@NotesFromComei
iAccount based inFrance
About this account
- Account based in
- France
- Connected via
- United States App Store
Account-level information from X, not a live location or the device used for a specific post.
Exploring new ventures (@esquisse_camera and more). Prev: https://nitter.cf/t.co/dH5RWGv2iq & https://nitter.cf/t.co/LFI3YSMpUw (acquired).
Joined February 2011
- Tweets9.3K
- Following483
- Followers1.9K
- Likes39.3K
I doubt reality will be as perfect as this simulation, but the direction is obvious: stealth is becoming less and less relevant.
Warfare is increasingly about saturation. What matters is not whether you can hide one high-end platform, but whether you can overwhelm the enemy’s defenses faster than they can detect, track and kill your attackers. We will eventually be able to detect stealth aircraft and drones and the cat and mouse game probably won’t hold for long.
Ironically, we may end up going back to tried-and-tested methods. Cheap, non-stealthy attack vectors. Lots of them. And cheap, vision-based defense systems to kill them.
Every country pouring billions into sixth-generation fighters should be thinking about that. Why accept internal weapons bays, compromised aerodynamics, limited payload and all the other constraints of low-observable design if the enemy can see you anyway?
Warfare is becoming K-shaped too: insanely expensive, high-tech systems for deterrence at one end, and huge numbers of cheap, relatively simple attackers and defenders at the other.
Basically, I've been iterating slowly on an AF/tracking system for @esquisse_camera these past few months but Opus 5.5 helps me get so much done.
This is for now simulated data and within a tight budget. Next step is deployment on Raspberry Pi (hopefully next week).
I have a weird theory that we will soon realize ai has been here for longer than we thought. It’s now ubiquitous and anything is analyzed as "it could be ai" but it actually dates back from a few years ago
When 4.5 came out I thought it would take a few decades for anything to be possible for $200/mo.
When Fable 5 came out, I thought it would be sooner than that.
Since Astra / Opus 5.5 it's my operating default. I assume I can build anything.
There is a huge debate in France around whether the new hot novel of the rentrée, C’était ça ou Mourir, by Thélyson Orelien, was written with AI or not.
Of course, it’s been excluded from most literary prizes. There are also allegations of plagiarism, which is much less excusable.
But what I find interesting is how the French intelligentsia keeps repeating that they don’t believe it was written with AI.
A few interesting learnings from this:
1. Elites are AI-blind (or dumb). Anyone who reads a lot of ChatGPT will immediately recognise the patterns. And yes, it’s relatively trivial to tell whether a text has been written with AI or not.
2. It got politicized extremely quickly, and the debate completely missed the point. The left immediately defended the author by saying he was being attacked because he’s Haitian. Which is not the point. Excellent thread by @Hazukashi1 here:
nitter.cf/Hazukashi1/status/2103…
3. Most importantly: everyone is debating whether AI was used, instead of asking the much more interesting question: what if the book is actually good?
If AI can write this well, why couldn’t it win literary prizes? Maybe the future of writing is prompting. After all, who would say using a pencil is cheating because you can erase what you wrote?
Until a few months ago, the general consensus was that AI would basically be a super automator. And it is. But it’s also far, far more than that: it’s coming for everything.
We are going to be cats for AI agents. Everybody will be happy. AIs will take care of us, and we will be fed doing nothing.
My favorite way to learn is to ask ChatGPT to write or read me a tailored explanation of whatever subject I want to get proficient in.
It kinda feels like The Matrix. I can just download whatever I want into my brain.
I started studying electrical engineering for the robot I'm building. It's the first time since I dropped out a year ago that I've been in the process of deeply understanding a field I heard about in high school. It's really, really another world to have AI explaining all the physics behind this to me with patience and custom pedagogy. I feel that it must be so cool to study in school nowadays with this level of personalization. It also reminds me that project-based learning should be the default way of teaching.
Gamers have a massive advantage in today’s economy. As it’s now clear that we live in a simulation, they are the ones who are the most used to it.
Côme retweeted
In less than 24 months there will be a clip of a high profile CRO explaining this reality to a (much smaller) room of sales reps.
Côme retweeted
Fin septembre 1976 - tel un savant fou, Giorgio Moroder compose I Feel Love.
La matrice a cinquante ans.
Il enchaînera avec la BO de Midnight Express où la basse continue de précéder la mélodie.
One thing driving enshittification is Instagram. People pay fortunes for the right photo, just to show off their lifestyle.
They don’t care if the food is bad, it just has to look good. They don’t care about the toiletries, as long as there’s a big patio where they can pose for an iPhone portrait.
Asia, especially Japan, Korea and parts of China, still seems to care about delivering genuinely good service. But in Europe and the US, more and more places feel built for Instagram rather than for the person actually paying.
The good news is that none of this is necessary to make lifelong memories. The right song in a rental car at dusk, or cheap pasta in the sun with zero pretension, can be just as good.
What is the point of luxury hotels in Europe?
I had lunch at the Mandarin Oriental Conservatorium in Amsterdam
The staff did not speak Dutch
The food was sub-tier
The service was slow
The vanity had wood rot
The bathrooms stank
Toilet lids were loose
Yet rooms start at ~€1,200 a night
This is enshittification: paying for nothing but a brand name