Hacker News Reader

manxon Lessons from 15 Years of Indie App Development

Can anyone recommend good learning material for the non-tech part? I'm good at building stuff, but not on the other parts.

manxon All Hacker News posts dataset on Google BigQuery

You can use it for free from kaggle: https://www.kaggle.com/code/felixdietze/hacker-news-score-an...

manxon Show HN: I build and deployed AI Food Recipe Assistant quickly

Great! What did you use for image generation? How much does it cost you?

Edit: It's DALL-E 3.

manxon Ask HN: Which Books or Articles Do You Keep Rereading and Why?

Excellent question. I asked the same question for Movies: https://news.ycombinator.com/item?id=41315330

For me it's:

- https://drewdevault.com/2018/07/09/Simple-correct-fast.html

In combination with "Simple Made Easy":

- https://www.youtube.com/watch?v=SxdOUGdseq4

manxon Mergiraf: a syntax-aware merge driver for Git

Great to see more work in automatic git conflict resolution! I'll definitely give this a try. My own attempt at such a tool involved character based diffing and patching. It is able to solve most trivial conflicts that git cannot: https://fdietze.github.io/blend/

I never found the time to make a cli out of this.

manxon ChatGPT Search

Has anyone created a RAG on Wikipedia yet?

manxon Using reinforcement learning and $4.80 of GPU time to find the best HN post

Very interesting! Identifying great new content is a big unsolved problem for HN IMHO. Unfortunately, scores are not a good metric to predict, because they are not comparable (see https://felx.me/2021/08/29/improving-the-hacker-news-ranking...). A better metric might be "upvoterate", defined as how much more or less likely users are to upvote a story compared to the average story. More about that here: https://github.com/social-protocols/quality-news?tab=readme-...

manxon Using reinforcement learning and $4.80 of GPU time to find the best HN post

Scores are not a good metric to be compared. I did some data analysis and wrote about it here: https://felx.me/2021/08/29/improving-the-hacker-news-ranking...

manxon Streaming joins are hard

I think it should be possible to create a compiler which transforms arbitrary sql queries into a set of triggers and temporary tables to get incremental materialized views which are just normal tables. Those can be indexed, joined etc. no extra services needed. Such an approach should in theory work for multiple relational database systems if it's all adhering to standards.

manxon Nixiesearch: Running Lucene over S3, and why we're building a new search engine

I thought about creating a search engine using https://github.com/phiresky/sql.js-httpvfs, commoncrawl and cloudflare R2. But never found the time to start...

manxon The Static Site Paradox

What are good WYSIWYG editors for static sites? Any recommendations?

manxon AI is an impediment to learning web development

Humanity was only able to produce one generation who knows how computers work.

manxon Algorithm

How much faster your algorithm is than state of the art on big problem instances.

manxon Why You Should Learn Linux (As a Developer)

You might enjoy using https://www.nushell.sh/

manxon Why You Should Learn Linux (As a Developer)

Same for NixOS

manxon Linux desktop market share climbs to 4.45%

NixOS does this too.

manxon Show HN: InstantDB – A Modern Firebase

Looks great! How is the migration story when schema changes are needed? How do you deal with old clients?

manxon Ask HN: Learning Linux from Scratch

https://www.arcolinux.info/ might be interesting to you

manxon Elon Musk and the Danger to Democracy

This article reads like Elon became a victim of a filler bubble on his own platform without realizing it. nobody really sees their own filter bubble.

manxon Where Facebook's AI Slop Comes From

So many stakeholders with different incentives: advertisers, facebook, regular users, paid content creators, ...

How could we design incentives which produce a positive outcome for everyone?

manxon Where Facebook's AI Slop Comes From

The question now becomes: how to classify useful content. I doubt that's possible without downvotes and some more advanced algorithms than majority voting. Right now, those systems are full of probably unintended feedback loops.

manxon Gen AI: Too much spend, too little benefit [pdf]

What is your setup?

manxon Show HN: I do nothing in my dayjob for 95% of the Time so I made this

There should be a version for the terminal which looks like logs.

manxon What's New in Neovim 0.10

neovide is a nvim gui that feels good

manxon Most of Europe is glowing pink under the aurora

Do you mind sharing a photo?

manxon Show HN: LLMs following grammar for general-purpose PLs

It would be interesting if these constraints could be extended to give even more guarantees, like type safety. Such constraints probably need some kind of backtracking.

manxon The best way to have complex discussions?

Or the almost 200 other tools and experiments: https://docs.google.com/spreadsheets/d/1wQShF0J3lGmIFACTvt5F...

It's a difficult but important problem to solve and we need more people in this space.

manxon Wild orangutan seen healing his wound with a plant

The book "Guns Germs and Steel" has some good answers to that question.

manxon Why SQLite Uses Bytecode

I'm wondering if one could write this bytecode directly (or with a higher level imperative language) instead of SQL. Often, the programmer knows exactly which index lookups need to happen in a loop, while it seems like a burden to express that in SQL. This might also be an opportunity to create a different type safe dsl for database access.