Labnotes - Weekend Reading — Ctrl click back

Tech Stuff

PartyKit This looks interesting. Similar concept to Vercel/Netlify — free model with higher priced tiers — but for building real-time stateful server apps. Uses Cloudflare workerd runtime, Durable Objects to maintain state, with support for HTTPS and WebSocket, and some other cool tricks.

brillout/awesome-react-components What makes this list better than other awesome lists: to add an entry you also have to remove one:

We keep this list fresh by requiring all PRs to remove one or more non-awesome entries from this list.

How Next.js breaks React Fundamentals (Examples)

I truly believe Colocation, Encapsulation, and Arbitrary composition are crucial properties for efficient development. They are necessary for team cooperation and maintaining mid-large codebases. … That's the reason I'm writing this. So we can design better framework APIs.

Unclack Small macOS utility that mutes your keyboard while you type!

TDD: You're Probably Doing It Just Fine 👏

• You can write good code without TDD.
• Even the act of trying to write one test is already enough to clarify your thinking about what the system (or some part of it) ought to do.
• You don’t have to do TDD 100% all the time perfectly in order to benefit from doing it some of the time and pretty well.
• You might find tests helpful when it comes time to change the code after you’ve written it the first time.

What You Need to Know about Modern CSS (Spring 2024 Edition) Covers new additions to CSS, like container queries, :has(), view transitions, nesting, and more.

Security Writer 🤯 Ctrl click back!

Sharing a really simple productivity/workflow tip that just blew my colleague’s mind.

Say you click through a link and it takes you there, you want to remain on that page but also want your previous page back, just ctrl click the back button and it’ll open your previous page with all related history in the new tab, and not switch to it.

Can’t tell you how often I use this, especially for research sessions.

WeExpire

WeExpire is an opensource tool for creating emergency notes that can be read by your trusted contacts only after your death or if you are seriously injured.

CatSalad “Tabs vs Spaces”

Eye for Design

Design Spells Micro-interactions, easter eggs, and other seemingly extra design details that infuse life, personality, and fun back into the web.

Social Proof Learn from the best examples of social proof in the industry and how to apply it to increase the perceived value of your product ethically.

Blair Holloway

Why we crave healthier computing How throughout history, increasing computing capability has lead to increasing burden on users.

Rob Norris “Raise your hand if you ever made a paper snake out of tear-off tractor-feed margins.”

Peoples

Achieving Greatness Without Falling Apart

Psychologists Ellen Winner pointed this out decades ago when she studied prodigies across domains. She found that they were almost always borderline obsessive. They had what she called a “rage to master.” But what really mattered is what that drive was built upon. …

Business Side

The Pentagon’s Silicon Valley Problem Israel bet big on AI and … well … it didn’t pan out: “AI, it turned out, knew everything about the terrorist except what he was thinking.” That’s the story of October 7, but this article also talks about the US DoD, Google, Palantir, the Vietnam war, and several other casualties of the “tech first” way of thinking.

Fake Scrum Stats Memes & Humor

Machine Intelligence

Jan Run LLMs locally on your computer — a privacy-focus, local-first, open-source solution. I’ve been using it for several days now, and it’s pretty nice as a second-pair LLM that doesn’t share your secrets with the world. I’m using Mistral 7B instruct, which you can download yourself, or choose any model you like. Oh, and it also supports remote APIs — OpenAI GPT 3.5 and GPT 4.0.

Quiet-STaR 'Think before you speak' applied to AI:

We present Quiet-STaR, a generalization of STaR in which LMs learn to generate rationales at each token to explain future text, improving their predictions. We address key challenges, including 1) the computational cost of generating continuations, 2) the fact that the LM does not initially know how to generate or use internal thoughts, and 3) the need to predict beyond individual next tokens.

NYC’s AI Chatbot Tells Businesses to Break the Law Tired: open source license. Wired: open lies license. “The Microsoft-powered bot says bosses can take workers’ tips and that landlords can discriminate based on source of income”

Why great AI produces lazy humans I trust the numbers in this article:

Recruiters with higher-quality AI were worse than recruiters with lower-quality AI. They spent less time and effort on each résumé, and blindly followed the AI recommendations. They also did not improve over time. On the other hand, recruiters with lower-quality AI were more alert, more critical, and more independent.

How Quickly Do Large Language Models Learn Unexpected Skills?

They chose new metrics showing that as parameters increased, the LLMs predicted an increasingly correct sequence of digits in addition problems. This suggests that the ability to add isn’t emergent — meaning that it undergoes a sudden, unpredictable jump — but gradual and predictable. They find that with a different measuring stick, emergence vanishes.

Someone Got GPT2 Running Entirely in a Spreadsheet Add to the list of things Excel can do!

Insecurity

Technologist vs spy: the xz backdoor debate A rundown of the most insidious backdoor that was just discovered by an industrious postgres developer:

Well — we just witnessed one of the most daring infosec capers of my career. Here’s what we know so far: some time ago, an unknown party evidently noticed that liblzma (aka xz) — a relatively obscure open-source compression library — was a dependency of

Facebook snooped on users' Snapchat traffic in secret project, documents reveal

A secret program called "Project Ghostbusters" saw Facebook devise a way to intercept and decrypt the encrypted network traffic of Snapchat users to study their behavior.

taber

The reason unsubscribing from a list may "take 7-10 business days to process" is the systems that cause those emails to be barfed out are just batch jobs somebody slapped together in 2002 where one day's batch is "get all the emails for the week from the Big Database" and then schedules the "Actually Send The Emails" batch jobs without checking the Big Database again

Everything Else

GeriAQuin “coffee for those that don’t have #peanuts allergies”

Col

To save money on storage, buy a pen from Amazon and use the box it came in to store blankets and sheets.

UnclePJ

I read somewhere that Fitbits are just modern day Tamagotchis and the stupid animal you’re trying to keep healthy is just you. I think about that a lot.

mos_8502

A ship crashed into a Baltimore bridge and demolished the lies about immigration “Despite the right's ridiculous DEI conspiracy theories, the Baltimore bridge disaster reminded us that immigrants are what makes America great.” All of the men confirmed/presumed dead in this tragedy have emigrated from Honduras, Guatemala, El Salvador, and Mexico.

Why isn’t dental health considered primary medical care?

Ailments of the mouth can put the body at risk for a slew of other ills. Some practitioners think dentistry should no longer be siloed.

Only 90’s Cats Will Understand