Full Width [alt+shift+f] Shortcuts [alt+shift+k]
Sign Up [alt+shift+s] Log In [alt+shift+l]

New here?

Welcome! BoredReading is a fresh way to read high quality articles (updated every hour). Our goal is to curate (with your help) Michelin star quality articles (stuff that's really worth reading). We currently have articles in 0 categories from architecture, history, design, technology, and more. Grab a cup of freshly brewed coffee and start reading. This is the best way to increase your attention span, grow as a person, and get a better understanding of the world (or atleast that's why we built it).

35
.title { text-wrap: balance } Spending for October, generated by piping hledger → R Over the past six months, I’ve tracked my money with hledger—a plain text double-entry accounting system written in Haskell. It’s been surprisingly painless. My previous attempts to pick up real accounting tools floundered. Hosted tools are privacy nightmares, and my stint with GnuCash didn’t last. But after stumbling on Dmitry Astapov’s “Full-fledged hledger” wiki1, it clicked—eventually consistent accounting. Instead of modeling your money all at once, take it one hacking session at a time. It should be easy to work towards eventual consistency. […] I should be able to [add financial records] bit by little bit, leaving things half-done, and picking them up later with little (mental) effort. – Dmitry Astapov, Full-Fledged Hledger Principles of my system I’ve cobbled together a system based on these principles: Avoid manual entry – Avoid typing in each transaction. Instead, rely on CSVs from the...
4 months ago

Improve your reading experience

Logged in users get linked directly to articles resulting in a better reading experience. Please login for free, it takes less than 1 minute.

More from Tyler Cipriani: blog

Subliminal git commits

Luckily, I speak Leet. – Amita Ramanujan, Numb3rs, CBS’s IRC Drama There’s an episode of the CBS prime-time drama Numb3rs that plumbs the depths of Dr. Joel Fleischman’s1 knowledge of IRC. In one scene, Fleischman wonders, “What’s ‘leet’”? “Leet” is writing that replaces letters with numbers, e.g., “Numb3rs,” where 3 stands in for e. In short, leet is like the heavy-metal “S” you drew in middle school: Sweeeeet. / \ / | \ | | | \ \ | | | \ | / \ / ASCII art version of your misspent youth. Following years of keen observation, I’ve noticed Git commit hashes are also letters and numbers. Git commit hashes are, as Fleischman might say, prime targets for l33tification. What can I spell with a git commit? DenITDao via orlybooks) With hexidecimal we can spell any word containing the set of letters {A, B, C, D, E, F}—DEADBEEF (a classic) or ABBABABE (for Mama Mia aficionados). This is because hexidecimal is a base-16 numbering system—a single “digit” represents 16 numbers: Base-10: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 16 15 Base-16: 0 1 2 3 4 5 6 7 8 9 A B C D E F Leet expands our palette of words—using 0, 1, and 5 to represent O, I, and S, respectively. I created a script that scours a few word lists for valid words and phrases. With it, I found masterpieces like DADB0D (dad bod), BADA55 (bad ass), and 5ADBAB1E5 (sad babies). Manipulating commit hashes for fun and no profit Git commit hashes are no mystery. A commit hash is the SHA-1 of a commit object. And a commit object is the commit message with some metadata. $ mkdir /tmp/BADA55-git && cd /tmp/BAD55-git $ git init Initialized empty Git repository in /tmp/BADA55-git/.git/ $ echo '# BADA55 git repo' > README.md && git add README.md && git commit -m 'Initial commit' [main (root-commit) 68ec0dd] Initial commit 1 file changed, 1 insertion(+) create mode 100644 README.md $ git log --oneline 68ec0dd (HEAD -> main) Initial commit Let’s confirm we can recreate the commit hash: $ git cat-file -p 68ec0dd > commit-msg $ sha1sum <(cat \ <(printf "commit ") \ <(wc -c < commit-msg | tr -d '\n') \ <(printf '%b' '\0') commit-msg) 68ec0dd6dead532f18082b72beeb73bd828ee8fc /dev/fd/63 Our repo’s first commit has the hash 68ec0dd. My goal is: Make 68ec0dd be BADA55. Keep the commit message the same, visibly at least. But I’ll need to change the commit to change the hash. To keep those changes invisible in the output of git log, I’ll add a \t and see what happens to the hash. $ truncate -s -1 commit-msg # remove final newline $ printf '\t\n' >> commit-msg # Add a tab $ # Check the new SHA to see if it's BADA55 $ sha1sum <(cat \ <(printf "commit ") \ <(wc -c < commit-msg | tr -d '\n') \ <(printf '%b' '\0') commit-msg) 27b22ba5e1c837a34329891c15408208a944aa24 /dev/fd/63 Success! I changed the SHA-1. Now to do this until we get to BADA55. Fortunately, user not-an-aardvark created a tool for that—lucky-commit that manipulates a commit message, adding a combination of \t and [:space:] characters until you hit a desired SHA-1. Written in rust, lucky-commit computes all 256 unique 8-bit strings composed of only tabs and spaces. And then pads out commits up to 48-bits with those strings, using worker threads to quickly compute the SHA-12 of each commit. It’s pretty fast: $ time lucky_commit BADA555 real 0m0.091s user 0m0.653s sys 0m0.007s $ git log --oneline bada555 (HEAD -> main) Initial commit $ xxd -c1 <(git cat-file -p 68ec0dd) | grep -cPo ': (20|09)' 12 $ xxd -c1 <(git cat-file -p HEAD) | grep -cPo ': (20|09)' 111 Now we have an more than an initial commit. We have a BADA555 initial commit. All that’s left to do is to make ALL our commits BADA55 by abusing git hooks. $ cat > .git/hooks/post-commit && chmod +x .git/hooks/post-commit #!/usr/bin/env bash echo 'L337-ifying!' lucky_commit BADA55 $ echo 'A repo that is very l33t.' >> README.md && git commit -a -m 'l33t' L337-ifying! [main 0e00cb2] l33t 1 file changed, 1 insertion(+) $ git log --oneline bada552 (HEAD -> main) l33t bada555 Initial commit And now I have a git repo almost as cool as the sweet “S” I drew in middle school. This is a Northern Exposure spin off, right? I’ve only seen 1:48 of the show…↩︎ or SHA-256 for repos that have made the jump to a more secure hash function↩︎

5 months ago 49 votes
The Pull Request

A brief and biased history. Oh yeah, there’s pull requests now – GitHub blog, Sat, 23 Feb 2008 When GitHub launched, it had no code review. Three years after launch, in 2011, GitHub user rtomayko became the first person to make a real code comment, which read, in full: “+1”. Before that, GitHub lacked any way to comment on code directly. Instead, pull requests were a combination of two simple features: Cross repository compare view – a feature they’d debuted in 2010—git diff in a web page. A comments section – a feature most blogs had in the 90s. There was no way to thread comments, and the comments were on a different page than the diff. GitHub pull requests circa 2010. This is from the official documentation on GitHub. Earlier still, when the pull request debuted, GitHub claimed only that pull requests were “a way to poke someone about code”—a way to direct message maintainers, but one that lacked any web view of the code whatsoever. For developers, it worked like this: Make a fork. Click “pull request”. Write a message in a text form. Send the message to someone1 with a link to your fork. Wait for them to reply. In effect, pull requests were a limited way to send emails to other GitHub users. Ten years after this humble beginning—seven years after the first code comment—when Microsoft acquired GitHub for $7.5 Billion, this cobbled-together system known as “GitHub flow” had become the default way to collaborate on code via Git. And I hate it. Pull requests were never designed. They emerged. But not from careful consideration of the needs of developers or maintainers. Pull requests work like they do because they were easy to build. In 2008, GitHub’s developers could have opted to use git format-patch instead of teaching the world to juggle branches. Or they might have chosen to generate pull requests using the git request-pull command that’s existed in Git since 2005 and is still used by the Linux kernel maintainers today2. Instead, they shrugged into GitHub flow, and that flow taught the world to use Git. And commit histories have sucked ever since. For some reason, github has attracted people who have zero taste, don’t care about commit logs, and can’t be bothered. – Linus Torvalds, 2012 “Someone” was a person chosen by you from a checklist of the people who had also forked this repository at some point.↩︎ Though to make small, contained changes you’d use git format-patch and git am.↩︎

6 months ago 64 votes
Git the stupid password store

.title {text-wrap:balance;} GIT - the stupid content tracker “git” can mean anything, depending on your mood. – Linus Torvalds, Initial revision of “git”, the information manager from hell Like most git features, gitcredentials(7) are obscure, byzantine, and incredibly useful. And, for me, they’re a nice, hacky solution to a simple problem. Problem: Home directories teeming with tokens. Too many programs store cleartext credentials in config files in my home directory, making exfiltration all too easy. Solution: For programs I write, I can use git credential fill – the password library I never knew I installed. #!/usr/bin/env bash input="\ protocol=https host=example.com user=thcipriani " eval "$(echo "$input" | git credential fill)" echo "The password is: $password" Which looks like this when you run it: $ ./prompt.sh Password for 'https://thcipriani@example.com': The password is: hunter2 What did git credentials fill do? Accepted a protocol, username, and host on standard input. Called out to my git credential helper My credential helper checked for credentials matching https://thcipriani@example.com and found nothing Since my credential helper came up empty, it prompted me for my password Finally, it echoed <key>=<value>\n pairs for the keys protocol, host, username, and password to standard output. If I want, I can tell my credential helper to store the information I entered: git credential approve <<EOF protocol=$protocol username=$username host=$host password=$password EOF If I do that, the next time I run the script, it finds the password without prompting: $ ./prompt.sh The password is: hunter2 What are git credentials? Surprisingly, the intended purpose of git credentials is NOT “a weird way to prompt for passwords.” The problem git credentials solve is this: With git over ssh, you use your keys. With git over https, you type a password. Over and over and over. Beleaguered git maintainers solved this dilemma with the credential storage system—git credentials. With the right configuration, git will stop asking for your password when you push to an https remote. Instead, git credentials retrieve and send auth info to remotes. On the labyrinthine options of git credentials My mind initially refused to learn git credentials due to its twisty maze of terms that all sound alike: git credential fill: how you invoke a user’s configured git credential helper git credential approve: how you save git credentials (if this is supported by the user’s git credential helper) git credential.helper: the git config that points to a script that poops out usernames and passwords. These helper scripts are often named git-credential-<something>. git-credential-cache: a specific, built-in git credential helper that caches credentials in memory for a while. git-credential-store: STOP. DON’T TOUCH. This is a specific, built-in git credential helper that stores credentials in cleartext in your home directory. Whomp whomp. git-credential-manager: a specific and confusingly named git credential helper from Microsoft®. If you’re on Linux or Mac, feel free to ignore it. But once I mapped the terms, I only needed to pick a git credential helper. Configuring good credential helpers The built-in git-credential-store is a bad credential helper—it saves your passwords in cleartext in ~/.git-credentials.1 If you’re on a Mac, you’re in luck2—one command points git credentials to your keychain: git config --global credential.helper osxkeychain Third-party developers have contributed helpers for popular password stores: 1Password pass: the standard Unix password manager OAuth Git’s documentation contains a list of credential-helpers, too Meanwhile, Linux and Windows have standard options. Git’s source repo includes helpers for these options in the contrib directory. On Linux, you can use libsecret. Here’s how I configured it on Debian: sudo apt install libsecret-1-0 libsecret-1-dev cd /usr/share/doc/git/contrib/credential/libsecret/ sudo make sudo mv git-credential-libsecret /usr/local/bin/ git config --global credential.helper libsecret On Windows, you can use the confusingly named git credential manager. I have no idea how to do this, and I refuse to learn. Now, if you clone a repo over https, you can push over https without pain3. Plus, you have a handy trick for shell scripts. git-credential-store is not a git credential helper of honor. No highly-esteemed passwords should be stored with it. This message is a warning about danger. The danger is still present, in your time, as it was in ours.↩︎ I think. I only have Linux computers to test this on, sorry ;_;↩︎ Or the config option pushInsteadOf, which is what I actually do.↩︎

7 months ago 50 votes
Hexadecimal Sucks

Humans do no operate on hexadecimal symbols effectively […] there are exceptions. – Dan Kaminsky When SSH added ASCII art fingerprints (AKA, randomart), the author credited a talk by Dan Kaminsky. As a refresher, randomart looks like this: $ ssh-keygen -lv -f ~/.ssh/id_ed25519.pub 256 SHA256:XrvNnhQuG1ObprgdtPiqIGXUAsHT71SKh9/WAcAKoS0 thcipriani@foo.bar (ED25519) +--[ED25519 256]--+ | .++ ... | | o+.... o | |E .oo=.o . | | . .+.= . | | o= .S.o.o | | o o.o+.= + | | . . .o B * | | . . + & . | | ..+o*.= | +----[SHA256]-----+ Ben Cox describes the algorithm for generating random art on his blog. Here’s a slo-mo version of the algorithm in action: ASCII art ssh fingerprints slo-mo algorithm But in Dan’s talk, he never mentions anything about ASCII art. Instead, his talk was about exploiting our brain’s hardware acceleration to make it easier for us to recognize SSH fingerprints. The talk is worth watching, but I’ll attempt a summary. What’s the problem? We’ll never memorize SHA256:XrvNnhQuG1ObprgdtPiqIGXUAsHT71SKh9/WAcAKoS0—hexadecimal and base64 were built to encode large amounts of information rather than be easy to remember. But that’s ok for SSH keys because there are different kinds of memory: Rejection: I’ve never seen that before! Recognition: I know it’s that one—not the other one. Recollection: rote recall, like a phone number or address. For SSH you’ll use recognition—do you recognize this key? Of course, SSH keys are still a problem because our working memory is too small to recognize such long strings of letters and numbers. Hacks abound to shore up our paltry working memory—what Dan called “brain hardware acceleration.” Randomart attempts to tap into our hardware acceleration for pattern recognition—the visiuo-spacial sketchpad, where we store pictures. Dan’s idea tapped into a different aspect of hardware acceleration, one often cited by memory competition champions: chunking. Memory chunking and sha256 The web service what3words maps every three cubic meters (3m²) on Earth to three words. The White House’s Oval Office is ///curve.empty.buzz. Three words encode the same information as latitude and longitude—38.89, -77.03—chunking the information to be small enough to fit in our working memory. The mapping of locations to words uses a list of 40 thousand common English words, so each word encodes 15.29 bits of information—45.9 bits of information, identifying 64 trillion unique places. Meanwhile sha256 is 256 bits of information: ~116 quindecillion unique combinations. 64000000000000 # 64 trillion (what3words) 115792089237316195423570985008687907853269984665640564039457584007913129639936 # 116 (ish) quindecillion (sha256) For SHA256, we need more than three words or a dictionary larger than 40,000 words. Dan’s insight was we can identify SSH fingerprints using pairs of human names—couples. The math works like this1: 131,072 first names: 17 bits per name (×2) 524,288 last names: 19 bits per name 2,048 cities: 11 bits per city 17+17+19+11 = 64 bits With 64 bits per couple, you could uniquely identify 116 quindecillion items with four couples. Turning this: $ ssh foo.bar The authenticity of host 'foo.bar' can't be established. ED25519 key fingerprint is SHA256:XrvNnhQuG1ObprgdtPiqIGXUAsHT71SKh9/WAcAKoS0. Are you sure you want to continue connecting (yes/no/[fingerprint])? Into this2: $ ssh foo.bar The authenticity of host 'foo.bar' can't be established. SHA256:XrvNnhQuG1ObprgdtPiqIGXUAsHT71SKh9/WAcAKoS0 Key Data: Svasse and Tainen Jesudasson from Fort Wayne, Indiana, United States Illma and Sibeth Primack from Itārsi, Madhya Pradesh, India Maarja and Nisim Balyeat from Mukilteo, Washington, United States Hsu-Heng and Rasim Haozi from Manali, Tamil Nadu, India Are you sure you want to continue connecting (yes/no/[fingerprint])? With enough exposure, building recognition for these names and places should be possible—at least more possible than memorizing host keys. I’ve modified this from the original talk, in 2006 we were using md5 fingerprints of 160-bits. Now we’re using 256-bit fingerprints, so we needed to encode even more information, but the idea still works.↩︎ A (very) rough code implementation is on my github.↩︎

9 months ago 61 votes

More in programming

AI: Where in the Loop Should Humans Go?

This is a re-publishing of a blog post I originally wrote for work, but wanted on my own blog as well. AI is everywhere, and its impressive claims are leading to rapid adoption. At this stage, I’d qualify it as charismatic technology—something that under-delivers on what it promises, but promises so much that the industry still leverages it because we believe it will eventually deliver on these claims. This is a known pattern. In this post, I’ll use the example of automation deployments to go over known patterns and risks in order to provide you with a list of questions to ask about potential AI solutions. I’ll first cover a short list of base assumptions, and then borrow from scholars of cognitive systems engineering and resilience engineering to list said criteria. At the core of it is the idea that when we say we want humans in the loop, it really matters where in the loop they are. My base assumptions The first thing I’m going to say is that we currently do not have Artificial General Intelligence (AGI). I don’t care whether we have it in 2 years or 40 years or never; if I’m looking to deploy a tool (or an agent) that is supposed to do stuff to my production environments, it has to be able to do it now. I am not looking to be impressed, I am looking to make my life and the system better. Another mechanism I want you to keep in mind is something called the context gap. In a nutshell, any model or automation is constructed from a narrow definition of a controlled environment, which can expand as it gains autonomy, but remains limited. By comparison, people in a system start from a broad situation and narrow definitions down and add constraints to make problem-solving tractable. One side starts from a narrow context, and one starts from a wide one—so in practice, with humans and machines, you end up seeing a type of teamwork where one constantly updates the other: The optimal solution of a model is not an optimal solution of a problem unless the model is a perfect representation of the problem, which it never is.  — Ackoff (1979, p. 97) Because of that mindset, I will disregard all arguments of “it’s coming soon” and “it’s getting better real fast” and instead frame what current LLM solutions are shaped like: tools and automation. As it turns out, there are lots of studies about ergonomics, tool design, collaborative design, where semi-autonomous components fit into sociotechnical systems, and how they tend to fail. Additionally, I’ll borrow from the framing used by people who study joint cognitive systems: rather than looking only at the abilities of what a single person or tool can do, we’re going to look at the overall performance of the joint system. This is important because if you have a tool that is built to be operated like an autonomous agent, you can get weird results in your integration. You’re essentially building an interface for the wrong kind of component—like using a joystick to ride a bicycle. This lens will assist us in establishing general criteria about where the problems will likely be without having to test for every single one and evaluate them on benchmarks against each other. Questions you'll want to ask The following list of questions is meant to act as reminders—abstracting away all the theory from research papers you’d need to read—to let you think through some of the important stuff your teams should track, whether they are engineers using code generation, SREs using AIOps, or managers and execs making the call to adopt new tooling. Are you better even after the tool is taken away? An interesting warning comes from studying how LLMs function as learning aides. The researchers found that people who trained using LLMs tended to fail tests more when the LLMs were taken away compared to people who never studied with them, except if the prompts were specifically (and successfully) designed to help people learn. Likewise, it’s been known for decades that when automation handles standard challenges, the operators expected to take over when they reach their limits end up worse off and generally require more training to keep the overall system performant. While people can feel like they’re getting better and more productive with tool assistance, it doesn’t necessarily follow that they are learning or improving. Over time, there’s a serious risk that your overall system’s performance will be limited to what the automation can do—because without proper design, people keeping the automation in check will gradually lose the skills they had developed prior. Are you augmenting the person or the computer? Traditionally successful tools tend to work on the principle that they improve the physical or mental abilities of their operator: search tools let you go through more data than you could on your own and shift demands to external memory, a bicycle more effectively transmits force for locomotion, a blind spot alert on your car can extend your ability to pay attention to your surroundings, and so on. Automation that augments users therefore tends to be easier to direct, and sort of extends the person’s abilities, rather than acting based on preset goals and framing. Automation that augments a machine tends to broaden the device’s scope and control by leveraging some known effects of their environment and successfully hiding them away. For software folks, an autoscaling controller is a good example of the latter. Neither is fundamentally better nor worse than the other—but you should figure out what kind of automation you’re getting, because they fail differently. Augmenting the user implies that they can tackle a broader variety of challenges effectively. Augmenting the computers tends to mean that when the component reaches its limits, the challenges are worse for the operator. Is it turning you into a monitor rather than helping build an understanding? If your job is to look at the tool go and then say whether it was doing a good or bad job (and maybe take over if it does a bad job), you’re going to have problems. It has long been known that people adapt to their tools, and automation can create complacency. Self-driving cars that generally self-drive themselves well but still require a monitor are not effectively monitored. Instead, having AI that supports people or adds perspectives to the work an operator is already doing tends to yield better long-term results than patterns where the human learns to mostly delegate and focus elsewhere. (As a side note, this is why I tend to dislike incident summarizers. Don’t make it so people stop trying to piece together what happened! Instead, I prefer seeing tools that look at your summaries to remind you of items you may have forgotten, or that look for linguistic cues that point to biases or reductive points of view.) Does it pigeonhole what you can look at? When evaluating a tool, you should ask questions about where the automation lands: Does it let you look at the world more effectively? Does it tell you where to look in the world? Does it force you to look somewhere specific? Does it tell you to do something specific? Does it force you to do something? This is a bit of a hybrid between “Does it extend you?” and “Is it turning you into a monitor?” The five questions above let you figure that out. As the tool becomes a source of assertions or constraints (rather than a source of information and options), the operator becomes someone who interacts with the world from inside the tool rather than someone who interacts with the world with the tool’s help. The tool stops being a tool and becomes a representation of the whole system, which means whatever limitations and internal constraints it has are then transmitted to your users. Is it a built-in distraction? People tend to do multiple tasks over many contexts. Some automated systems are built with alarms or alerts that require stealing someone’s focus, and unless they truly are the most critical thing their users could give attention to, they are going to be an annoyance that can lower the effectiveness of the overall system. What perspectives does it bake in? Tools tend to embody a given perspective. For example, AIOps tools that are built to find a root cause will likely carry the conceptual framework behind root causes in their design. More subtly, these perspectives are sometimes hidden in the type of data you get: if your AIOps agent can only see alerts, your telemetry data, and maybe your code, it will rarely be a source of suggestions on how to improve your workflows because that isn’t part of its world. In roles that are inherently about pulling context from many disconnected sources, how on earth is automation going to make the right decisions? And moreover, who’s accountable for when it makes a poor decision on incomplete data? Surely not the buyer who installed it! This is also one of the many ways in which automation can reinforce biases—not just based on what is in its training data, but also based on its own structure and what inputs were considered most important at design time. The tool can itself become a keyhole through which your conclusions are guided. Is it going to become a hero? A common trope in incident response is heroes—the few people who know everything inside and out, and who end up being necessary bottlenecks to all emergencies. They can’t go away for vacation, they’re too busy to train others, they develop blind spots that nobody can fix, and they can’t be replaced. To avoid this, you have to maintain a continuous awareness of who knows what, and crosstrain each other to always have enough redundancy. If you have a team of multiple engineers and you add AI to it, having it do all of the tasks of a specific kind means it becomes a de facto hero to your team. If that’s okay, be aware that any outages or dysfunction in the AI agent would likely have no practical workaround. You will essentially have offshored part of your ops. Do you need it to be perfect? What a thing promises to be is never what it is—otherwise AWS would be enough, and Kubernetes would be enough, and JIRA would be enough, and the software would work fine with no one needing to fix things. That just doesn’t happen. Ever. Even if it’s really, really good, it’s gonna have outages and surprises, and it’ll mess up here and there, no matter what it is. We aren’t building an omnipotent computer god, we’re building imperfect software. You’ll want to seriously consider whether the tradeoffs you’d make in terms of quality and cost are worth it, and this is going to be a case-by-case basis. Just be careful not to fix the problem by adding a human in the loop that acts as a monitor! Is it doing the whole job or a fraction of it? We don’t notice major parts of our own jobs because they feel natural. A classic pattern here is one of AIs getting better at diagnosing patients, except the benchmarks are usually run on a patient chart where most of the relevant observations have already been made by someone else. Similarly, we often see AI pass a test with flying colors while it still can’t be productive at the job the test represents. People in general have adopted a model of cognition based on information processing that’s very similar to how computers work (get data in, think, output stuff, rinse and repeat), but for decades, there have been multiple disciplines that looked harder at situated work and cognition, moving past that model. Key patterns of cognition are not just in the mind, but are also embedded in the environment and in the interactions we have with each other. Be wary of acquiring a solution that solves what you think the problem is rather than what it actually is. We routinely show we don’t accurately know the latter. What if we have more than one? You probably know how straightforward it can be to write a toy project on your own, with full control of every refactor. You probably also know how this stops being true as your team grows. As it stands today, a lot of AI agents are built within a snapshot of the current world: one or few AI tools added to teams that are mostly made up of people. By analogy, this would be like everyone selling you a computer assuming it were the first and only electronic device inside your household. Problems arise when you go beyond these assumptions: maybe AI that writes code has to go through a code review process, but what if that code review is done by another unrelated AI agent? What happens when you get to operations and common mode failures impact components from various teams that all have agents empowered to go fix things to the best of their ability with the available data? Are they going to clash with people, or even with each other? Humans also have that ability and tend to solve it via processes and procedures, explicit coordination, announcing what they’ll do before they do it, and calling upon each other when they need help. Will multiple agents require something equivalent, and if so, do you have it in place? How do they cope with limited context? Some changes that cause issues might be safe to roll back, some not (maybe they include database migrations, maybe it is better to be down than corrupting data), and some may contain changes that rolling back wouldn’t fix (maybe the workload is controlled by one or more feature flags). Knowing what to do in these situations can sometimes be understood from code or release notes, but some situations can require different workflows involving broader parts of the organization. A risk of automation without context is that if you have situations where waiting or doing little is the best option, then you’ll need to either have automation that requires input to act, or a set of actions to quickly disable multiple types of automation as fast as possible. Many of these may exist at the same time, and it becomes the operators’ jobs to not only maintain their own context, but also maintain a mental model of the context each of these pieces of automation has access to. The fancier your agents, the fancier your operators’ understanding and abilities must be to properly orchestrate them. The more surprising your landscape is, the harder it can become to manage with semi-autonomous elements roaming around. After an outage or incident, who does the learning and who does the fixing? One way to track accountability in a system is to figure out who ends up having to learn lessons and change how things are done. It’s not always the same people or teams, and generally, learning will happen whether you want it or not. This is more of a rhetorical question right now, because I expect that in most cases, when things go wrong, whoever is expected to monitor the AI tool is going to have to steer it in a better direction and fix it (if they can); if it can’t be fixed, then the expectation will be that the automation, as a tool, will be used more judiciously in the future. In a nutshell, if the expectation is that your engineers are going to be doing the learning and tweaking, your AI isn’t an independent agent—it’s a tool that cosplays as an independent agent. Do what you will—just be mindful All in all, none of the above questions flat out say you should not use AI, nor where exactly in the loop you should put people. The key point is that you should ask that question and be aware that just adding whatever to your system is not going to substitute workers away. It will, instead, transform work and create new patterns and weaknesses. Some of these patterns are known and well-studied. We don’t have to go rushing to rediscover them all through failures as if we were the first to ever automate something. If AI ever gets so good and so smart that it’s better than all your engineers, it won’t make a difference whether you adopt it only once it’s good. In the meanwhile, these things do matter and have real impacts, so please design your systems responsibly. If you’re interested to know more about the theoretical elements underpinning this post, the following references—on top of whatever was already linked in the text—might be of interest: Books: Joint Cognitive Systems: Foundations of Cognitive Systems Engineering by Erik Hollnagel Joint Cognitive Systems: Patterns in Cognitive Systems Engineering by David D. Woods Cognition in the Wild by Edwin Hutchins Behind Human Error by David D. Woods, Sydney Dekker, Richard Cook, Leila Johannesen, Nadine Sarter Papers: Ironies of Automation by Lisanne Bainbridge The French-Speaking Ergonomists’ Approach to Work Activity by Daniellou How in the World Did We Ever Get into That Mode? Mode Error and Awareness in Supervisory Control by Nadine Sarter Can We Ever Escape from Data Overload? A Cognitive Systems Diagnosis by David D. Woods Ten Challenges for Making Automation a “Team Player” in Joint Human-Agent Activity by Gary Klein and David D. Woods MABA-MABA or Abracadabra? Progress on Human–Automation Co-ordination by Sidney Dekker Managing the Hidden Costs of Coordination by Laura Maguire Designing for Expertise by David D. Woods The Impact of Generative AI on Critical Thinking by Lee et al.

yesterday 4 votes
AMD YOLO

AMD is sending us the two MI300X boxes we asked for. They are in the mail. It took a bit, but AMD passed my cultural test. I now believe they aren’t going to shoot themselves in the foot on software, and if that’s true, there’s absolutely no reason they should be worth 1/16th of NVIDIA. CUDA isn’t really the moat people think it is, it was just an early ecosystem. tiny corp has a fully sovereign AMD stack, and soon we’ll port it to the MI300X. You won’t even have to use tinygrad proper, tinygrad has a torch frontend now. Either NVIDIA is super overvalued or AMD is undervalued. If the petaflop gets commoditized (tiny corp’s mission), the current situation doesn’t make any sense. The hardware is similar, AMD even got the double throughput Tensor Cores on RDNA4 (NVIDIA artificially halves this on their cards, soon they won’t be able to). I’m betting on AMD being undervalued, and that the demand for AI has barely started. With good software, the MI300X should outperform the H100. In for a quarter million. Long term. It can always dip short term, but check back in 5 years.

yesterday 2 votes
whippet lab notebook: untagged mallocs, bis

Earlier this weekGuileWhippet But now I do! Today’s note is about how we can support untagged allocations of a few different kinds in Whippet’s .mostly-marking collector Why bother supporting untagged allocations at all? Well, if I had my way, I wouldn’t; I would just slog through Guile and fix all uses to be tagged. There are only a finite number of use sites and I could get to them all in a month or so. The problem comes for uses of from outside itself, in C extensions and embedding programs. These users are loathe to adapt to any kind of change, and garbage-collection-related changes are the worst. So, somehow, we need to support these users if we are not to break the Guile community.scm_gc_malloclibguile The problem with , though, is that it is missing an expression of intent, notably as regards tagging. You can use it to allocate an object that has a tag and thus can be traced precisely, or you can use it to allocate, well, anything else. I think we will have to add an API for the tagged case and assume that anything that goes through is requesting an untagged, conservatively-scanned block of memory. Similarly for : you could be allocating a tagged object that happens to not contain pointers, or you could be allocating an untagged array of whatever. A new API is needed there too for pointerless untagged allocations.scm_gc_mallocscm_gc_mallocscm_gc_malloc_pointerless Recall that the mostly-marking collector can be built in a number of different ways: it can support conservative and/or precise roots, it can trace the heap precisely or conservatively, it can be generational or not, and the collector can use multiple threads during pauses or not. Consider a basic configuration with precise roots. You can make tagged pointerless allocations just fine: the trace function for that tag is just trivial. You would like to extend the collector with the ability to make pointerless allocations, for raw data. How to do this?untagged Consider first that when the collector goes to trace an object, it can’t use bits inside the object to discriminate between the tagged and untagged cases. Fortunately though . Of those 8 bits, 3 are used for the mark (five different states, allowing for future concurrent tracing), two for the , one to indicate whether the object is pinned or not, and one to indicate the end of the object, so that we can determine object bounds just by scanning the metadata byte array. That leaves 1 bit, and we can use it to indicate untagged pointerless allocations. Hooray!the main space of the mostly-marking collector has one metadata byte for each 16 bytes of payloadprecise field-logging write barrier However there is a wrinkle: when Whippet decides the it should evacuate an object, it tracks the evacuation state in the object itself; the embedder has to provide an implementation of a , allowing the collector to detect whether an object is forwarded or not, to claim an object for forwarding, to commit a forwarding pointer, and so on. We can’t do that for raw data, because all bit states belong to the object, not the collector or the embedder. So, we have to set the “pinned” bit on the object, indicating that these objects can’t move.little state machine We could in theory manage the forwarding state in the metadata byte, but we don’t have the bits to do that currently; maybe some day. For now, untagged pointerless allocations are pinned. You might also want to support untagged allocations that contain pointers to other GC-managed objects. In this case you would want these untagged allocations to be scanned conservatively. We can do this, but if we do, it will pin all objects. Thing is, conservative stack roots is a kind of a sweet spot in language run-time design. You get to avoid constraining your compiler, you avoid a class of bugs related to rooting, but you can still support compaction of the heap. How is this, you ask? Well, consider that you can move any object for which we can precisely enumerate the incoming references. This is trivially the case for precise roots and precise tracing. For conservative roots, we don’t know whether a given edge is really an object reference or not, so we have to conservatively avoid moving those objects. But once you are done tracing conservative edges, any live object that hasn’t yet been traced is fair game for evacuation, because none of its predecessors have yet been visited. But once you add conservatively-traced objects back into the mix, you don’t know when you are done tracing conservative edges; you could always discover another conservatively-traced object later in the trace, so you have to pin everything. The good news, though, is that we have gained an easier migration path. I can now shove Whippet into Guile and get it running even before I have removed untagged allocations. Once I have done so, I will be able to allow for compaction / evacuation; things only get better from here. Also as a side benefit, the mostly-marking collector’s heap-conservative configurations are now faster, because we have metadata attached to objects which allows tracing to skip known-pointerless objects. This regains an optimization that BDW has long had via its , used in Guile since time out of mind.GC_malloc_atomic With support for untagged allocations, I think I am finally ready to start getting Whippet into Guile itself. Happy hacking, and see you on the other side! inside and outside on intent on data on slop fin

yesterday 2 votes
Creating static map images with OpenStreetMap, Web Mercator, and Pillow

I’ve been working on a project where I need to plot points on a map. I don’t need an interactive or dynamic visualisation – just a static map with coloured dots for each coordinate. I’ve created maps on the web using Leaflet.js, which load map data from OpenStreetMap (OSM) and support zooming and panning – but for this project, I want a standalone image rather than something I embed in a web page. I want to put in coordinates, and get a PNG image back. This feels like it should be straightforward. There are lots of Python libraries for data visualisation, but it’s not an area I’ve ever explored in detail. I don’t know how to use these libraries, and despite trying I couldn’t work out how to accomplish this seemingly simple task. I made several attempts with libraries like matplotlib and plotly, but I felt like I was fighting the tools. Rather than persist, I wrote my own solution with “lower level” tools. The key was a page on the OpenStreetMap wiki explaining how to convert lat/lon coordinates into the pixel system used by OSM tiles. In particular, it allowed me to break the process into two steps: Get a “base map” image that covers the entire world Convert lat/lon coordinates into xy coordinates that can be overlaid on this image Let’s go through those steps. Get a “base map” image that covers the entire world Let’s talk about how OpenStreetMap works, and in particular their image tiles. If you start at the most zoomed-out level, OSM represents the entire world with a single 256×256 pixel square. This is the Web Mercator projection, and you don’t get much detail – just a rough outline of the world. We can zoom in, and this tile splits into four new tiles of the same size. There are twice as many pixels along each edge, and each tile has more detail. Notice that country boundaries are visible now, but we can’t see any names yet. We can zoom in even further, and each of these tiles split again. There still aren’t any text labels, but the map is getting more detailed and we can see small features that weren’t visible before. You get the idea – we could keep zooming, and we’d get more and more tiles, each with more detail. This tile system means you can get detailed information for a specific area, without loading the entire world. For example, if I’m looking at street information in Britain, I only need the detailed tiles for that part of the world. I don’t need the detailed tiles for Bolivia at the same time. OpenStreetMap will only give you 256×256 pixels at a time, but we can download every tile and stitch them together, one-by-one. Here’s a Python script that enumerates all the tiles at a particular zoom level, downloads them, and uses the Pillow library to combine them into a single large image: #!/usr/bin/env python3 """ Download all the map tiles for a particular zoom level from OpenStreetMap, and stitch them into a single image. """ import io import itertools import httpx from PIL import Image zoom_level = 2 width = 256 * 2**zoom_level height = 256 * (2**zoom_level) im = Image.new("RGB", (width, height)) for x, y in itertools.product(range(2**zoom_level), range(2**zoom_level)): resp = httpx.get(f"https://tile.openstreetmap.org/{zoom_level}/{x}/{y}.png", timeout=50) resp.raise_for_status() im_buffer = Image.open(io.BytesIO(resp.content)) im.paste(im_buffer, (x * 256, y * 256)) out_path = f"map_{zoom_level}.png" im.save(out_path) print(out_path) The higher the zoom level, the more tiles you need to download, and the larger the final image will be. I ran this script up to zoom level 6, and this is the data involved: Zoom level Number of tiles Pixels File size 0 1 256×256 17.1 kB 1 4 512×512 56.3 kB 2 16 1024×1024 155.2 kB 3 64 2048×2048 506.4 kB 4 256 4096×4096 2.7 MB 5 1,024 8192×8192 13.9 MB 6 4,096 16384×16384 46.1 MB I can just about open that zoom level 6 image on my computer, but it’s struggling. I didn’t try opening zoom level 7 – that includes 16,384 tiles, and I’d probably run out of memory. For most static images, zoom level 3 or 4 should be sufficient – I ended up a base map from zoom level 4 for my project. It takes a minute or so to download all the tiles from OpenStreetMap, but you only need to request it once, and then you have a static image you can use again and again. This is a particularly good approach if you want to draw a lot of maps. OpenStreetMap is provided for free, and we want to be a respectful user of the service. Downloading all the map tiles once is more efficient than making repeated requests for the same data. Overlay lat/lon coordinates on this base map Now we have an image with a map of the whole world, we need to overlay our lat/lon coordinates as points on this map. I found instructions on the OpenStreetMap wiki which explain how to convert GPS coordinates into a position on the unit square, which we can in turn add to our map. They outline a straightforward algorithm, which I implemented in Python: import math def convert_gps_coordinates_to_unit_xy( *, latitude: float, longitude: float ) -> tuple[float, float]: """ Convert GPS coordinates to positions on the unit square, which can be plotted on a Web Mercator projection of the world. This expects the coordinates to be specified in **degrees**. The result will be (x, y) coordinates: - x will fall in the range (0, 1). x=0 is the left (180° west) edge of the map. x=1 is the right (180° east) edge of the map. x=0.5 is the middle, the prime meridian. - y will fall in the range (0, 1). y=0 is the top (north) edge of the map, at 85.0511 °N. y=1 is the bottom (south) edge of the map, at 85.0511 °S. y=0.5 is the middle, the equator. """ # This is based on instructions from the OpenStreetMap Wiki: # https://wiki.openstreetmap.org/wiki/Slippy_map_tilenames#Example:_Convert_a_GPS_coordinate_to_a_pixel_position_in_a_Web_Mercator_tile # (Retrieved 16 January 2025) # Convert the coordinate to the Web Mercator projection # (https://epsg.io/3857) # # x = longitude # y = arsinh(tan(latitude)) # x_webm = longitude y_webm = math.asinh(math.tan(math.radians(latitude))) # Transform the projected point onto the unit square # # x = 0.5 + x / 360 # y = 0.5 - y / 2π # x_unit = 0.5 + x_webm / 360 y_unit = 0.5 - y_webm / (2 * math.pi) return x_unit, y_unit Their documentation includes a worked example using the coordinates of the Hachiko Statue. We can run our code, and check we get the same results: >>> convert_gps_coordinates_to_unit_xy(latitude=35.6590699, longitude=139.7006793) (0.8880574425, 0.39385379958274735) Most users of OpenStreetMap tiles will use these unit positions to select the tiles they need, and then dowload those images – but we can also position these points directly on the global map. I wrote some more Pillow code that converts GPS coordinates to these unit positions, scales those unit positions to the size of the entire map, then draws a coloured circle at each point on the map. Here’s the code: from PIL import Image, ImageDraw gps_coordinates = [ # Hachiko Memorial Statue in Tokyo {"latitude": 35.6590699, "longitude": 139.7006793}, # Greyfriars Bobby in Edinburgh {"latitude": 55.9469224, "longitude": -3.1913043}, # Fido Statue in Tuscany {"latitude": 43.955101, "longitude": 11.388186}, ] im = Image.open("base_map.png") draw = ImageDraw.Draw(im) for coord in gps_coordinates: x, y = convert_gps_coordinates_to_unit_xy(**coord) radius = 32 draw.ellipse( [ x * im.width - radius, y * im.height - radius, x * im.width + radius, y * im.height + radius, ], fill="red", ) im.save("map_with_dots.png") and here’s the map it produces: The nice thing about writing this code in Pillow is that it’s a library I already know how to use, and so I can customise it if I need to. I can change the shape and colour of the points, or crop to specific regions, or add text to the image. I’m sure more sophisticated data visualisation libraries can do all this, and more – but I wouldn’t know how. The downside is that if I need more advanced features, I’ll have to write them myself. I’m okay with that – trading sophistication for simplicity. I didn’t need to learn a complex visualization library – I was able to write code I can read and understand. In a world full of AI-generating code, writing something I know I understand feels more important than ever. [If the formatting of this post looks odd in your feed reader, visit the original article]

yesterday 4 votes
Introducing the blogroll

This website has a new section: blogroll.opml! A blogroll is a list of blogs - a lightweight way of people recommending other people’s writing on the indieweb. What it includes The blogs that I included are just sampled from my many RSS subscriptions that I keep in my Feedbin reader. I’m subscribed to about 200 RSS feeds, the majority of which are dead or only publish once a year. I like that about blogs, that there’s no expectation of getting a post out every single day, like there is in more algorithmically-driven media. If someone who I interacted with on the internet years ago decides to restart their writing, that’s great! There’s no reason to prune all the quiet feeds. The picks are oriented toward what I’m into: niches, blogs that have a loose topic but don’t try to be general-interest, people with distinctive writing. If you import all of the feeds into your RSS reader, you’ll probably end up unsubscribing from some of them because some of the experimental electric guitar design or bonsai news is not what you’re into. Seems fine, or you’ll discover a new interest! How it works Ruben Schade figured out a brilliant way to show blogrolls and I copied him. Check out his post on styling OPML and RSS with XSLT to XHTML for how it works. My only additions to that scheme were making the blogroll page blend into the rest of the website by using an include tag with Jekyll to add the basic site skeleton, and adding a link with the download attribute to provide a simple way to download the OPML file. Oddly, if you try to save the OPML page using Save as… in Firefox, Firefox will save the transformed output via the XSLT, rather than the raw source code. XSLT is such an odd and rare part of the web ecosystem, I had to use it.

yesterday 2 votes