HNHacker News
TopNewBestAskShowJobs

kbrannigan

687 karma · joined October 27, 2016

submissionscomments
kbrannigan··on Ask HN: Remote island can only have 1 PC, 1 book, 1 programming language
But what if you were ask to at least 1 commit per day?
kbrannigan··on Muse Ember: Our smallest, most efficient model
This is starting to feel just like the early PC days. A bunch of Computer models coming out, all shapes and color, but eventually it settled around mouse keyboard, monitor

most people, use the office suite: Word, Spreadsheet , presentations, database.

I bet we're going to see the exact same thing with LLMs at specialized models instead of the everything models.

One for lawyers, coders, Accountants, Designers, etc…

kbrannigan··on Ask HN: What is the best non mainstream programming language or stack in 2026?
I tried rust but the compilation times often make it hard to get into a flow state. I have no idea why it takes so long to compile. I thought low level languages compiled faster.
kbrannigan··on Ask HN: Will future archeologist know about us, with everything encrypted?
So in a way carving things in stone or clay was probably the right idea. A silly thought, maybe we had computers before
kbrannigan··on Ask HN: Will future archeologist know about us, with everything encrypted?
So in a way, old inscriptions in clay tablets, carvings in Hieroglyphs , Nsibidi and other ancient languages are technically encrypted.

We needed to decrypt them first.

That actually makes a lot of sense. Thanks for that insight.

kbrannigan··on Police are using AI to pull over drivers for 'suspicious' behavior
What is a criminal? What is crime? In the USA, it was a crime for an enslaved Black American to learn to read and write? Attempting to escape was also a crime.

What about the crimes happening in company boardrooms rather than on the street corners?

People often call crime, individual offenses and think of certain racial groups, while ignoring real crimes committed by corporations led by other racial groups.

The crazy thing is that the greatest harms are caused by people who are not target of surveillance technology .

kbrannigan··on OpenCode 2.0
With websites i went back to 90s styles. Hyper dense in information. Anything clickable is a link. A button is for Write actions.

Interactive element are distinct from text.

If webpage page is an interactive canvas , make the canvas the main thing ad text as support.

If you start mixing the two idk..

kbrannigan··on A Modern Web App in C
Model are trained on them in 2026 .

Me Raw php .

Many people here have machines that can handle all of AOL , yahoo, msn of 2001 combined in their closet .

I have been going backwards in time as an experiment . I use mature extremely documented tools . Has to be at least 20 years old

kbrannigan··on Is mathematics over, or just graduating?
That is a very interesting perspective. What is the purpose of that corpus. Are all theorems or axioms useful?
kbrannigan··on Is mathematics over, or just graduating?
How big are mathematics? Is it as big as the universe?
kbrannigan··on Who is cleaning up all the garbage LLMs generate?
Hmm, hence the issue. It’s like chopping down a tree to make a box of matches. I wish the next generations of models don’t rely too much on markdown documentations which seems to be a human way of not writing code. Sometimes the code is just a little bit more structured to be honest , often times I feel like I might as well just write the code.
kbrannigan··on Who is cleaning up all the garbage LLMs generate?
What does that mean. Taking them down because the project failed or just replacing it?
kbrannigan··on Ask HN: What Makes Tokens Expensive?
It's not always larger models. Often time it's a smaller model fit for a specific use case. In the future we will see 16B models that are for : Legal, Healthcare, Science, Software Engineering, Education, Forestry, Music production
kbrannigan··on Astra and Fable still hack on simple variants of alignment evals from 2025
Such a massive leap at averaging possible use cases from previous data collected.

My guess is : collect all the prompt and their satisfaction score. group them by similarity . For each group pretrain the next model on that . Get these results ready.

Next model generation feed them back those answers.

kbrannigan··on Ask HN: Would you hire a new engineer, or get 300k worth of tokens for the team?
I really would like to be a fly on wall during these executive meetings? Here is what i realized.

Spaghetti code written by 5 developers will need 5 + 1 developers to manage.

Same in the AI age spaghetti systems built by agents, will need a greater agent to untangle the mess that itself has created

kbrannigan··on Ask HN: Initial Thoughts on GPT-6 Astra?
First impression. It seems smarter at synthesing information faster. Something Gemini was really good at but it does it better.

It speaks more naturally, unlike claude which litteraly just vomit jargon and random analogies.

It's very expensive. After 15 message I burned through my 5 hour limits.

Demos online are very biased toward fancy user interfaces and animations and video games

kbrannigan··on Check if a file was made with Claude
How long before they change the terms and conditions to subtly claim ownership of your files? When you write code they already insert Co author attribution/

Say I write a text by hand And then I tell it to clean up the grammar and fix some sentences Did it make it?

This is also interesting for those companies that siphoned the entire open web

kbrannigan··on Claude Fable 5.1 and Claude Mythos 5.1
The human brain is fascinating Three years ago The idea of having A robot writing production level code in 10 minutes that would have needed a team of 5 people and 2 months. Was pure Scifi

Now it's boring , not good enough

Wow there should be a term of that .

kbrannigan··on Is it just me, or has Claude Opus gotten worse recently?
When it's overly verbose like that. Does that count against your usage as well? Also during its thinking does it output hidden thinking tokens?

I am now wondering

kbrannigan··on The brain may be about to have its Ozempic moment
To get a job at our company take three pills of brainOzempic every day to maximize productivity.

We have our NeuroLinkBrain Implant plugin to AI agent , so that you can vibe code 24/7 . To maximize shareholder value.

kbrannigan··on Anthropic's best AI model struggles to attract users as cheaper tools thrive
I'm not trust I was meant to be sarcastic or if you're serious I cant tell Is it true?
kbrannigan··on Anthropic's best AI model struggles to attract users as cheaper tools thrive
We used to get paid to code now we think we need to pay a subscription fee just to write software. They push marketing campaigns saying Manual coding no more, just vibe code, gain 10x speed for $100.

Then they hit you with hourly and weekly limits, you are wondering when you are going to get cut off. Since LLM at probabilistic it often feels like pulling the lever of a slot machine, hoping our prompt is the jackpot. To make sure we win, we come up with systems, convoluted agents, context pipelines, rags to load . It feels like it's working, then bam you reach weekly limits.

(Just pay more if you want to keep winning).

I think developers need to wake up.

I myself started to use AI like a fancy debugger ,explainer. I make it walk me though every single line of code it writes.

I notice that i run into limits less, if i get cutoff, i can still make changes

kbrannigan··on Anthropic's best AI model struggles to attract users as cheaper tools thrive
But it worked and it was so useful that LLM companies siphoned their data. It was how i learned programming
kbrannigan··on Anthropic's best AI model struggles to attract users as cheaper tools thrive
remember 4 year ago we use to : have stack overflow open, documentation, obscure forums plus other tabs.

An ide open with 20 tabs open each file a component, a class or an interface We also use to hold entire codebases in our brain.

kbrannigan··on Anthropic's best AI model struggles to attract users as cheaper tools thrive
How many big tech companies let you talk to a human to get support. Automation is wonderful to cut cost for them but for the users being unable to get support is a horrible experience. But you cannot go elsewhere because they are the only player in town.

How can small companies with 1000x less money able to provide live support, but if you pay 20, 100, 200 dollars for a subscription you dont have a phone number to call ?

kbrannigan··on I ran Photoshop on a £0.60 computer chip
This is wonderful and amazing. Computing can be very simple. I think constraints and hardware restrictions can unlock creativity.

Imagine running an GPT-3 Generation model on a raspberry pi . This would be great

kbrannigan··on Felony charges for citizen deleting phone data at US Border
It's a very insidious, first you look all over your neighborhood flock cameras Hundreds of them came out of nowhere Nobody fought back .

it's as long as you're not doing anything wrong you don't have to worry about it Then once changed the definition of what wrong means.

If you're not a criminal you don't have to worry about it That's for your safety Then they changed the definition of what a crime is

kbrannigan··on Claude Code weekly limits reduce by a third tomorrow
That make sense. Let me ask you this When the project reaches a level of complexity . Do you simply reach for better models or do you reengineer it or does the project scope stop at the egdge of the model's capabilities.

I am asking because in my personal projects after a while they becomes a giant messy ball of wires and i basically trust the model to untangle it for me , by the time it untangles properly, I run into my token limits.

kbrannigan··on Claude Code weekly limits reduce by a third tomorrow
Any model currently in 2026 is better than the ones of 2024. What is the quality difference really.

I mean i think if a developer has a good handle of the code the difference is marginal .

Unless we 100% offload the thinking to the model and act like a prompt manager. Maybe

kbrannigan··on Ask HN: Do you feel comfortable admitting that you use AI?
If it was up to companies selling subscriptions. Nobody should even look at the code anymore. Just spend 200 bucks each month and you will be all set.

I admit i use AI for coding definitely too but after getting disillusioned with it.

I will definitely vibe code a prototype to quickly validate an idea. Once i see how it might work. I start from scratch and become more manual.

The more quality i want in a feature to be the more manually i get involved.

Page 1 of 8Next →