Hacker News Reader

Nice to see a Festool miter saw in his shop, Lynch knew what he was doing.

a-r-ton Amelia Earhart's Reckless Final Flights

There is a good Veritasium episode on her last flight going deep into technical details of what went wrong: https://www.youtube.com/watch?v=zTDFhWWPZ4Q

a-r-ton AI code is legacy code from day one

If we are speculating here, why not just go straight to an LLM serving all requests directly? No code needed.

a-r-ton How I built an AI-generated news site with daily articles

https://en.wikipedia.org/wiki/Dead_Internet_theory

a-r-ton OpenAI Audio Models

Glad to hear it's on your radar. I'd imagine phone call transcription is a significant use case.

a-r-ton OpenAI Audio Models

There are two options that I know of:

1. Merge both channels into one (this is what Whisper does with dual-channel recordings), then map transcription timestamps back to the original channels. This works only when speakers don't talk over each other, which is often not the case.

2. Transcribe each channel separately, then merge the transcripts. This preserves perfect channel identification but removes valuable conversational context (e.g., Speaker A asks a question, Speaker B answers incomprehensively) that helps model's accuracy.

So yes, there are two technically trivial solutions, but you either get somewhat inaccurate channel identification or degraded transcription quality. A better solution would be a model trained to accept an additional token indicating the channel ID, preserving it in the output while benefiting from the context of both channels.

a-r-ton OpenAI Audio Models

Transcribing Twilio's dual-channel recordings using OpenAI's speech-to-text while preserving channel identification.

a-r-ton OpenAI Audio Models

Hi Jeff, are there any plans to support dual-channel audio recordings (e.g., Twilio phone call audio) for speech-to-text models? Currently, we have to either process each channel separately and lose conversational context, or merge channels and lose speaker identification.

a-r-ton Ask HN: Why aren't LLMs used for email spam detection?

The MS announcement is limited to scams/phishing. Google mentions both scams and spam, but somehow I still get 15-20 spam emails a day that even the smallest LLM should be able to classify correctly.

a-r-ton Ask HN: Can I really create a company around my open-source software?

You mentioned Django, but I couldn't find much on your website about how it works with your forms. Will I have to manually replicate all form fields on the client side? Will it work with Django form validation and show errors? I'd suggest creating a documentation page for each framework you mentioned to explain how it will work together.

a-r-ton HP to Acquire Parts of Humane for $116M

The goal from the start was to raise $230M and then get acqui-hired for $116M by HP?

a-r-ton HP to Acquire Parts of Humane for $116M

https://support.humane.com/hc/en-us/articles/34374173951373-...

"Your Ai Pin will continue to function normally until 12pm PST on February 28, 2025. After this date, it will no longer connect to Humane’s servers, and .Center access will be fully retired."

a-r-ton We were wrong about GPUs

Off topic, but the font in the article is hard on the eyes.

a-r-ton Frozen human brain tissue was successfully revived for the first time

From the article: "The researchers also applied their technique to three-millimeter cubes of brain tissue from a 9-month-old girl with epilepsy. The tissue maintained its pre-freezing structure and remained active in a laboratory culture for at least two weeks after being thawed."

a-r-ton Generative A.I. arrives in the gene editing world of CRISPR

So the six finger hands were just a foreshadowing?

a-r-ton Ask HN: Should Banks Phish Their Own Customers

Something like a quiz similar to [0] with a small reward would be better.

[0] https://phishingquiz.withgoogle.com/

a-r-ton Game of Life, simulating itself, infinitely zoomable

If anyone is interested in reading about this idea, I recommend Stephen Wolfram's "A New Kind of Science" [0] and Max Tegmarks's "Our Mathematical Universe" [1].

[0] https://www.wolframscience.com/nks/

[1] https://space.mit.edu/home/tegmark/mathematical.html

a-r-ton Direct File officially opens in 12 pilot states

My state is not in the pilot, so I can't try it myself. Have anyone here used it yet? How was the experience?

a-r-ton Both pilots of an A320 fell asleep in the cockpit for 28 minutes

Do a barrel roll :)

a-r-ton Show HN: Replay your typing in a few lines of JavaScript

You can introduce multiple types of changes, for example: addition at the end, removal of N characters from the end, and complete string replacement, etc. Then for each change you can calculate which type is more efficient to store.

a-r-ton It's 29—Delphi, I mean

Delphi was the first GUI-based development experience for me back in '96-'97. It was the next step after learning Turbo Pascal and writing command line programs. I still get warm feelings every time I see it mentioned somewhere. The built-in and third-party components were my favorite part.

a-r-ton OpenAI negotiations to reinstate Altman hit snag over board role

Apple is sitting on $162B in cash right now and would definitely benefit from OpenAI's IP.

a-r-ton Apple agrees to $25M settlement with US over hiring of immigrants

Banana for scale: Apple makes that much in revenue in about 30 minutes.

a-r-ton NSA publishes ten most common misconfigurations in networks

11. Using encryption algorithms compromised by NSA

a-r-ton How companies like Amazon and JPMorgan spy on their staff

> Is "sleeping" the macbook and closing it shut, enough?

For Apple silicon-based (and newer Intel-based), yes: https://support.apple.com/guide/security/hardware-microphone...

a-r-ton Japanese scientists working on creating human eggs and sperm in the lab

I've seen things you people wouldn't believe...

a-r-ton The Decomposition of Rotten Tomatoes

A lot of great movies bombed at the box office. Marketing plays a big role.

a-r-ton The Decomposition of Rotten Tomatoes

538 compares polling data against actual election results. What are you going to compare a critic's ratings against? If you compare them to the average critic or audience rating for a given movie, you will end up with a system that punishes any independent thought.

a-r-ton Recursively summarizing enables long-term dialogue memory in LLMs

This is what ChatGPT's Code Interpreter does (writes code in Python and then runs it to check for errors). I'm not sure if it's enabled for everyone yet though.

a-r-ton Recursively summarizing enables long-term dialogue memory in LLMs

Because if you had a logic layer capable of validating LLM's output, you wouldn't need the LLM.