom conference in Washington DC. His reaction was to jump up and down, point at the screen, and yell “That sucks! That sucks! Wow, that’s neat! That sucks!”
I tried explaining how we’d performed an experiment proving pie menus were faster than linear menus, but he insisted the liner menus in NeXT Step were the best possible menus ever.
When I explained to him how flexible NeWS was, he told me "I don't need flexibility -- I got my window system right the first time!"
But who was I to rain on his parade, two weeks after the first release of NeXT Step 0.8? He just wasn't in the mood to be told that he could have a better user interface.
So I gave him one of the a "NeRD" buttons I'd made for NeWS NeRDs, which he appreciated.
Up to that time, NeXT was the most hyped piece of vaporware ever, and doubters were wearing t-shirts saying “NeVR Step”!
Even after he went back to Apple, Steve Jobs never took a bite of Apple Pie Menus, the forbidden fruit. There’s no accounting for taste!
I interned at zed during the summer of 2022, when the editor was pre-alpha. Nathan, Max, Antonio are great guys and build software with care. I'm happy to see the editor receive the success it deserves, because the team has poured so much world-class engineering work into it.
I worked with Antonio on prototyping the extensions system[0]. In other words, Antonio got to stress test the pair programming collaboration tech while I ran around in a little corner of the zed codebase and asked a billion questions. While working on zed, Antonio taught me how to talk about code and make changes purposefully. I learned that the best solution is the one that shows the reader how it was derived. It was a great summer, as far as summers go!
I'm glad the editor is open source and that people are willing to pay for well-engineered AI integrations; I think originally, before AI had taken off, the business model for zed was something along the lines of a per-seat model for teams that used collaborative features. I still use zed daily and I hope the team can keep working on it for a long time.
[0]: Extensions were originally written in Lua, which didn't have the properties we wanted, so we moved to Wasm, which is fast + sandboxed + cross-language. After I left, it looks like Max and Marshall picked up the work and moved from the original serde+bincode ABI to Wasm interface types, which makes me happy: https://zed.dev/blog/zed-decoded-extensions. I have a blog post draft about the early history of Zed and how extensions with direct access to GPUI and CRDTs could turn Zed from a collaborative code editor into a full-blown collaborative application platform. The post needs a lot of work (and I should probably reach out to the team) before I publish it. And I have finals next week. Sigh. Some day!
This hits home for me because my great grandfather was given away at 11 years old, and his brother who was 12 also, because the family could no longer afford to keep them. They were given to a family with a farm in a nearby county. Unfortunately due to illiteracy and lack of official records, no one knows to this day who my grandfathers actual parents were. All I know is that he and his brother were dropped off with a new family and worked on the farm until they left for the military at 18.
This kind of article reminds me of just how good we have it in America today, despite all its faults.
"Good Vibrations" was actually recorded using an Electro-Theremin [1] (emphasis mine). It was essentially the same but sported more traditional knob controls. Also if you ever hear a Theremin-esque noise in an Elmer Bernstein soundtrack like "Heavy Metal" it was actually an Ondes Martenot [2] which is distinct from and less similar to the classic Theremin.
I had the opportunity to work the Foy Desk a few times during my undergrad at Auburn in the early 2000s - mostly as a volunteer while the regular workers would be in meetings. At the time we had a multi-page list of common questions and answers, the Internet (as it was then), as well as access to university computer systems for things like class schedule lookups.
The most common questions I got then were from other students, most around when a certain class started or where it was located. This is was the early 2000s and, while a lot of this was available via OASIS (the Auburn student system) for any student, many either didn't have the computer savvy to use it or ... didn't have a computer at home at all!
The most unusual call I took was from a student who was lost in Haley Center (the largest building on Auburn's campus - at the time, not sure about now as I haven't been back in decades - and somewhat difficult to navigate if you aren't familiar with its layout). The poor kid sounded absolutely panicked. I actually had to pull up a map and walk the him turn-by-turn until he found the main hallway again.
As an aside, it's neat to see a few other Auburn alums on here. WDE!
I worked on the design of wireless USB chips around 2008 - 2010. They worked - you really could get USB 2.0 full rate connections wirelessly and we had some neat demos.
I would say the major problem it had with adoption was that wired USB also provided power. (A lot more people use usb to charge their phone than to sync their phone.)
So great - wireless connectivity... but you still have to plug the device into a cable at some point (or have replaceable batteries), which makes the value proposition a lot less clear.
Beyond that it suffered from the usual adoption chicken-and-egg problem. Laptop manufacturers didn't want to add it because it was an expense that didn't drive sales since there weren't any must-have peripherals that used it, and peripheral manufacturers didn't want to make wireless usb devices since they couldn't be used with a standard laptop (at least not without a WUSB dongle - which raised the cost).
your comment reminds me of when i interned at a large cloud services provider
i was very much infatuated by the idea of total reliability on a global scale and couldn't wait to talk about the CAP theorem with anyone who would listen. Read much of the research. I still love to flip through Designing Data-Intensive Applications by Martin Kleppmann occasionally.
As I was learning about said company's server racks there was a component that I asked an engineer on my team about – "Isn't that a single point of failure?"
He looked at me, exhausted and stressed out, and just said "Take it far enough, the Earth is a single point of failure."
This sounds big enough to require a black start. Unfortunately, those are slow and difficult.
If an entire nation trips offline then every generator station disconnects itself from the grid and the grid itself snaps apart into islands. To bring it back you have to disconnect consumer loads and then re-energize a small set of plants that have dedicated black start capability. Thermal plants require energy to start up and renewables require external sources of inertia for frequency stabilization, so this usually requires turning on a small diesel generator that creates enough power to bootstrap a bigger generator and so on up until there's enough electricity to start the plant itself. With that back online the power from it can be used to re-energize other plants that lack black start capability in a chain until you have a series of isolated islands. Those islands then have to be synchronized and reconnected, whilst simultaneously bringing load online in large blocks.
The whole thing is planned for, but you can't really rehearse for it. During a black start the grid is highly unstable. If something goes wrong then it can trip out again during the restart, sending you back to the beginning. It's especially likely if the original blackout caused undetected equipment damage, or if it was caused by such damage.
In the UK contingency planning assumes a black start could take up to 72 hours, although if things go well it would be faster. It's one reason it's a good idea to always have some cash at home.
I worked for a while on extremeophile Archaeal viruses - the type that infect organisms that manage to live in volcanic hot springs, for instance. These are ecological niches that are old, and extremely divergent. There's little genetic exchange between life around the hot springs, and life within them.
The typical route of discovering those viruses was first genetic. When you get a genome (especially back when this work was initiated), you'd BLAST all the gene sequences against all known organisms to look for homologs. That's how you'd annotate what the gene does. Much more often than not, you'd get back zero results - these genes had absolutely no sequence similarity to anything else known.
My PI would go through and clone every gene of the virus into bacteria to express the protein. If the protein was soluble, we'd crystallize it. And basically every time, once the structure was solved, if you did a 3D search (using Dali Server or PDBe Fold), there would be a number of near identical hits.
In other words, these genes had diverged entirely at the sequence level, but without changing anything at the structural (and thus functional) level.
Presumably, if AlphaFold is finding the relationship, there's some information preserved at the sequence level - but that could potentially be indirect, such as co-evolution. Either way, it's finding things no human-guided algorithm has been able to find.
Oh, fun! These came up on HN a few years ago[0]. Recycling some of comment from then:
The "Tower Sound and Communications" (TSC) company that recorded many of these was located a few miles up the road from my home town. The booming male voice on the recordings also sounded familiar to me, too. I'm fairly certain I heard him on the local radio station that my father played over the PA system in our family grocery store when I was growing up.
Turns out that Cecil "Lee" Rutherford, the voice on the recordings, did VO work for local radio stations near my home town, too. He died in November, 2020.
He was involved in some ventures that persist today. His company EchoSat[1] (which I'd heard of because I had some involvement in the convenience store / retail petroleum industry) merged with an IT security firm to become "ControlScan", doing PCI testing stuff because gas stations and credit cards.
Quoting the obituary[2]:
He started Tower Sound and Communications while in Greenville to pursue a venture that would eventually spearhead "in store" broadcasting for companies such as Kmart (he became the voice of Kmart) and Jamesway which evolved into another corporation in KY called EchoSat that would use satellite technology in helping with multiple stores for POS processing and security.
There's an interview with Lee Rutherford in 2011.[3] He absolutely still has that "radio voice".
These videos of robotic cow milking machines, feed mixers and distributers and pushers, and manure roombas are amazing!
Cows like to push and play with their food to get to the yummy grain bits, so the feed robot pushes the food back so they can eat it all.
And the Poopoombas had to learn to be more aggressive about pushing cows out of the way and not stopping every time they bumped or got kicked, because otherwise the cows would assign them the lowest status in the pecking order, and they could only cower in the corner.
Here are the videos from the article and some more:
The milking process of the Lely Astronaut A5 - EN:
For some reason the value of Pi given in the C++ code is wrong!
It's given in the source as 3.14159274101257324219
when the right value to the same number of digits is
3.14159265358979323846. Very weird. I noticed when I went to look at the C++ to see how this algorithm was actually implemented.
I lied about my age and was given janitor access in the mid 2000s. There was a special /j/ board to coordinate on, but it broke relatively early, and you mostly had to hang out in the #janiteam channel on Rizon. I think almost everybody else was underage as well. There was a minimal web overlay that let you delete/escalate posts. You couldn't see people's IPs, but you could see how many outstanding ban requests they had. These numbers helped me deduce that many boards' most infamous personalities were all the same guy.
We were all offered the chance to become mods in 2010, but moot wanted to see our faces on a Skype call. I thought that was a step too far and just gradually stopped caring after that. Seems like I made the right choice.
On the whole it was barely held together technically and organisationally, mostly run by moot's personal friends, and fun all around. Things were far less serious then.
I posted Bracket City to HN on February 24th and the game went live yesterday on The Atlantic (!)
The game will stay free to play (and not require logging in). Also, I'm still making all the puzzles!
HN provided the first real infusion of players that weren't my mom's friends. So thanks everyone.
FWIW The Atlantic's team is amazing and got this live exactly 2 weeks from when we signed the deal.
This happened quick and I feel very lucky. The HN community of solvers keeps me honest with much helpful technical and editorial feedback. I love it all -- here or at mayor@bracket.city
Finishing up my PhD thesis on low-resource audio classification for ecoacoustics. Our partners deployed 98 recorders in remote Arctic/sub-Arctic regions, collecting a massive (~19.5 years) dataset to monitor wildlife and human noise.
Labeled data is the bottleneck, so my work focuses on getting good results with less data. Key parts:
- Created EDANSA [1], the first public dataset of its kind from these areas, using a improved active learning method (ensemble disagreement) to efficiently find rare sounds.
- Explored other low-resource ML: transfer learning, data valuation (using Shapley values), cross-modal learning (using satellite weather data to train audio models), and testing the reasoning abilities of MLLMs on audio (spoiler: they struggle!).
For years, I've been good heartedly losing the blog SEO ranking fight to a great developer and writer who has the same name as me. A football player eclipses us both if you just google our shared name, but if you add any sort of "developer" or "programming", he's clearly got me beat for the top marks. It makes sense — he writes about tech much more consistently than I do, and his articles are likely much more helpful than my sporadic and eclectic posts.
Naturally, being vain, when I saw this post, I immediately looked up my own blog and was chuffed to see it at #292.
When I was fiddling with LED circuits back in the 70s (!) I experimented with turning the LED on and off with a square wave. If you turned it on and off rapidly enough, your eye did not notice it, and your eye perceived it as fully bright. (Adjusting both the frequency and duration of the "on" part.)
Hence, you could get some decent power savings doing this.
I wonder if this is commonly known. I've mentioned it to a couple EEs over the years, and they were able to reduce the power consumption of their devices.
Fun fact: 2^133477987019 is the smallest power of two that ends with 40 even digits. In fact, it ends with 46 even digits - which is surprising, given that it is significantly smaller than 2^(2^46). The last 50 digits of this number are ...32070644226208822284248862288402404246620406284288. This number has over 40 billion digits, though, so it seems kind of unlikely that we will ever find another number where all the digits are even. The relevant OEIS sequence is here: https://oeis.org/A096549
Context: I wrote a search program that is substantially faster - it takes just a few minutes to get up to 2^(10^13), although my laptop's limited memory is starting to be a problem (my intermediate result file is already nearly 1GB in size). Unfortunately, it seems there are no results up to 2^15258789062500, which is a 4.5-trillion digit number.
As a theater director and actor I spend a lot of time thinking about "charisma". What is it that makes some people interesting to look at, even when they're not doing anything?
Physical attractiveness can play into it, but there are some very charismatic actors who aren't attractive. Acting skill plays into it, but often charismatic actors are only mediocre at "acting". It's commonly associated with confidence, but some charismatic actors have a habit of playing un-confident roles. (Which is not the same as a lack of personal confidence, but what is it they're doing that conveys both "confident" and "insecure" simultaneously?")
It's often said to be about commitment, a sense that they're really "present" and really focused on you. That's certainly something we want actors to do: the more they care about their scene-partners, the more the audience will be drawn to both. (That's true even when the focus is based on a negative emotion, like hatred, but it has to be a really targeted kind of hate and not just a general anger.)
I can teach a lot about the theory of acting, but I have only a vague idea of why it works. When it does, it can be really potent, though it can be be incredibly hard to get. People will often do the same wrong thing harder. A lot of the silly acting games they teach are about getting you to at least do a different thing than what you were doing, hoping that somehow you'll accidentally discover the right track.
I'm not sure any of this is really "charisma" in the sense that this writer means it. I certainly support his overall gist: soft skills are massively underrated.
Had the dotcom bubble not burst I’d likely be an attorney. I’d accepted an offer at a law firm in San Francisco to work in their Securities practice, largely taking companies public or doing M&A.
In March of 2000, the firm called and said: “Good news bad news. Good news: you still have a job [unlike a lot of my law school classmates]. Bad news: we don’t need any more Securities lawyers, but we have lots of room in our Bankruptcy practice.”
Being a Bankruptcy lawyer didn’t sound like fun. A law professor’s brother was starting a B2B startup. He offered me a job. The startup was a colossal failure, but I was hooked on the idea of a group of people starting something from nothing.
Next ~8 years were painful with lots of ideas that went no where, but it all worked out. So, in the end, always remember that but for the dotcom bubble bursting, I’d be keeping track of my time in six minute increments.
