Hacker News

Article URL: https://medium.com/@szabomarius/on-wasted-effort-and-vibe-coding-52beee533025

Comments URL: https://news.ycombinator.com/item?id=43751242

Points: 1

# Comments: 0

Article URL: https://www.musicmint.ai/vocal-separation

Comments URL: https://news.ycombinator.com/item?id=43751240

Points: 1

# Comments: 1

Article URL: https://en.wikipedia.org/wiki/Swill_milk_scandal

Comments URL: https://news.ycombinator.com/item?id=43751230

Points: 1

# Comments: 0

Article URL: https://blogs.igalia.com/carlosgc/2025/04/21/graphics-improvements-in-webkitgtk-and-wpewebkit-after-the-switch-to-skia/

Comments URL: https://news.ycombinator.com/item?id=43751217

Points: 1

# Comments: 0

Article URL: https://link.springer.com/article/10.1007/s11127-025-01277-2

Comments URL: https://news.ycombinator.com/item?id=43751215

Points: 1

# Comments: 0

Follow up post from: https://news.ycombinator.com/item?id=43742639

TLDR: My attempts of jamming a somewhat smart conversational chatbot into a QR code (with a UI)

Here's what makes SYNEVA work:

I went with a pattern-matching approach instead of neural nets or tiny transformers (I tried that in archived versions, we DO NOT have space for this sadly). The entire system is built around our S class that's basically:

S {

constructor() { this.m = this.L() || {}; // Memory/learned responses this.v = {}; // Word frequency tracker this.c = []; // Conversation history this.h = ""; // Last message } }

*Memory & Learning*

SYNEVA can actually "learn" during conversations! We can't use localstorage because this is a BASE64 URI but we can still use SessionStorage. When you chat, it:

Creates context keys from the last 2 normalized words Stores your input/response pairs in this.m Tracks word frequencies in this.v for relevance scoring Persists this data inside the URL hash

*Response Generation*

For responses, it uses a multi-stage fallback system:

Pattern matching against phrases/questions (the few it has) Context-based lookup from some learned responses Word frequency analysis for fuzzy matching Default responses as last resort if nothing works

This lets it seem somewhat intelligent despite the size constraints.

*UI Layer*

Added a kinda retro looking green-on-black terminal interface and managed to get a typing animation!

Everything's packed into 2.8kb including HTML/CSS/JS and it's all in a QR code

Previous attempts with neural networks went upto 5.3kb even with 8-bit quantization and word-level processing. This pattern-matching approach with context awareness turned out way more efficient while still being somewhat "smart".

You can actually share your "trained" chatbot too - just share the URL with its hash, and your conversation context gets preserved :P

I'd love to know your thoughts or if you can make something better in the constraints, feel free to contribute

Comments URL: https://news.ycombinator.com/item?id=43751199

Points: 1

# Comments: 0

Article URL: https://medium.com/@dwolela/citizenm-miami-brickell-the-hotel-that-knows-exactly-what-its-doing-mostly-58ad15dc3583

Comments URL: https://news.ycombinator.com/item?id=43751195

Points: 1

# Comments: 1

Article URL: https://www.ft.com/content/b4c868aa-28e7-4f35-8357-88e59a4fa9c2

Comments URL: https://news.ycombinator.com/item?id=43751177

Points: 1

# Comments: 1

Article URL: https://sam.elborai.me/articles/tabular-programming/

Comments URL: https://news.ycombinator.com/item?id=43751168

Points: 1

# Comments: 0

Article URL: https://pubmed.ncbi.nlm.nih.gov/40227745/

Comments URL: https://news.ycombinator.com/item?id=43751157

Points: 1

# Comments: 0

Article URL: https://www.benkuhn.net/impact/

Comments URL: https://news.ycombinator.com/item?id=43751135

Points: 1

# Comments: 0

Article URL: https://nvd.nist.gov/vuln/detail/CVE-2025-30401

Comments URL: https://news.ycombinator.com/item?id=43751122

Points: 1

# Comments: 0

Article URL: https://thefoggiest.dev/2025/04/21/daily-driving-a-linux-phone

Comments URL: https://news.ycombinator.com/item?id=43750756

Points: 1

# Comments: 0

Article URL: https://phys.org/news/2025-04-ai-reimagines-gravitational.amp

Comments URL: https://news.ycombinator.com/item?id=43750753

Points: 1

# Comments: 0

Article URL: https://www.enlightenment.org/news/2025-03-23-terminology-1.14.0

Comments URL: https://news.ycombinator.com/item?id=43750696

Points: 1

# Comments: 0

Article URL: https://www.whitehouse.gov/lab-leak-true-origins-of-covid-19/

Comments URL: https://news.ycombinator.com/item?id=43750694

Points: 2

# Comments: 0

Article URL: https://techcrunch.com/2025/04/20/openais-o3-ai-model-scores-lower-on-a-benchmark-than-the-company-initially-implied/

Comments URL: https://news.ycombinator.com/item?id=43750691

Points: 1

# Comments: 0

Article URL: https://octet-stream.net/b/scb/2025-04-21-ai-a-fork-in-the-road-for-open-source.html

Comments URL: https://news.ycombinator.com/item?id=43750681

Points: 1

# Comments: 0

Article URL: https://fastpanel.direct

Comments URL: https://news.ycombinator.com/item?id=43750679

Points: 2

# Comments: 0

I’ve always been curious about big, complex topics; things like quantum computing, AI ethics, or philosophy, but I’d often get stuck at the very beginning. I’d open a dozen tabs, skim Wikipedia, and still not feel like I had a good mental model of what I was even looking at.

So I built a tool to help with that. You type in any topic, and it generates a simple mind map; just a few high-level subtopics to start. Then, as you click and explore those nodes, it expands outward with more specific, connected concepts. The map grows with your curiosity.

It’s intentionally lightweight and incremental. The goal isn’t to overwhelm you with everything at once, but to help you gradually build a clearer picture and follow your own curiosity in a structured way.

I’ve been using it to explore new areas, plan research, and even prep for writing projects. If you're the kind of person who likes to "see" what they're learning, I’d love your feedback.

Comments URL: https://news.ycombinator.com/item?id=43750661

Points: 1

# Comments: 0