Eduardo Krustofski

@dhh my kid is playing Zork, on a MacBook running Omarchy. He just understood how the shell works because it works just like in Zork. Thanks for playing a part in this.

@Charles_SEO Regarding Hubspot: do you mean that just by using it you get penalized?

Welcome to Day #11 of the Core Update 🚨 So far, we have seen: - The initial surge of listicle / parasite domains from the spam update falling back off again - Subdomain / canonical abuse getting more prevalent than ever, especially in international SERPs - The biggest winner so far? YouTube... - Reddit saw some initial declines, and is now GOING TO THE MOON, from our tracking this is because it hadn't rolled out as strongly internationally as it did in US/UK/CA but has now gone fully global 🌏 - Thin, low quality pages (and by extension, techniques like pSEO) are still being targeted - There seems to be large overlaps between these last two updates (and the testing we saw in the weeks leading up to it) and the August 2025 Spam Update - HubSpot, that got hit hard by Aug 2025 and several previous updates, has continued losing positions and traffic With all that being said, this update is still likely largely, or at least trying to, feed the ingestion of the AIOs; This is mostly through information gain scoring, aka, how much the content matches consensus vs. how much it's contributing new information against existing ranked sources. It's also getting increasing difficult to isolate variables with such a messy algo anyway... AI Overviews are increasing, again, and Google recently released research on "TurboQuant", a new compression framework that gives near instant retrieval for models while using significantly less processing resources - If they haven't already rolled out that tech for AIOs, they will do, which substantially lowers the cost to deliver and brings us one step closer to a default Google AI search experience.

@denisyurchak hi Denis, can you send me a DM? We are looking for someone that has an eSIM API. If that is not your use case, feel free to ignore 🫣

Whatever results I had promised myself I could deliver will first have to deal with the damage this might have caused. The image I posted is how things are right now. Let’s see how far I can take it.

A subdomain that was not in use anymore was still pointing to an AWS server. Someone found this out and created half a million pages of adult content. When I found out, this had been going on for almost a month.

After more than 15 years of not doing anything related to SEO, I have decided to catch up. As a pet project I decided to help a friend’s business. His website has been up for many years and has some SEO work done years ago. When I opened Search Console I got a surprise.

@wesbos Presente 🫡

Only cool people can reply to this

@lmanchu @mitchellh I finally understood why coding with LLMs feels taxing when really reviewing their output. Just like there is a big difference between listening and active listening, there is one between vibing and active understanding code.

@mitchellh Exactly. "100% agent-written" often hides the 1000% human-directed effort. At IrisGo, we see the cognitive load shifting from execution to architectural integrity and strategic nudging. That's the real engineering challenge now.

Btw, for Ghostling (libghostty demo), I didn't write a single line of... anything. Agents wrote 100% of everything you see incl. Nix flakes, CI jobs, etc. I reviewed every line of code manually and constantly nudged the agents in the right direction. I used a mix of Opus+Codex. Note: libghostty itself is of course heavily hand written (with agent assistance all over too, though). I'm just talking about Ghostling itself. Even for CI setup (GitHub actions), I had the agent sit in a `gh` CLI loop watching failures, fixing, pushing, fixing, pushing, etc. Doing GHA with agents are the only way I stay sane, honestly. I just re-read the full main.c from top to bottom and I'm very satisfied. I would've done some things differently but if an engineer I worked with PRed all this, I would've accepted it. Its good enough. Some people have pointed out the commenting is over the top and indicative of AI. This is actually my personal comment style, and I told the agent to comment heavily (its even in the AGENTS.md, look for yourself!). Anyone who has worked with me professionally or in OSS knows that I comment everything all the time.

I just found out that when you take a screenshot of a twit the X app adds its URL. Nice touch.

@patio11 @BeenThereCap I personally have said to my family that if someone claims to be me they should ask for proof or hang up. No matter the urgency. And this was before 2015. It will only get worse 🫠

@patio11 @BeenThereCap Here in México we’ve had scam calls from people claiming to be certain person and they needing urgent money because of <important and urgent reason>. And it works. This was before a.i.

If you are reading my words here, it will be worth it for you in your personal life to read Patrick's thread below.

@elmundoverdees En dos cines y en imax. Pero hay

Se acabó ... Que final no mms. No es el epitome de la ciencia ficción, pero se le acerca. #projecthailmary

@karpathy I feel like I need a square monitor. And no, I don’t want to put too monitors side by side.

Expectation: the age of the IDE is over Reality: we’re going to need a bigger IDE (imo). It just looks very different because humans now move upwards and program at a higher level - the basic unit of interest is not one file but one agent. It’s still programming.

@TheRooster The UI is a mess. They had to go out of their way to make it like that. No reason other than maxxstyling.

I’ll confess something about Marathon I’m having fun. And I know why:

@twilio we just got two charges due to auto-recharge. But our MTD usage is way below what those two charges were for. How can I tell what caused them?

@josevalim @mateuscolvr It was the website in general but it is working now.

@mateuscolvr @edmz Only that page or the website in general?

Most people treat the static vs dynamic debate as a foregone conclusion, but in practice type systems restrict the expressive power of programming languages in ways that leak complexity to developers. New article: Type systems are leaky abstractions — the case of Map.take!/2

@josevalim I’m in Mexico, btw.

@edmz It is working for me :S Anyone else sees it down?

@josevalim (Is Twitter still penalizing links? I assume that’s why you put the link in a second twit) Link seems to be down

Read the article here: dashbit.co/blog/type-syst… Spoiler alert: it is not about which is better, static or dynamic, but rather an attempt to shine a light on the underlying trade-offs. I'm under no illusion I will change anyone's mind on this topic.