It's WILD how redpills are abundant in most social media. YouTube just recommended me a video. I though it was strange, so I opened up in an anonymous tab. Found that it's a channel that makes stories. But ALL OF THEM are misogynistic; and every video shows them as if they're true.

When did the Google Drive logo changed to be the Triforce?

I am 43 years old. And I am tired. When I was young, I though misogyny was a solved problem. That people who still believed in "gender roles" would be just old people, failing to modernize their ideas. And now, the "redpill movement" became popular. I though that the internet would solve education

Is it all lost? Instagram is full of redpills, fascists, racists, etc; Reddit is full of the same. Twitter is hell on earth. Xenophobia is rising again, including where I live right now. AI made the whole "open source model" nil and void because in hours you can literally rewrite a whole app.

This is SO FREAKING STUPID. The Claude problem is on their side. I asked for help, and their "help chat" doesn't work on Firefox Then their "help chat" is a bot (why wouldn't be, right?) and it says that I need to open an issue. It, of course, can't open to me. Then it asks me to open in the chat

The Internet. A tool made for sharing files. HTML are files that are interpreted in a specific way on the browser. That's all. Claude's web UI simply refuse to download my files. Like... since ever. When they can't do the thing that internet was MADE to do, on the internet... what does that say?

Now there are people arguing that the Claude Code leak is a security issue, because "everybody can see the code and try different attacks". Are we really that dumb? Did we forget all the good practices of the last 30 years? That security by obscurity is not real? What the hell is happening?

LinkedIn, last week, was full of idiots saying that AI changed the game completely, and that's the solution for all the problems in the world. Today, it's full of idiots saying that AI is not the answer for anything, and it's dangerous. Of course, this was predictable... except if you're and idiot

Talk do @mauricioszabo.bsky.social divulgada 🎉 "Usando Clojure para Gerar Javascript para rodar Clojure que Executa Ruby (e Python)" Interactive coding (estilo LISP/Smalltalk) para linguagens tipo Ruby e Python? Sim, é possível, melhor do que você imagina, já existe, e você já pode usar!

I hate that I had to accept "Techbro" is pervasive in tech. I just saw a post about someone pleading for some "integrated medical system" where patient's data would be available globally in my country... a country known for corruption, and for bad private healthcare system. And NO ONE was against.

After a couple months of silence, I finally could capture how I feel about LLMs, and wrote a bit about it: mauricio.szabo.link/blog/2026/03... Ok, maybe "more than just a bit" - but yeah, honestly, I HATE how the rules can be ignored when the subject is "AI", somehow...

I asked Claude for a specific feature on my plug-in. It emitted almost 2k lines of code. My plug-in - all of it - including tests - have about 8k lines of code. It seems that I won't be using Claude to code this feature...

I REALLY want to study a way where I can use AI as my "parser" instead of my "writer". Where it helps me understand the problem and guide me to the right places to fix it, instead of fixing it for me. So I can evaluate a code, see an error, and pass that to AI instead of guiding it to the error

MoltBot was acquired by Meta. A vibe-coded, insecure, wasteful, and even a scam, of a piece of software that delivers nothing useful and wastes A LOT of energy and money from all the involved people, was sold. Probably by millions. What the fucking are we doing?

I think when we started the Pulsar editor, we got only negative messages. "Why do it?" or "It's impossible", and "It's outdated", "It's buggy", etc. Nowadays, when I search for Pulsar... I mostly get amazing messages. "Oh, it's so much better than Atom", etc. I am honestly very happy!

move slow and repair things

Honestly, I still can't believe that the Pulsar Editor exists. I wrote SO MUCH on how hard, close to impossible, was to keep a fork of Atom. On how it was dying, buggy, slow. How the syntax highlighters were weird, how the API never evolved, etc. And now, years later, I'm helping to keep a fork

Congrats, @circleci.com - you are broken. You had a GREAT CI system in the past. And I mean it. That was until you started to "fix" what wasn't "broken". That was before your automated systems start to randomly block without explanation, and your support didn't care.

I was JUST asking a couple of friends some minutes before on why companies are so proud to announce they are "AI-First Companies". When I was younger, dehumanization was a bad word. Nobody wanted to be associated with it. Now, it's a "feature", it's "desirable", it's "marketing" and "innovation".

I kind of found recently that gambling is supposed to release dopamine and/or be fun... which is something I never experienced. I think this part of my brain must be broken, which is not a bad thing, but it's just one more that I found I can't relate with most people.

I feel, after investigating a little about what's public about Epstein, that Christians were right: THERE IS a hell. They just misplaced it - it's not after we die, it's here. The demons walk around, they have most of the power, and money, in the world. And we're their slaves thinking we're free.

Goodbye, @firefox.com. I tried to like you, but it's not working anymore. Tabs are buggy (can't reorder, can't detach,they get "stuck" in the bar when I try to drag them). My token stopped working. Webcam detection is TOO SLOW. Sites break, lots of crashes.... Sad to change back, but I need to.

Did we all suddenly became dumber with the whole AI stuff? Everything that we were able to do BEFORE AI suddenly needs it. Things like "MIDI to MP3" return A LOT of "AI sites" that do the conversion, something we were able to do without needing it. PDF to text? Same. OCR? Same. What is happening>

I can't understand why more tests framework just don't present how they show their failures. It's mostly how I judge which framework to use. An error like NOT EQUALS: [0, 2, 1], [0, 1, 2] is not useful. I want diffs. I want custom ways to compare stuff. I want my test framework to HELP me.

I hate - HATE - these automated, AI-based tools that detect fraud. I am just locked out of @circleci.com. The reason? I tried to build an AppImage. That's it. I had a first try, it was aborted; I pushed a change... and I can't login anymore. The idea of "making my own CI" grows stronger...

I somehow was able to upgrade MidiEditor to work with QT6. Unfortunately, most of the work was made by an AI. Which is what I feel it's good of: making "random changes until it compiles" works well if you can automate the process! If anyone still used MidiEditor: github.com/mauricioszab...

I'm so surprised how Google apps became so unfriendly... Take Android auto, for example. Amazing app... except that maps' search isn't synchronized between my android and my car. I can't also "disconnect", and can't say "don't auto-connect". It's such insanity to have an app that I can't QUIT.

Just saw an amazing quote: As software ate the world, we lost the protections from the real world. Amazing how much this rings true. And sad because, as a software development, I keep thinking if I'm not part of the problem, instead of the solution.

How Google can be SO INCOMPETENT in some cases? I just got some "security warning, your password was seen on public sites!" on my Google Account. So I got the details... ... and it's all private IPs, 192.168.<something>, and more than half of them are "localhost:<some-port>"

Sometimes I feel like an alien. I see job offers, and they all say something like "we have AI in our DNA" or some other shit like this... ... and I immediately don't want the job. I don't even want to know the salary. Like, close the offer, ignore the company. Why is this sold as an "advantage"?

What the hell happened with C++ while I wasn't looking? I was trying to compile something to WASM. I used Emscripten, but the build failed. Now I'm debugging bash that calls cmake that calls emake that calls python that calls em++ that calls clang...

I find quite amusing that USA by its own definition of "terrorist state" is now one, bombing innocents in international waters to literally terrorize governments I also find quite convenient that USA's definition of terrorism state excludes USA from the definition I would laugh, if it wasn't scary

Silksong is SO POLISHED! There are so many cool easter eggs, small details, jokes, secrets, things to explore... Even Hornet's personality is amazing - strong, determined, firm, kind, responsible, respectful, but also bold and proud. I REALLY didn't expect the game to be this good!

Oh my gosh, @teamcherry.bsky.social, you did it again! Congratulations, Silksong is amazing!

it does seem weird to declare war on people and then demand they mourn your dead

I was prepared for Silksong to never happen. I was prepared for it to be under-released, not living up to the hype. I was even prepared for it but be kind of like a "Hollow Knight mod", with the same movements, just a re-skin of the game. I was honestly not prepared for it to be THIS GOOD!

I reported more than 18 people to Instagram related to hate speech, since they changed their policies. Not a SINGLE ONE of them was considered "hate". Why have any moderation team at all, at this point?

Ender Lilies appeared to be a great game in the beginning. Unfortunately, it's frustrating, and not in the good way. First, it seems that "contact damage" was an afterthought-it doesn't really fit. Second, it tries TOO HARD to be linear. The map opens, but then you hit dead ends all the time.

It seems that you can promote eugenics and get REALLY CLOSE to fascism if you present an attractive woman right? Gosh, I so hate this stuff...

I wish I could like Beyond All Reason (an open-source RTS game). It's polished, it's beautiful, it works, have a strong AI, very interesting ideas, etc. It's also HIGHLY complex. 5 different "kinds" of units, 4 "tech-2" level units, one "tech-3". Impossible to have all these in a single game.

Talk "Usando Clojure para Gerar Javascript para rodar Clojure que Executa Ruby (e Python)" por @mauricioszabo.bsky.social Interactive coding (estilo LISP/Smalltalk) para linguagens tipo Ruby e Python? Sim, é possível, é melhor do que você imagina, já existe, e você já pode usar!

I still can't believe that we don't have some kind of documentation tool where searches simply work, and that uses LLMs to discover if you're documenting something that already exists, reads source and automatically adds documentation, and semantically searches what exists to summarize a result...

So, Mortal Kombat will have a new film. It's weirdly named "Mortal Kombat II", despite being the 4th. The third had the same name as the first, "Mortal Kombat". If you can't even give a meaningful name to your movie, maybe... stop doing it? It's not like ANY of the movies were remarkable, right?

LLM and Python are really a perfect match, like no other. When you try to run a Python project, it never works first try, seems like nothing is deterministic. It might work now, but it won't work tomorrow, depending on a lot of variables Exactly the same as LLMs.

There's something deeply cursed in Node.JS + Electron. If you do the wrong combinations of libraries, things fail to install or compile. EVEN IF you don't bump the actual thing that's failing, but for some reason, bumping something completely unrelated crashes. And, no, it's not transitive deps.

Every time I am angry with my current job, or some code that I am doing that I can't make it work and that would be simpler in <insert any other language here> I will remember that I am glad I don't work with Node!

What the hell is happening to the world? At this point, I am VERY GLAD to be living in a country nobody cares about. Hopefully I won't be caught in some cross-fire...

I wish I knew how to draw... I have so many weird ideas for webcomics, and can't draw a single finger. Imagine a webcomic showing "The Legend of Zelda" but Link is a "speedrunner": Gannon waiting for Link, while he sneaks behind him, opens the door, and snatches the Triforce

Recently, I reported more than 20 comments in Instagram. Not a SINGLE ONE of them was removed. Not a SINGLE ONE of them were "against the guidelines", even the ones that LITERALLY SAID to test dangerous chemicals in people. Meta is rotten to the core...

I really can't understand how Google can rot everything it touches. Google Slides is now full of weird bugs. You select to animate something, but sometimes it changes the animation's duration to be 0s. Meaning there's NO animation. Also, why isn't there an option to change the default animation?