[16/01/25] on AI clickbait

This guy makes nothing but bangers. I wanted to write a summary, but I don't think there is much to summarise, since all of his points are valid and compact.

And so, when something inevitably goes wrong, it's going to be treated as "the AI lied to us, bad AI, we should punish the AI" instead of "whatever idiot hooked that AI up to the thing that went wrong was negligent and shouldn't have done that."

source: Ai ClickBait, and how hard it is to discuss AIs intelligently

[02/11/24] on web dev

This week, I had the genius idea to finally build a portfolio website. I already knew JavaScript was an absolutely ridiculous language, but now I found out that Web Dev is screwed as a whole.

My biggest frustration was images, which had a white line at the bottom by default? I thought this was an issue with my coordinate system, which of course it wasn't; my code never fails, so I started googling around to find out how to fill up a div with an image. My code was exactly the same as the big tutorial websites? So I googled my issue more specifically and found a stack overflow thread (OP was downvoted, and I still do not know why?). This fixed the problem. But when I looked up what those settings actually meant, it said nothing about the line. (Bug Preview: https://codepen.io/agpvdwdz-the-solid/pen/RwXJxwO)

My second frustration was exactly this. The fact that each element has a list of hidden properties that just randomly screw me over. One of the things I like about programming is that I control everything that happens. CSS and HTML felt like the opposite. Everything I typed had effects that I did not specify. In multiple languages, this is the case, many libraries have default settings within their functions. However, these are always fully documented. When looking up the documentation, all functions start off with func(x, y, z=default, v=default, w=default). The MDN Web Doc, recommended by all threads when you Google HTML/CSS docs, does not have a section on this. W3 schools hide it all the way at the bottom and only if you look up a specific element. This should have been properly documented after 30 years, right? Or, at least, be more intuitive.

This ties into my last point, these languages are not new and are used almost everywhere (thanks electron). How can it be that they suck this much? I haven't even talked about variables, (I mean root :: {—var} and var(—var) are absolutely insane ways of using and declaring variables right?), file structure, and responsiveness (there are like thousands of different devices nowadays, and instead of responsiveness being a default setting it is a ###### white line!!!). Why are people talking about crypto, NFTs, and the metaverse when discussing Web 3.0? I want websites to be fixed instead! Give me a python + .md + OOP-css browser and I will buy your Oculus Rift.

AI intermezzo
As a “back-end” developer, I always thought LLM coding tools were made for front-end devs since they definitely could not solve any intermediate back-end problem in a sophisticated way (please check out the Internet of Bugs' videos for demonstration). However, every time I build something in HTML, CSS, or JS, I find out it can't do that as well. Who are these people who use AI? If it is not front-end and back-end? I feel like I'm missing out on something :) (Or, the more reasonable alternative, everyone is lying)

However, not everything can be blamed on HTML and CSS, some things are just my poor programming skills, or computer skills in general. For example, when googling the default screen size, I did not realize that my screen was not in the list of most used screen sizes, so everything was completely misplaced. I did not know this 3 years ago, which made me give up web design back then, but I realized it now. Furthermore, I firmly believe that some of these errors I made are just skill issues, and if I understand the web better, I will appreciate it more as well.

[04/10/24] on chess heuristics

This is the first time I recommend a channel as a whole. This is because the creator mentioned did not make a playlist for the series I wanted to recommend.

The series is about chess bots, more importantly chess heuristic. The series is in a classic "chat/my-viewers decide X" format, since his twitch chat decides on what heuristics to use when evaluating the search tree of the game.

This leads to some interesting questions, both on chess, and heuristics as a whole

source: Green Lemon Games

[01/10/24] on 138x development

This video is on Thibault Duplessis, Lichess' core developer. A truly inspiring developer who mastered the idea of simplicity in his code. His whole tech stack is built on elegance and maintainability. Since I am only familiar with scala from his stack (which apparently stands for Scalable Language) I was really interested in the tech that was mentioned in the video.

I also really like the fact that Thibault was critical on his current stack, and reflected on the choices he made in the past. Giving advice to people who try a similar project. Maybe I should try to build an actual personal website + YT blog with this stack. But on the other hand, learning socket programming in scala to maintain simplicity is not really the way to go?

Ugh, so in love with this guy, the whole elegance of only using features you need, small frameworks, 0 bloat, the constant focus on tech depth... Pls watch it :)

source: How 1 Software Engineer Outperforms 138 - Lichess Case Study

[26/9/24] on unicorns and think tanks

This post is 10 months late, since the drama in the OpenAI company is already over. Still, I think this video perfectly illustrates how OpenAI went from a company focusing on safe AI to the fastest growing tech product in the world.

Furthermore, the fact that Sam Altman is the former president of Y Combinator is one of those facts which I now will quote in every possible conversation.

“His ethos, what he was good at, was taking ideas and commercialising them.”

*the most successful start up accelerator of silicon valley AKA just some VCs

edit: OpenAI to remove non-profit control and give Sam Altman equity

source: Good Work - What happened with OpenAI

[21/9/24] on quants

"Quants work with things buried deep in the math, uncovering insights through massive data analysis and machine learning. There’s extensive research on predicting weather patterns, such as the temperature in Nebraska five days ahead. If it’s going to be three degrees hotter than expected, even a minor change like that can influence oil pipelines running through Nebraska, potentially affecting the cost per liter by a fraction of a cent. These slight adjustments can ripple through the markets."

source: Good Work - What do Wall Street quants actually do?

[20/9/24] on python rats

Random ass 300 view youtube video on Python Rats. Interesting concept. Pretty scary considering the amount of code needed to be checked if you install a repo.

source: I HATE PYTHON RATS :(

[17/9/24] on starting your own company

"All of the worst programmers I have ever worked with, some of them very senior at large companies, spent most or all of their careers in teams where they avoided being wholly responsible for anything. I recall one of them briefly trying to do his own thing, but he didn't like it and quickly bailed. These folks, mostly men, all considered themselves too good for QA or first-line support. They never worried about losing money because customers stopped using their software.

Now, not every programmer with that background was awful, but most of the awful ones shared that background. Interestingly, none of the best programmers I've worked with came from that mold. All the really good programmers either did their own thing at some point or worked at very small companies where they were running the whole show."

source: internet of bugs

[10/9/24] Hello