Do not go gentle into that good night

If you cannot bring yourself to say "I was wrong" to a model, you are not fit to be a senior engineer in 2026

Not a writer, but something feels off about describing mass homicide as "social tension".

Mediocrity does not need a villain. It only needs sustained inattention.

RT @ShriramKMurthi: Startled to find out that there are young people who haven't read James Iry's magnificent "A Brief, Incomplete, and Mos…

If somebody thinks LLMs are conscious, I would expect them being capable of having deep conversations about the nature of consciousness, which is hardly ever the case.

You cannot outsource your understanding even if you outsource your thinking to Andrej Karpathy.

from the tree, difficult to say. from the parent? definitely not.

"ex machina" logical fallacy: generated output used in place of an argument

You do not find important work by browsing options; you find it by pulling on a thread that keeps bothering you.

I strongly believe there are entire companies right now under heavy AI psychosis and its impossible to have rational conversations about it with them. I can't name any specific people because they include personal friends I deeply respect, but I worry about how this plays out. I lived through the great MTBF vs MTTR (mean-time-between-failure vs. mean-time-to-recovery) reckoning of infrastructure during the transition to cloud and cloud automation. All those arguments are rearing their ugly heads again but now its... the whole software development industry (maybe the whole world, really). It's frightening, because the psychosis folks operate under an almost absolute "MTTR is all you need" mentality: "its fine to ship bugs because the agents will fix them so quickly and at a scale humans can't do!" We learned in infrastructure that MTTR is great but you can't yeet resilient systems entirely. The main issue is I don't even know how to bring this up to people I know personally, because bringing this topic up leads to immediately dismissals like "no no, it has full test coverage" or "bug reports are going down" or something, which just don't paint the whole picture. We already learned this lesson once in infrastructure: you can automate yourself into a very resilient catastrophe machine. Systems can appear healthy by local metrics while globally becoming incomprehensible. Bug reports can go down while latent risk explodes. Test coverage can rise while semantic understanding falls. Changes happens so fast that nobody notices the underlying architecture decaying. I worry.

Writing has always been hard, but AI fixed this. Now reading is hard.

Imagine a Lego constructor that you needed to emotionally manipulate so that the blocks stuck together. That'd be psychotic, and still...

I have been enjoying programming so far precisely because the computer followed my instructions without me having to scare it into compliance.

It helps to remember that training data today includes a range of human idiotic / emotional, and maybe even psychotic responses. Keep saying "please", "thank you", and praise and encourage the model to trigger more constructive responses. Manipulating feelings of a simulated character is a novel professional competency.

What decides where our attention goes?

Not fatigue. Not frustration. Just the bone-deep weariness of a soul that has swallowed one too many italicized revelations — drowning in the wreckage of sentences that bleed too loudly, that perform their wounds for an audience that never asked to witness.

Sad reality: AI commoditized eloquence. Good analogy and dramatic language used to require actual skill. Example: "A isn't just B, it's C". Now anyone can. Use it and most people (including myself!) will see it as AI slop because they very often indeed are. Eloquence is dead.