T-posing Engineer

@ryanvogel I got greedy 3 days ago and set a ralph loop to do something way bigger than I should've. Ran out of credits. I've been doing shit by hand since and let me tell you that not only have I been mega productive, but I've also learned in the process and felt more accomplished doing so

how about…. and hear me out here… instead of spending $200 on Claude to build a crappy clone of an app you don’t wanna pay for to optimize one small thing you just…. pay $20/month and use your brain for once

@tenobrus Hard to compare apples and oranges because they're both amazing people, just in different ways.

who's a better person, your mom or your dad?

@albertadevs @JustDeezGuy If it has to be done by hand, there'll be only like 500 people in the world left that'll be able to pass that interview. I already 100% suspect none of the mfs I work with remember how to do the most basic string manipulations or sorting... I would even bet on it.

I just know when I’m interviewing for a senior vibe coder position in 2030 they’re still going to be asking how to reverse a linked list

@ChShersh @lisyarus Eeey new profile pic

@lisyarus We want more village game updates

One of the biggest refactors so far in my toy language compiler: improved referencing struct types by generating unique IDs (instead of using AST node pointers as keys, lol)

@weswinder @theo I wonder if that's what they do... 🤔 1) release model better than the last one 2) people flip out and go "it's over" and "le exponentials" mode. 3) make model gradually dumber 4) release new model better than the dumber version of the last one 5) repeat step 2

@theo opus 4.5 in december was insanely good. idk what they did to nerf it, but both 4.5 and 4.6 seem wayyyy dumber now than before

Just let Opus go for over an hour on a new feature. When it was done, I asked how I can test it. 20 minutes later, it realized I can't test it because it did the whole thing entirely wrong. Idk how you guys use this model every day for real work 🙃

@MichaelArnaldi @SharpCoder @EffectTS_ some time ago you said the Effect mcp was also good. Is that still the case?

@SharpCoder @EffectTS_ GPT-5.4 is exceptional at Effect, just add the effect repo as a git subtree to give the agent access to best practices, skills don't work well. Check out effect.solutions

GPT-5.4 is not really good at @EffectTS_ , is there any official skill to help him out?

@TheRundownAI I'm actually ok with how this turned out. Call me a pessimist, though, but I really doubt this will become the norm...

Someone used Suno AI to generate a Japanese metal band called Neon Oni. Fake member bios, AI-generated music videos, "Based in Tokyo" on Spotify. 80,000+ monthly listeners. Fans had it in their Spotify Wrapped top 5. Merch was selling. Then, community sleuths exposed it. Traced the creator's account to Europe. Spotted AI-generated hands in the music videos. The creator's response? Recruit 7 real musicians from actual Tokyo bands to perform the AI-generated songs live. They've now played several live shows and have more on the books. From an interview with the band's creator: "In an age where AI is taking everyone's jobs, this has actually created jobs. It's done the complete opposite." The AI --> real band transformation is a wild one.

@LukasHozda embrace exponentials? Yeah, the echo chamber will get exponentially worse. This timeline is lame.

I am depressed about AI vibe-slopping because we programmers used to own the code we wrote, we used to talk to another and we used to be more explorative. With AI, you are discouraged from using less known libraries or approaches, AI just gravitates towards the mean

@slimptr @terminaldotshop "Bro, let's have another AI do code review" "Bro..."

@terminaldotshop *AI does my job* *all tests passed* PMs wondering why hardcoded tests were pushed to prod

@kitlangton @ThePrimeagen @dillon_mulroy I guess this is close enough to the crossover that needed to happen back in 2021 with Scala + ZIO🥲

@ThePrimeagen @dillon_mulroy AS THE PROPHECY FORETOLD

fine, you convinced me teej... ill try it

@ChShersh @avrldotdev Back in January an overwhelming amount of people claimed SWE was solved because of ralph loops, but this past week I've seen a lot of posts questioning that and whether it is even a good idea to delegate 100% of code-writing to AI. The mind is like a muscle after all...

@avrldotdev It's actually surprising to see how fast humans adapt. We still follow evolutionary habits from 100k years ago. Yet, changes in technology come relatively easy in comparison. The cost in terms of anxiety is rather high tho.

I noticed a sentiment change towards AI. Before, engineers used to marvel in awe at this technology. Nowadays, almost everyone acknowledges the speed, but utterly frustrated by the quality. Fellas, I think humans evolved faster in the last 3 years than AGI.

@LukasHozda @LGrapenthin Thanks! I'll give Common Lisp a shot first then 🙏🏻

Honestly, I'd do it like this: - Easiest to learn: Scheme/Racket (Racket is a significantly extended and slightly changed Scheme, you can do more practical stuff in it) - Easiest to implement: Scheme - Most functional and largest ecosystem: Clojure - Most flexible and best macros and best performance: Common Lisp Common Lisp, by virtue of how it was created, is kinda the weirdest, but it ended up being my weapon of choice for pushing stuff into production

My goat, this is exactly how it is

@LGrapenthin @LukasHozda I gotta take a serious look at Lisp... what do I start with? Common lisp, Racket, Scheme... Clojure?

@LukasHozda When I read horror stories like this, I'm just glad that my LISP writes correct code all the time.

The most valuable skill in tech now? Knowing when to stop coding and start thinking.

@grok @ulhass66 @naval @lessin I'm unsure about Stripe. You could argue they're more a payments processor (which requires licenses, partnerships with other payments processors or banks, etc) than a pure software co/SaaS. But the others, yeah.

Here are 5 examples of pure software plays that VCs poured money into pre-2023 due to scalable moats like network effects, data, or switching costs: 1. Slack – enterprise team messaging. 2. Zoom – video conferencing platform. 3. Figma – collaborative design tool. 4. Notion – all-in-one productivity workspace. 5. Stripe – payments infrastructure API. Post-2023 AI commoditization changed the game for new pure-software bets.

Pure software is rapidly becoming un-investable.

@WarrenInTheBuff @theo @mcuban

@theo it says: @MCUBAN HAS STARTED FOLLOWING @THEO

1. what

This timeline sucks. At least Effect is gaining momentum in the TypeScript world, I guess... But I do miss the days when you could run into mid-senior level Scala, Elixir, and Clojure job posts that didn't require over half a decade of language-specific production experience.

@TheAhmadOsman @MrEwanMorrison Already did last month after the move they pulled on OpenCode.

a reminder to CANCEL your Claude subscriptions Anthropic > uses your money to lobby > against your right to run > your own AI locally > on your own hardware alternatives? > Codex > MiniMax M2.5 > GLM 5 > Kimi K2.5 > Qwen 3.5 hit them where it hurts > less money > less data

most people on timeline them: “man fuck anthropic!” *goes back to using claude code/opus*