I work at Google on these systems everyday (caveat this is my own words not my employers)). So I simultaneously can tell you that its smart people really thinking about every facet of the problem, and I can't tell you much more than that.
However I can share this written by my colleagues! You'll find great explanations about accelerator architectures and the considerations made to make things fast.
Edit:
Another great resource to look at is the unsloth guides. These folks are incredibly good at getting deep into various models and finding optimizations, and they're very good at writing it up. Here's the Gemma 3n guide, and you'll find others as well.
In the last twenty years, aircraft surveillance has moved from controller-based interrogation to automatic broadcast. The Automatic Dependent Surveillance-Broadcast (ADS-B) is one of the most common methods for aircraft to report their state information like identity, position, and speed. Like other Mode S communications, ADS-B makes use of the 1090 megahertz transponder to transmit data. The protocol for ADS-B is open, and low-cost receivers can easily be used to intercept its signals. Many recent air transportation studies have benefited from this open data source. However, the current literature does not offer a systematic exploration of Mode S and ADS-B data, nor does it explain the decoding process.
This book tackles this missing area in the literature. It offers researchers, engineers, and enthusiasts a clear guide to understanding and making use of open ADS-B and Mode S data. The first part of this book presents the knowledge required to get started with decoding these signals. It includes background information on primary radar, secondary radar, Mode A/C, Mode S, and ADS-B, as well as the hardware and software setups necessary to gather radio signals. After that, the 17 core chapters of the book investigate the details of all types of ADS-B signals and commonly used Mode S signals. Throughout these chapters, examples and sample Python code are used extensively to explain and demonstrate the decoding process. Finally, the last chapter of the book offers a summary and a brief overview of research topics that go beyond the decoding of these signals.
Does seem to support cross-architecture image building, which is a feature that's kept me on Debian's multistrap and now mmdebstrap.
I should give this a go!
The README's reference page has some good write-ups et cetera! There's a good "re-introduction" on Lennart (creator of systemd, mkosi) site, from the current maintainer.
The Internet used to be semi literally a place you went - a desktop in the corner of some room, not central on a desk, not in your pocket. And with a ritual to access it on top of that and the dial up sounds and all.
It's more present but also more invisible now, yeah.
Drew Breunig has been doing some fantastic writing on this subject - coincidentally at the same time as the "context engineering" buzzword appeared but actually unrelated to that meme.
How to Fix Your Context - https://www.dbreunig.com/2025/06/26/how-to-fix-your-context.... - gives names to a bunch of techniques for working around these problems including Tool Loadout, Context Quarantine, Context Pruning, Context Summarization, and Context Offloading.
Absolutely. It serves as a filter, if people are being honest. It also highlights the bizarre dating culture and view of life we've adopted. This dating culture has produced a good deal of rotten fruit.
The ultimate purpose of dating is to meet your future spouse. We're turned it into some kind of senseless sexual escapade, and this has poisoned the relations between men and women. It makes them exploitative and dehumanizing in spirit: sprinkling them with the waters of "consent" doesn't change that, as the subjective cannot abolish the objective. We've reduced sex to something that is merely pleasurable and contradicted its intrinsic and essential function which is procreative by employing an array of technologies that impede and interfere with healthy procreative processes. This creates a mindset not unlike that of a drug user who is obsessed with getting another hit with no thought given to the damage, or the bulimic who wants the sensual satisfaction of eating, but not the calories.
The psychophysical reality of sexual intercourse is much more than some passing physical pleasure. It mobilizes processes in us that are completely oriented toward bonding and the strengthening of the relationship in preparation for children. Whence the stereotype that men will often exit quickly in the morning after a one night stand with a strange woman? Because both can feel, if only subconsciously, that the processes of bonding are taking place, and who wants to bond — and in such a profound and intimate way — with someone they've just met? In this regard, the character of Julianna in Vanilla Sky makes an astoundingly profound and accurate remark for a movie coming out of Hollywood: "Don't you know when you sleep with someone, your body makes a promise whether you do or not?" Our capacity for sexual intimacy is likewise dulled.
(Masturbation is even worse. Those processes bond us with a fictional harem of the imaginary and close us within ourselves. For social animals like us, this is a recipe for misery.)
We thwart and ignore our biological nature to our own detriment. The procreative prime spans the mid-twenties into the early 30s. Statistically, most people should be having families by their mid-20s. Our culture confuses people and creates a pointless obstacle course that leads them to postpone such things either because they're too immature (and encouraged to remain so, also by this unserious dating culture) or because they believe they must achieve some arbitrary milestones first. Furthermore, family and community support has been dashed by a culture of hyperindividualism.
The causes of demographic decline are not a mystery. People simply either don't think deeply enough, or they don't want to make the cultural changes necessary to restore normalcy.
Internet operating systems is a niche subject that I am fascinated with, and I put a list of them on my website. Im going to add OS.js to the list soon.
>> Complexity has to live somewhere. If you are lucky, it lives in well-defined places. In code where you decided a bit of complexity should go, in documentation that supports the code, in training sessions for your engineers. You give it a place without trying to hide all of it. You create ways to manage it. You know where to go to meet it when you need it. If you're unlucky and you just tried to pretend complexity could be avoided altogether, it has no place to go in this world. But it still doesn't stop existing.
>> With nowhere to go, it has to roam everywhere in your system, both in your code and in people's heads. And as people shift around and leave, our understanding of it erodes.
>> Complexity has to live somewhere. If you embrace it, give it the place it deserves, design your system and organisation knowing it exists, and focus on adapting, it might just become a strength.
which covers over a decade, contains many hundreds of entries from Concorde pilots, crew, flight engineers, cabin crew, maintenance personnel, and air traffic control recounting anecdotes and amazing things about the plane and events surrounding it. I started reading it and spent about three absorbing hours fascinated and amazed, unable to stop until it was time to go to bed.
I'm going to resume when I have a few more hours, it's gold.
Nice article about artifacts that make the past more immediate, that allow us to connect our experiences to people hundreds or thousands of years ago.
My favorite example is the writings of Onfim, who was a little boy in the 1200s in present day Russia whose scribbling and homework were exquisitely preserved on birch bark fragments. It’s so immediately recognizable as a little boy’s endearing doodles about knights and imaginary beasts, yet its 800 years old.
It's slow and immensely comprehensive, but also very accessible and so much fun! His love for the subject and enthusiasm are contagious.
Sharing this because thanks to HN I learned about some of my other favourite podcasts: Science & Futurism with Isaac Arthur and The History of Rome by Marc Duncan. I finished THoR twice and I listen to Isaac's podcast every week. I keep coming back to Dante.
yeah I had that. For me I had an early childhood like these kids except my parents weren't even that active in accelerating me. My dad gave me a science book for middle schoolers when I was 4 and taught me my times tables. That, along with my incessant reading that got me reading history and literature at an adult level by the age of at least 10 meant that I was years ahead for most of my school career.
I felt like I was wasting so much time to be in school, and all I got from it was being relentlessly bullied, to the point where I stopped engaging at all out of protest. I started to take pride in being as lazy as possible and getting top marks anyway and teachers who told me to work harder to reach my potential pissed me off because I knew everything they wanted me to work hard on was just too mind numbing to be worth it for me. In classes where I was too objectively ahead to be ignored, like Latin where I had literally already completed the first two textbooks they were using, they deliberately held me behind for everyone else to catch up. So it just seemed like a fucking scam when they told me to "reach my potential"
I crashed and burned at about 17 because of this. Suddenly more long research projects that required discipline that I had purposefully thrown away were required rather than just acing exams effortlessly. I got moved to a private school where the kids around me had actually been receiving a proper education and the standards were higher. And I took one of my worst subjects as a higher level class (maths) and failed exams for the first time in my life. My identity was so tied to being a genius and I was so unused to not succeeding effortlessly I pretty much gave up. Like this shattering of my reality just made me full on depressed.
Since then I've actually become stupider from the extra lack of effort and now I have severe issues dealing with frustration when not being immediately good at things and maintaining discipline or any focus whatsoever. I have so much resentment for the schooling system because I truly believe if I had been moved 2 years up as a kid I would have relished the challenge, kept gaining new skills and not created this learned helplessness for myself.
Open source is a system of prefect logic built on the foundation of a few flawed assumptions.
- Money doesn't matter
- Contributors are benevolent and altruistic
- Commercial interests can't/won't game the process
- Support and security is someone else's responsibility
- Building useful and viable software is a fun hobby
- All software should and will eventually be Open Source
I dropped out of college in 2012 and was one of the very lucky few who managed to find software engineering work almost immediately [1].
I had a bit of a complex about not having a degree, and a few times I tried going back only to drop out again because I would get bored; by the time I had gone back, I already knew enough stuff to be qualified as an engineer, and as such I didn't feel like I was getting a lot out of school and I would paradoxically do pretty poorly because I was half-assing everything.
It wasn't until I found out about WGU in 2021 where I actually decided to finish my degree, primarily because WGU lets me work at my pace. Since I already knew a lot about computer science, I was able to speed through the classes that would have been very boring to me, and I finished my degree really quickly as a result. I don't feel like my education is appreciably worse than people who did things in a traditional brick and mortar school, but I'm not 100% sure if I'm a test for this.
It made me realize that, at least for people like me, EdTech can be extremely powerful stuff. School can be a lot more engaging when it's personalized, instead of the frustrating "one size fits all" of traditional lecturing.
[1] I say "lucky" because I think it was exactly that: luck. Yeah I learned this stuff on my own for fun but finding an employer who was willing to hire someone without credentials was never guaranteed and I feel extremely fortunate to have accidentally timed my dropout about perfectly.
EDIT: For those confused, WGU means "Western Governors University" in this case.
A word of warning, client side support of name constraints may still be incomplete. I know it works on modern Firefox and Chrome, but there's lots of other software that uses HTTPS.
This repo links to BetterTLS, which previously audited name constraint support, but BetterTLS only checked name constraint support at the intermediary certificates not at the trust anchors. I reported[1] the oversight a year back, but Netflix hasn't re-engineered the tests.
Knowing how widely adopted name constraints are on the client side would be really useful, but I haven't seen a sound caniuse style analysis.
Personally, I think the public CA route is better and I built a site that explores this[2].
Entirely eliminating the feeling of guilt is the first step to beating procrastination.
There is a strong negative feedback cycle where you procrastinate, feel guilty about it, and then procrastinate further because you are subconsciously seeking to escape the guilt through distraction.
Happines in society lives in the slack - that is, having rules, but not having them too tightly enforced and with a lot of leeway and context.
The modern mindset prevents that - it's beaurocratic, rigid and extremely oppressive. Technology just helps with that, but if you look at even posts online you'll see that it's the mindset that DEMANDS punishment for any rule breaking no matter how small and how stupid it is.
When predicting changes in attitudes, it's always safer to assume changes will happen to behaviors that started recently than behaviors that started long ago. In the case of mourning, we've done it since before we could write, and many animals do it too, so it is unlikely we'd stop.
Hacker News is a communal content fountain. Lots and lots of websites use the HN rss feed to stock content, and where that content gains traction, it gets propagated elsewhere. Slashdot, Reddit, tons of "news" aggregators, and so on, in a giant graph of relations where HN sits at the center, with the firehose of content getting actually curated and discussed here at the source.
It means your upvotes and downvotes on this site are exponentially more impactful than any site downstream, because anything making it to the front page here will automatically get reposted to hundreds or thousands of other sites.
There is virality, but there are also savvy content producers that know the overall graph of content dissemination, and they certainly utilize that to their benefit.
There are a handful of similar sites and content aggregators that sit in the center like HN that determine what goes viral and what goes stagnant. It's what drew me to HN in the first place, after all the sites I liked were essentially showing me the same thing, and I had to figure out where they were getting their links from.
I love TiddlyWiki. I've been developing a highly non-linear document for some conceptual documentation with it at work.
If you want to see what you can really do with TW, or want to learn how, check out Grok TiddlyWiki (https://groktiddlywiki.com/read/)
I built a personal chess knowledge base with it too. I added a content-type that let me just paste in a plain-text representation of the game and it would display the game with an interactive board.
I very much liked this quote on choosing happiness:
> In the short term, you would be much happier if you accepted and admitted to yourself that the reason you don't have what you want is simply because you do not want it badly enough. The sooner you accept that, the happier you'll be. Then the next question is: Do you want to be happy or do you want to achieve what you want?
It hasn't been updated for Subject Alternative Names or EC keys, but it's sort of like a Linux From Scratch for SSL CA ops that can help you understand what tools like smallstep-cli are doing behind the scenes.
> Those who spend their time flying through imaginary worlds do well to "remember where the off switch is"
If this world is a simulation, and someone among us is the player-character, forgetting that there is an off switch is a feature for them that increases immersion by making any failure to suspend disbelief (which I as a probable NPC suffer from regulary) a moot issue. As long as we think that this is reality, its believability is subordinate to its survivability.
Sport, in general, is about the quest for excellence. It's a tradition that dates back millennia, across nearly all cultures, which suggests it is a deeply-rooted and possibly evolutionary trait in humans. Racing in general, and specifically the 100m dash, the crown jewel of track and field, might be considered one of the purest expressions of athletic excellence.
> Part of what made it so enthralling to read was this grand act of human perversity that a single woman would endure
This is part of why I enjoy watching a couple of RuneScape YouTubers. The sheer time commitments they put in for seemingly minor gains is like gore but for amputating off their own time effort & energy and throwing it into a bottomless maw.
However I can share this written by my colleagues! You'll find great explanations about accelerator architectures and the considerations made to make things fast.
https://jax-ml.github.io/scaling-book/
In particular your questions are around inference which is the focus of this chapter https://jax-ml.github.io/scaling-book/inference/
Edit: Another great resource to look at is the unsloth guides. These folks are incredibly good at getting deep into various models and finding optimizations, and they're very good at writing it up. Here's the Gemma 3n guide, and you'll find others as well.
https://docs.unsloth.ai/basics/gemma-3n-how-to-run-and-fine-...