> ## Content Index > Fetch the complete content index at: https://labnotes.org/llms.txt > Use this file to discover other available public pages before exploring further. # Weekend Reading — FRAP aka as "the zoomies" - URL: https://labnotes.org/weekend-reading-frap-aka-as-the-zoomies/ - Published: 2026-08-29T15:54:36.000Z - Updated: 2026-08-29T15:54:36.000Z - Author: Assaf Arkin [**Clarissa**](https://gamedev.lgbt/users/ClarusPlusPlus/statuses/117159328858203808?ref=labnotes.org) "What if we kissed on the Cray supercomputer loveseat?" --- [**Vibe coding and quality**](https://ilyabirman.net/meanwhile/all/vibe-coding-and-quality/?ref=labnotes.org) "When you hear that some app was “vibe coded”, how much quality do you expect?" [I was just wondering](https://mas.to/users/assaf/statuses/117156843644522426?ref=labnotes.org) about this very topic. I think LLMs can do a remarkably good job at writing code, but "vibe coding" sounds like quality was neglected. [@bensheldon@ruby.social](https://ruby.social/@bensheldon?ref=labnotes.org) proposed [a wonderful idea](https://ruby.social/users/bensheldon/statuses/117156900374465826?ref=labnotes.org): "stewarded". I'm going to start saying the code was "**stewarded by me**". I think that gives people a better understanding that, while I didn't write every single line, I was quite specific in making sure the app meets my standards of quality. --- ## Tech Stuff [**Obsidian**](https://obsidian.md/?ref=labnotes.org) I came around to liking it and now I'm using it daily. I heard about Obsidian long ago, but somehow couldn’t get myself to enjoy it. I shifted around various apps: Apple Notes, Bear, Craft, Mem, etc. Eventually, I trimmed them all down to now using only Obsidian. I don't use the Links, Graph, or Canvas features (yet), but as a note taking app and useful reference tool, I keep it running at all times. Overall I would recommend. ⭐ ⭐ ⭐ ⭐ ⭐ ![](https://labnotes.org/content/images/2026/08/image-1787939131196.png) [**Reading notes on Naming Things by Tom Benner**](https://masalmon.eu/2026/08/27/naming-things-reading-notes/?ref=labnotes.org) > An understandable name has high comprehension (it can be understood quickly) and high recall (it can be remembered easily). [**omp — a coding agent with the IDE wired in**](https://omp.sh/docs/quickstart?ref=labnotes.org) Two weeks ago I upgraded from Pi to Oh My Pi (omp). And I'm loving it ❤️ Pi comes naked for you to install a variety of plugins. That's great in principle (and for people who love setting up their own Linux server), but in practice how do you which plugins to use? OMP comes with a lot of these capabilities built in, a better option for people who just want to get work done. Queue `/me`. Unlike Claude Code, it's not a flakey code abomination that fails in every other release. In short, OMP feels like the macOS of coding agents. ![](https://labnotes.org/content/images/2026/08/image-1787939866018.png) [**Verschlimmbesserung: The Word Your Software Updates Need**](https://geekyschmidt.com/post/2026-08-25-verschlimmbesserung/?ref=labnotes.org) > Verschlimmbesserung: An attempted improvement that only makes things worse. It is time we start naming this phenomenon in software engineering. [**Hister**](https://hister.org/?ref=labnotes.org) Turns the pages you visit and the files you keep into a private, full content search index. From the developer of [SearX](https://searx.space/?ref=labnotes.org), open-source, no mandatory cloud service, bundles web + CLI + MCP. ![](https://labnotes.org/content/images/2026/08/image-1787585306091.png) [**Punchcard**](https://everydaymactools.com/punchcard/?ref=labnotes.org) Menu bar app that tracks your day, printing a receipt of where you spent your time. No subscription, works offline, only tracks the name of the app you're using, not the window title. One time purchase of $9. ![](https://labnotes.org/content/images/2026/08/image-1787760947363.png) [**20 Lessons From 20+ Years in Tech**](https://newsletter.techworld-with-milan.com/p/20-lessons-from-20-years-in-tech?ref=labnotes.org) > This will be kind of a special issue where I share some important lessons from my 20 years of experience in the tech world. Actually, a bit more, as my father was a software developer his whole life, I know for myself that I was part of this world. > > During the first part of my life, until I was 14, I lost everything I owned twice. Then I spent more than 20 years building software for a living. These two things are connected more than it looks like. [**Prisma 8**](https://www.prisma.io/docs/prisma-postgres/quickstart/prisma-orm?ref=labnotes.org) This week I did a migration from Prisma 7 to Prisma 8\. It's … much different. I like Prisma 8, but don't go into this migration until you have an hour or two to spare. The API is all around different: smarter, but it takes some getting used to. Or, if you can, just ask the LLM to do it for you. ![](https://labnotes.org/content/images/2026/08/image-1787939206875.png) [**Obscura**](https://obscura.sh/?ref=labnotes.org) Tired of Playwright/Puppeteer eating up memory and taking forever to fire up? How about an open-source headless browser engine written in Rust? Runs JavaScript via V8, speaks the Chrome DevTools Protocol, and works as a drop-in replacement for headless Chrome. Eats up 30MB of memory (200+MB for headless Chrome), startup time is instant (2+ seconds for Chrome), page loads drops to 85ms from +500ms. Mind you there's no rendering engine, so you can't do visual regression testing, but for other purposes this is a great fit. ![](https://labnotes.org/content/images/2026/08/image-1787585501517.png) [**The Backlog Was a Coping Mechanism**](https://allen.hutchison.org/2026/08/25/the-backlog-was-a-coping-mechanism/?ref=labnotes.org) With AI, the entire model of software engineering chagnes: > The failure mode is just as old. Avery Pennarun makes the point plainly in his treatise on bug tracking and triage. Bugs pile up when filing outpaces fixing. Declaring bug bankruptcy doesn't change the slope of that line. Every engineer I know has worked in a tracker where the honest read on most open issues was that nobody would ever look at them again. We tolerated it because the alternative was forgetting, and forgetting felt worse. [**TermDOM**](https://termdom.org/?ref=labnotes.org) Build a TUI using the DOM: a JavaScript/TypeScript library that renders HTML and CSS to the terminal. ![](https://labnotes.org/content/images/2026/08/image-1787760975999.png) [**Condensing Twenty Years of Wisdom in One Markdown**](https://kig.re/2026/08/19/condensing-twenty-years-of-wisdom-in-one-markdown.html?ref=labnotes.org) That works for me: > If you cannot review the output, you have to constrain the input. That’s it. That’s the whole trick. You move your judgment *upstream*, out of code review and into the context the agent reads before it writes a single line. [**Graft**](https://nanonets.com/graft?ref=labnotes.org) Pre-built dependency map of your codebase, injected into your agent's prompt so it doesn't read 30 files to change 3\. Tree-sitter ASTs, no server/db/embeddings. Big win for big code bases. ![](https://labnotes.org/content/images/2026/08/image-1787590283978.png) [**Run OpenBSD on DigitalOcean for $4/month**](https://nil.wallyjones.com/run-openbsd-on-digitalocean-for-4month/?ref=labnotes.org) Now that servers are getting more expensive due to the RAMaggedon, I'm looking for cool tricks to keep server costs reasonable. [**flóra**](https://tech.lgbt/users/floofylily/statuses/117139654454035968?ref=labnotes.org) > going to the javascript store anyone want anything? ![](https://labnotes.org/content/images/2026/08/image-1787585063682.png) --- ## Eye for Design [**NebulaSans**](https://www.nebulasans.com/?ref=labnotes.org) Modern, humanist sans-serif with a neutral aesthetic for legibility in both digital and print applications. I'm using it in [Expense](https://expense.labnotes.org/?ref=labnotes.org) now, and I think it looks better — a touch more white space than the font I used before, so easier to scan. ![](https://labnotes.org/content/images/2026/08/image-1787939432575.png) [**Who can use this color combination?**](https://www.whocanuse.com/?ref=labnotes.org) Visceral way to see how different color combinations affect people with different visual impairments. (via [brad\_frost](https://mastodon.social/users/brad%5Ffrost/statuses/117168284883883267?ref=labnotes.org)) ![](https://labnotes.org/content/images/2026/08/image-1787939977568.png) [**Sometimes Your Best Customers Barely Use Your Product**](https://www.tariqdeir.com/best-customers-barely-use-your-product/?ref=labnotes.org) > If someone has to keep opening my product every day to get value from it, perhaps I haven't actually removed much work from their life? For some apps that is the correct shape, you want to create something sticky and engaging. [**Simon B. Støvring**](https://mastodon.social/users/simonbs/statuses/117157492035238346?ref=labnotes.org) > If hotel guests need a five-step manual to use your state-of-the-art shower, you’ve done something wrong ![](https://labnotes.org/content/images/2026/08/image-1787760285191.png) --- ## The One-Person App [**Software for One**](https://www.ajwaxman.com/writing/software-for-one?ref=labnotes.org) Another link to a post about a one-person app. Right now I'm working on a [one-person app](https://expense.labnotes.org/?ref=labnotes.org). I think that's the future we're heading towards, and key benefit of LLMs is that we can solve our own problems without needing a target audience: > Ephemeral is fine. We used the sleep app for about four months. Our son sleeps through the night now, so we retired it. If the app had taken me months to build, that might sting. Because it took a week, I'm just glad it worked when we needed it. ![](https://labnotes.org/content/images/2026/08/image-1787965970707.png) [**The Singular Viable Product**](https://www.nyn.me/blog/the-singular-viable-product?ref=labnotes.org) MVP -> SVP > The machine that produced all this had a name too: the Funnel. You know it well. Start wide. Pick a large cohort because the tax demands one. Guess at its center. Ship the mean. Then spend years trimming and growing: cut the features the middle ignores, add the ones it begs for, and sculpt your way toward that mythical state of grace, product-market fit. > > Wide, then narrow. Cohort, then fit. Mean, then meaning. > > None of this was wrong. It was the only rational response to the economics. When building costs quarters and millions, you cannot afford to be wrong about the audience, so you aim for the fattest part of the distribution and pray. [**Anger, Anxiety and Agency**](https://lucumr.pocoo.org/2026/8/24/anger-anxiety-agency/?ref=labnotes.org) Stay curious my friend … > A lot of the gains from AI aren’t turning into productivity gains that are reflected in company profits but they’re showing up instead in the number of side projects shipped by everybody not on their company’s time. [**New app: Yesterday, an iPhone app for yesterday’s weather**](https://interconnected.org/home/2026/08/26/yesterday?ref=labnotes.org) Another helpful one-person app: > Weather! It's like yesterday but cooler.. warmer... it's going to rain.. that's all you need. So that's all Yesterday shows: how today's weather compares to yesterday, so you know when to grab a jacket. Plus a glanceable description too. ![](https://labnotes.org/content/images/2026/08/image-1787966131763.png) --- ## Peoples [**Why Is Everyone in Silicon Valley Talking Like That?**](https://archive.ph/NTb3Y?ref=labnotes.org) Silicon Valley is vibing to a new slang: > If you misspeak, maybe you’re “hallucinating.” Don’t know the answer to a question? That’s because it’s not in your “training data.” Feeling forgetful? Perhaps you have a case of “context rot,” a phrase that refers to the degradation of a bot’s responses over a long conversation. [**Three People Is a Team. Seven Is a Committee with Zero Decisions.**](https://www.v01.io/posts/2026/08/three-people-is-a-team/?ref=labnotes.org) > Including more people feels inclusive. It feels democratic. But it slows everything down. And worse, it diffuses ownership. When everyone is responsible, nobody is responsible. ![](https://labnotes.org/content/images/2026/08/image-1787585965384.png) --- ## Business Side [**Building without predicting**](https://sive.rs/fit?ref=labnotes.org) The desired paths will always reveal themselves: > A fabled example: They built a new park and argued about where to put the walkways. Around the edges? Through the middle? The winning answer: Open the park with no walkways. After a year, look where people have been walking. Find paths where the grass is worn away. Pave there. Don’t predict. [**Your coding agent got the onboarding your developers never did**](https://thenewstack.io/coding-agents-developer-neglect/?ref=labnotes.org) > But isn’t anyone else feeling uncomfortable with the willingness to give all these benefits to an agent when we wouldn’t do it for people? Where was all that budget, executive sponsorship, and the patience to build these capabilities when people were writing all the code? Organizational leaders sure are eager to please the machines, where they were reluctant to solve these problems for developers, and that, for me, is a telling signal. [**When life calls for causal analysis**](https://www.junkcharts.com/when-life-calls-for-causal-analysis/?ref=labnotes.org) > Causal inference is, in my view, an important area of business concern. Frequently, business metrics move in unexpected directions. Business managers want to take action to reverse the undesirable trends. Such action works only if the cause-effect relationship has been correctly diagnosed. People already do causal reasoning in simple ways; they can benefit from more sophisticated analytical machinery. --- ## Machine Intelligence [**Stop Trying to Distinguish AI-Generated Writing From Human Writing Because You Can’t**](https://journal.james-zhan.com/ai-writing/?ref=labnotes.org) My favorite is when people accuse me of using AI too much in my writing. Yes, my wriing would look like AI, because I'm following the most basic writing principles, which is exactly what an AI is programmed to do. Down to me using the em-dashe liberally: > If you’re writing something that’s meant to be read, to communicate, then just write it in a way that’s easy to understand; don’t intentionally obfuscate it. You’re not more educated, literate or smarter just because you use vocabulary that no one normally uses. [**Readers can't identify watermarked AI text**](https://www.seangoedecke.com/readers-cant-identify-watermarked-ai-text/?ref=labnotes.org) Related, people can't identify the presence of AI watermarks. Obviously this wasn't exactly a scientific study, but it's still pretty suggestive. If watermarks were really choosing random words that the model would never pick, you'd be able to sometimes tell from three side-by-side responses which one went down the weird watermarked road, right? [**Is the LLM smart or are you not?**](https://blog.troed.se/posts/the-anchoring-effect-and-cognitive-simplicity/?ref=labnotes.org) > The reality is that there are many such well-known phenomena in psychology and neuropsychology where our brains act much more “simple” and predictable than we would like to believe. We like to think of our consciousness as this impenetrable fortress of complex reasoning, but much of it operates on heuristics and biases that can be bypassed with the right “trick”. [**kottke.org**](https://mastodon.social/users/kottke/statuses/117151078952883396?ref=labnotes.org) > [Meat proxy](https://sfisms.org/meat-proxy?ref=labnotes.org) (noun): “A person who forwards AI-generated text, code, or other output without reading, understanding, or validating it. The person acts only as a relay between the AI system and the intended recipient.” --- ## Insecurity [**Just the rumour of a bug is enough to find an exploit these days**](https://anil.recoil.org/notes/rumour-is-the-exploit?ref=labnotes.org) Our security processes is in for a rude awakening: > Conventional security process involves [embargoing](https://www.redhat.com/en/blog/Understanding-security-embargoes-at-Red-Hat?ref=labnotes.org) the bug, and assumes that secrecy of the details protects users. However, all an agent needs today is a broad direction to search in, and it can do its own research. [Fang et al.](https://arxiv.org/abs/2404.08144?ref=labnotes.org) found that when given a CVE description, their GPT-4 agent [exploited 87%](https://surrealyz.github.io/classes/llmsec-fall24/slides/14-agents-exploit-vulnerabilities.pdf?ref=labnotes.org) of a 15-vulnerability benchmark, and without the description, just 7%. ![](https://labnotes.org/content/images/2026/08/image-1787966094778.png) [**Logging In Was Never Supposed to Be This Complicated**](https://www.theatlantic.com/technology/2026/08/password-manager-login-difficulty/688396/?ref=labnotes.org) In the continous race between security and usability, security didn't come on top, and that's because bad usability leads to bad security, but definitely usabilty lost: > Cybersecurity experts will tell you that passkeys—which typically entail logging in to apps or websites with a face or fingerprint scan instead of a password—are the future. When they work properly, they’re both painless and secure. … > > But that future isn’t here yet, and in the present, passkeys have added a new layer of frustration. Part of the problem is that virtually no one seems to know what they are. [**Cyber Vulnerabilities**](https://epoch.ai/data/cve?ref=labnotes.org) At some point Claude Mythos gets released and then … ![](https://labnotes.org/content/images/2026/08/image-1787760028356.png) --- ## Everything Else [**rainbowCrow**](https://infosec.exchange/users/cR0w/statuses/117162309857118624?ref=labnotes.org) "administrative controls in a nutshell" ![](https://labnotes.org/content/images/2026/08/image-1787938983833.png) [**Jan Wildeboer**](https://social.wildeboer.net/users/jwildeboer/statuses/117149924798309686?ref=labnotes.org) > From "640 KB ought to be enough for everyone" to "640 KB is the maximum you can afford". #ThanksAI :) [**Hacker News**](https://news.ycombinator.com/item?id=49459227&ref=labnotes.org) CEO jobs are safe until the humanoid robots have mastered this skill: > Based on my observations, the most important characteristics of a CEO is to be over 6ft, and to walk about looking important. **[Mark T. Tomczak](https://mastodon.fixermark.com/users/mark/statuses/117151498580102151?ref=labnotes.org)** > Thinking about the era of technology where the easiest way to display images was to take all the air out of a tube and point a death ray at your face. > > It's fine though. There was a thin screen between you and the death ray that caught all the death and turned it into pretty colors. [**Anne Fuller**](https://alaskan.social/users/ScoterD/statuses/116853327713773067?ref=labnotes.org) "Grateful for TomGauld, an erudite cartoonist (and for the IRL scientist who share this with me)" ![](https://labnotes.org/content/images/2026/08/image-1787939738244.png) [**Ben Lockwood, Ph.D.**](https://substack.com/@briefecology/note/c-285407221?ref=labnotes.org) > The useless loop of students using AI to write papers that professors use AI to grade is the ultimate endpoint of treating education as a product rather than a process. [**Frenetic random activity periods**](https://en.wikipedia.org/wiki/Frenetic%5Frandom%5Factivity%5Fperiods?ref=labnotes.org) FRAP aka as "the zoomies" [**B.C. swimmer crosses the Strait of Georgia in under nine hours**](https://www.cbc.ca/radio/asithappens/b-c-swimmer-crosses-strait-of-georgia-under-nine-hours-9.7316260?ref=labnotes.org) This interview is just gold: (via [masukomi)](https://connectified.com/users/masukomi/statuses/117150332573835763?ref=labnotes.org) > **How long do you think it's going to take you to fully recover from something this intense?** > > I don't think I've recovered from childhood, so probably never, but I could get back in the pool in a day or two if I was keen on swimming, which incidentally, as it turns out, I don't think I am right now. [**afterglow**](https://morphing.cloud/afterglow/?ref=labnotes.org) Run classic *After Dark* screen savers on modern macOS! (via [Jonathan Clede)](https://theforkiverse.com/ap/users/116455784965047262/statuses/117158085206484255?ref=labnotes.org) ![](https://labnotes.org/content/images/2026/08/image-1787760439981.png) [**Columbia House: The First 12 Cost A Penny, But …**](https://tedium.co/2026/08/24/columbia-house-history-retrospective/?ref=labnotes.org) > Every SaaS model that you’ve ever hated—whether the sketchy cancel strategies of Adobe or the Bending Spoonsification of every mature SaaS app on the internet—has roots in the Columbia House model. That model offered a very attractive starting point, and likely proved a starter for numerous CD and cassette collections nationwide. But it normalized aggressive, misleading behavior by large, mainstream companies. [**‘Show How 3M Is 0% at Fault:’ Expert Witness Used ChatGPT to Write Report Defending Company in Deadly Explosion Lawsuit**](https://www.404media.co/show-how-3m-is-0-at-fault-expert-witness-used-chatgpt-to-write-report-defending-company-in-deadly-explosion-lawsuit/?ref=labnotes.org) The case shows that the specific prompts used to create this type of expert testimony can be discoverable during a case, and that those prompts can be quite embarrassing. [**There Are Carrier Pigeon Apps Now Because Pigeon Technology Is Actually Superior**](https://gizmodo.com/there-are-carrier-pigeon-apps-now-because-pigeon-technology-is-actually-superior-2000801934?ref=labnotes.org) We're entering the era of Friction Maxxing. ![](https://labnotes.org/content/images/2026/08/image-1787760091949.png) [**Good writing is obvious, not original**](https://www.seangoedecke.com/good-writing-is-obvious-not-original/?ref=labnotes.org) > When you write, you should try to say things that are obviously true, and spend very little time worrying about whether you’re being original. Ironically, this is the best way to do truly original writing. [**Welcome To The Resistance: Meet The Workers Dodging (And Sabotaging) Their Employer's AI Mandates**](https://aftermath.site/ai-resistance-tips-workforce-llm-workers/?ref=labnotes.org) > "I work for a company that produces luxury items", one worker tells me. "Any time I get prompted that my 'token spend' is too low (0), I go in and ask ChatGPT random gardening questions. So far it has failed to adequately build a rain garden, but I’ve gotten it to describe a sweet potato once." [**Walmart's AI Won't Let You Buy That Extra Gallon of Milk, But It Might Make You Spend 40% More**](https://www.ndtvprofit.com/technology/walmarts-ai-wont-let-you-buy-that-extra-gallon-of-milk-but-it-might-make-you-spend-40-percent-higher-11947667?ref=labnotes.org) The pinnacle of innovation: using AI to guide humans through the process of relocating money from their own bank account to the merchant's bank account: > "Someone recently shared with me that they asked Sparky for a weekly meal plan of healthy foods with high protein options. Within a few seconds, Sparky shared recipes and meal kits with the ability to add all the ingredients they needed their basket with one click. Sparky even recognised the ingredients they had recently purchased, both online and store, so they didn't buy something they already had. It's building trust," Furner narrated in the post-earnings conference call. [**stux⚡️**](https://mstdn.social/users/stux/statuses/117150711246339287?ref=labnotes.org) “That’ll be 10 billion please” ![](https://labnotes.org/content/images/2026/08/image-1787590167695.png)