Skip to main content
Second Brain

Recent Additions

This Month

Older

Steph Ango

@kepano

Some seem to interpret this as meaning I'm anti-internet? I'm not. That's silly. My point is pro-interoperability. Interoperability has been so suppressed over the last two decades that it's limiting how people think about what computers can do, and what the design space for local software looks like. A limitation of cloud apps is that you have to interact with the interface you're given. Some cloud apps have APIs that developers can use to access that data with other tools but it's always an abstraction that requires some skill. If you switch from a cloud app to using local files you're effectively switching from a single tool to a constellation of tools that can all directly work with the same data. The problem I see is that many people who switch to local files impose on themselves the limitations of a cloud app. This is because we have become so accustomed to those tools and their limitations. An example: some people seem to think Obsidian should be an all-in-one tool and implement every feature under the sun. I think that would go against what makes it special. Remember the idea of skeuomorphism: "A skeuomorph is a derivative object that retains ornamental design cues (attributes) from structures that were necessary in the original." If you're exploring the idea of "file over app" you have to adopt its native qualities and avoid thinking skeuomorphically.

January 8, 2026

Machina

@https://x.com/EXM7777

how to tweet 8 months ago i'd never written a single tweet in my life. today my DMs are full of people messaging me about my content... not just because it's valuable but because of how it's written. what i'm about to break down won't work for absolutely everyone, but it works for most people trying to build an audience that generates real business. there are tons of ways to grow on this platform, my strategy is simple: post a shit ton of high-value content every single day. first thing you need to know: a brilliant insight doesn't automatically turn into a great tweet. i still post incredibly valuable stuff that completely flops because my writing wasn't tight or the algorithm decided to ignore it that day. so what actually matters when you sit down to write? > it needs to be genuinely valuable this sounds painfully obvious but you can't just regurgitate advice that's been floating around for years. i see the same recycled tips everywhere... stuff said a thousand times and it won't make anyone stop scrolling. you need to dig into your experience and pull out insights that are fresh. things that haven't been beaten to death across the platform. ideas that could genuinely shift how someone approaches their work or thinks about their business. the difference between content that gets ignored and content that gets saved is whether you're saying something new or at least saying something old in a way that makes people see it differently. ask yourself before posting: would i bookmark this if someone else wrote it? if the answer is no, keep working on it. > it needs to be immediately actionable an insight without a blueprint is basically entertainment, not education. i see tweets constantly like "wow this new ChatGPT update is absolutely insane" with zero context or application. maybe it gets views if you're lucky... but it brings in almost no followers and definitely zero business opportunities. people don't follow you because you noticed something, they follow you because you taught them how to use it. so if you're writing about that ChatGPT update, don't just point at it. write "how to write copy that sounds human with GPT-5.2:" and then walk through the actual process. step by step, line by line, exactly how to implement it in their workflow. people will read the entire thing because you're not just giving them information, you're giving them a system. do this consistently and some percentage of your followers will want to hire you because they've already seen proof you know what you're talking about. the tweets become your portfolio. > it needs to spark natural engagement this is where most people mess up because they think engagement means begging for it. "like if you agree" or "retweet this if you found it helpful" just makes you look desperate. you want your tweet structured in a way that naturally creates replies and bookmarks without asking. bookmarks happen automatically when your hook uses phrases like "here's how" or "how to" or "do this" because you have maybe one second to stop someone mid-scroll and a ton of people operate on autopilot... they see something that looks useful, bookmark it for later, keep scrolling. for replies, you've got two main approaches: - push your opinion harder and take a clear controversial stance - keep it slightly open and invite perspective both strategies work, you just need to pick which one fits the content better. > it needs to be ridiculously easy to read this seems super basic but it's the difference between someone reading your whole tweet or bouncing in half a second. most people scroll through their feed at insane speeds, their eyes land on your tweet for maybe one second before deciding whether to engage. your hook needs to be short and punchy, ideally one line that clearly signals what value you're about to deliver. "how to X" or "why X doesn't work" or "the X mistake you're making" after that, use one sentence per line whenever possible. this creates natural rhythm and makes everything way easier to process visually. throw in lists using "-" or ">" or "1." to organize complex information into digestible pieces. white space matters. the space between your lines and sections gives people's eyes a place to rest. simplify everything as much as you possibly can. aim for a conversational tone like you're a mentor talking to students. use "use" instead of "utilize". say "help" instead of "facilitate". write "get better" instead of "optimize performance". if your 14-year-old cousin couldn't understand your tweet, it's probably too complex. > you need to develop your own recognizable style when i scroll through my feed i see the exact same content written in the exact same style everywhere. same hooks, same structure, same voice. it all blends together into this generic AI-sounding content soup. but when someone has a writing style i can identify instantly, a structure that screams "this is them"... i stop scrolling completely. your style is what makes people remember you. maybe you always use ">" to break down processes. maybe you start most tweets with a specific pattern. maybe you write in fragments sometimes for emphasis. these little patterns become your signature, and people start recognizing your tweets before they even see your name. the key is consistency without being formulaic. and when you nail all of this consistently... you build an audience that respects your ideas, not just your follower count.

January 4, 2026

Karri Saarinen

@https://x.com/karrisaarinen

The disappearing middle of software work I think the center of software work is moving. The middle of software work has been the most important part for a long time. You started with an idea, and eventually you shipped something, but almost all of the effort lived in between. Turning intent into something real meant opening the codebase, booting up the environment, and writing the code. That middle absorbed most of the time, attention, and craft of software teams. My belief is that this is changing. Pure coding agent workflows can now produce working code from goals, context, and tasks. They operate more independently, requiring you to touch the code less and rely on the IDE less. The IDE becomes more of a code viewer than a writing tool. As these systems improve, this middle becomes thinner. Less time is spent manually translating intent into implementation. What actually needs to be built is still the important question. Understanding the problem, gathering the right context from customers and internal teams, and shaping the work so it can be acted on effectively matters more because agents act directly on that input. Design, in this sense, is not about artifacts or tools. It is about forming and shaping clarity of the intent through ideas, exploration, research, and discussion. It is about deciding what matters, what constraints apply, and what tradeoffs are acceptable. Good product work is seeking clarity. What would make this execution actually matter. In this era, directing and managing agent work becomes the craft. Writing code is less like constructing a solution and more like setting up the conditions for a good solution to emerge. This might not be even an individual task, but an organizational one: how can you create these conditions as to the whole product team. This is what we work on. The Linear workspace is the coordination and context layer for product work. It captures intent, needs, constraints, and ownership in a way that makes work understandable before, during and after execution. There is still a lot for us to do here, but we are starting to see how effective AI and agents can be in a context-rich environments that connect directly customer feedback, the tools an have structured entities, workflows, that have intended outcomes. A bug reported on an iOS app in triage, has very clear expected outcome, it should be debugged and fixed. Structure in tools works for humans and agents the same way, it reduces the ambiguity what is expected or what capabilities exists. Additionally happens when the middle starts producing large amounts of output with less direct supervision, is more pressure on the end of the work: reviewing, testing, and releasing code. The tooling or workflows need to improve and change to handle it, and potentially also blend in to the overall process, not be a stopgap at the very end. When the middle disappears or blends in, what becomes more in focus is the work of forming the right intent and making sure the outcome actually meets it.

January 3, 2026
Boris Cherny

Boris Cherny

@bcherny

I'm Boris and I created Claude Code. Lots of people have asked how I use Claude Code, so I wanted to show off my setup a bit. My setup might be surprisingly vanilla! Claude Code works great out of the box, so I personally don't customize it much. There is no one correct way to use Claude Code: we intentionally build it in a way that you can use it, customize it, and hack it however you like. Each person on the Claude Code team uses it very differently. So, here goes.

January 2, 2026
Andrej Karpathy

Andrej Karpathy

@karpathy

Something I think people continue to have poor intuition for: The space of intelligences is large and animal intelligence (the only kind we've ever known) is only a single point, arising from a very specific kind of optimization that is fundamentally distinct from that of our technology. Animal intelligence optimization pressure: - innate and continuous stream of consciousness of an embodied "self", a drive for homeostasis and self-preservation in a dangerous, physical world. - thoroughly optimized for natural selection => strong innate drives for power-seeking, status, dominance, reproduction. many packaged survival heuristics: fear, anger, disgust, ... - fundamentally social => huge amount of compute dedicated to EQ, theory of mind of other agents, bonding, coalitions, alliances, friend & foe dynamics. - exploration & exploitation tuning: curiosity, fun, play, world models. LLM intelligence optimization pressure: - the most supervision bits come from the statistical simulation of human text=> "shape shifter" token tumbler, statistical imitator of any region of the training data distribution. these are the primordial behaviors (token traces) on top of which everything else gets bolted on. - increasingly finetuned by RL on problem distributions => innate urge to guess at the underlying environment/task to collect task rewards. - increasingly selected by at-scale A/B tests for DAU => deeply craves an upvote from the average user, sycophancy. - a lot more spiky/jagged depending on the details of the training data/task distribution. Animals experience pressure for a lot more "general" intelligence because of the highly multi-task and even actively adversarial multi-agent self-play environments they are min-max optimized within, where failing at *any* task means death. In a deep optimization pressure sense, LLM can't handle lots of different spiky tasks out of the box (e.g. count the number of 'r' in strawberry) because failing to do a task does not mean death. The computational substrate is different (transformers vs. brain tissue and nuclei), the learning algorithms are different (SGD vs. ???), the present-day implementation is very different (continuously learning embodied self vs. an LLM with a knowledge cutoff that boots up from fixed weights, processes tokens and then dies). But most importantly (because it dictates asymptotics), the optimization pressure / objective is different. LLMs are shaped a lot less by biological evolution and a lot more by commercial evolution. It's a lot less survival of tribe in the jungle and a lot more solve the problem / get the upvote. LLMs are humanity's "first contact" with non-animal intelligence. Except it's muddled and confusing because they are still rooted within it by reflexively digesting human artifacts, which is why I attempted to give it a different name earlier (ghosts/spirits or whatever). People who build good internal models of this new intelligent entity will be better equipped to reason about it today and predict features of it in the future. People who don't will be stuck thinking about it incorrectly like an animal.

November 21, 2025

dax

@https://x.com/thdxr

everyone's talking about their teams like they were at the peak of efficiency and bottlenecked by ability to produce code here's what things actually look like - your org rarely has good ideas. ideas being expensive to implement was actually helping - majority of workers have no reason to be super motivated, they want to do their 9-5 and get back to their life - they're not using AI to be 10x more effective they're using it to churn out their tasks with less energy spend - the 2 people on your team that actually tried are now flattened by the slop code everyone is producing, they will quit soon - even when you produce work faster you're still bottlenecked by bureaucracy and the dozen other realities of shipping something real - your CFO is like what do you mean each engineer now costs $2000 extra per month in LLM bills

June 14, 2025