880 karma · joined May 1, 2021
Chat is such a “leaky” abstraction for LLMs
I think most people share the same negative experience as they only interact with LLMs through the chat UI by OpenAI and Anthropic. The real magic moment for me was still the autocompletion moment from the gh copilot.
1) they are too hard to type.
2) using them without surrounding thin space or hairspace breaks the horizontal rhythm and draws unnecessary attention to the punctuation; but thin and hair spaces are equally hard to type
3) Most people write markdown with mono space fonts, making these dashes and spaces indistinguishable.
I'm sensing another bitter lesson coming, where domain optimized AI will hold a short term advantage but will be outdated quickly as the frontier model advances.
I don't disagree that software engineering is not as rigorous as it was. But software has also spread into much wider area, allowing many more people to participate as programmers, all thanks to abstractions.
I'm drawn to the romantic/pure/spartan aspect of low level programming too but I also need to make a living by solving problems in practical and scrappy ways.
If the AI is so good at it, why are we still hiring human to do the job? It just shows how the interview process is not measuring the right thing to start with.
Despite fewer clicks, copyright fights, and sometimes iffy answers, AI could unlock new ways to summon all the world’s knowledge.
Because Clojure treats data as first-class citizens, we could build our own lightweight conflict resolution system using pure functions that operate on these transactions.
A sound effect would be nice. Just an idea.
UW has missed a great opportunity to show its current and future students and faculty that innovation is valued there. Their management decision will shutdown the minds of so many brilliant hackers, as we have already witnessed with the OP's decision to back down. I really hope the school can re-evaluate their decision and more importantly, the faculty should support the students, as they are together pursuing the goal of learning by doing. That's what higher education should offer, not conditioning people into rule makers and rule followers.
Sadly privacy is not a marketable feature, or at least it does not have the ROI as Apple originally believed. I feel the only way to level the play field is to reconsider our regulation framework and treat privacy as a fundamental benefit for consumers.
I believe the PWAs will correctly show up in the OS native app launchers.
What is ironic though, is that the morphing algorithms are similar to what modern SPA frameworks do, except that it can be addressed without introducing new concepts such as vdom, zones, signal etc.
I agree that the island architecture is limiting when it comes to large scale application with state shared across multiple areas in the UI. I'd be curious to hear others finding success in building apps.
p.s. I appreciate the smol island :)
Claude, v0, etc are incredible- but you can't install packages, run backends or edit code. That’s where Bolt.new stands out:
1. Full-Stack in the Browser: Bolt.new integrates cutting-edge AI models with an in-browser development environment powered by StackBlitz’s WebContainers
2. AI with Environment Control: Unlike traditional dev environments where the AI can only assist in code generation, Bolt.new gives AI models complete control over the entire environment including the filesystem, node server, package manager, terminal, and browser console. This empowers AI agents to handle the entire app lifecycle—from creation to deployment.
Inheritance can be useful when it captures a single shared aspect across multiple designs. It becomes unwieldy when it captures multiple shared aspects where each aspect may need to change independently.