Wow, that X post is the most succinct but persuasive and damning thing I've read about Google's ad business developments in years (I don't watch it particularly closely.) Thanks!
2,034 karma · joined January 30, 2017
Fun fact: Irix 5.3 in Nov 1994 was the first desktop OS to allow web URLs to be desktop shortcuts, pre-Windows95+MacOS similar capability. SGI licensed Netscape Mosaic so they could do that.
1. We, Oceana, are at "war" with Eurasia. They're the worst! Our leaders have made friends with Eastasia. We can do great things together with Eastasia! Eastasia is great! <<Time passes.>> We won! Eurasia is dead!
2. We feel threatened by Eastasia! Eastasia has been stealing from us! We must prepare for battle with Eastasia! Eurasia is great aren't they! Eurasia always been great! We've always been at war with Eastasia!
3. ???
4. Profit!
(apologies to 1984 + slashdot)
The author doesn't discuss kanban which is less overhead, but then makes collaboration acrossultiple scrum teams a bit trickier actually.
Presumably it's for medical research but that doesn't mean it similarly gets bought/sold for another purpose, sooner or later and possibly in bankruptcy like 23andme as mentioned by the parent poster who cared enough that I figured I'd warn them about less intuitive similar risks.
I think I have started to see similar language at doctors office paperwork recently but maybe I am misremembering. If you care about not getting sequenced, you have to read the fine print nowdays.
Or am I just a product to them?
This is in notable contrast to the "PC revolution" in the late 70s/early 80s, the marketing was all about equipping people, empowering individuals. It was a "personal" computer after all, not a big mainframe owned by corporate... designed to make you more powerful.
That messaging is wildly absent from the EA/SV-centric LLM/AI boom.
Instead it's about replacing you, the singularity, inevitability, fear-mongering, etc. And there is much more of a flavor of information asymmetry to the business model because of the Internet connectivity and SaaS structure to it all designed to enable the fastest possible feedback loops for the inventors combined with a bunch of pinkie promises with no actual controls possible on the user side at least for the frontier large models. What exactly did OpenAI peek at when doing Navier-Stokes and do they even know or would you trust them to tell you? You are right, it is a very different vibe.
Making an English language description of a problem and proof steps match the Lean proof is apparently harder than the halting problem and is not something that a Lean proof alone or the current dump or even the Navier-Stokes dump solves.
A good breakdown of the problem is here: https://arxiv.org/html/2610.08144v1 or also see here: https://terrytao.wordpress.com/2026/10/09/what-mathematician...
However the following articles' descriptions of drug treatments using alpha-agonists, such as clonidine and guanfacine, which provide a response rate of 55% (significant relief for 55% of people) does imply to me that there is some physiological basis and may be of interest to people reading this thread: https://neurodivergentinsights.com/how-to-deal-with-rejectio... https://neurodivergentinsights.com/history-of-rejection-sens... https://www.goldstarrehab.com/parent-resources/is-rejection-...
And I think there are stronger findings that autism/ADHD is at least partially genetic.
P.S. I would caution you against holding a premise of a hard "or", that mental conditions are either hardware or software, or the hidden premise that you can't patch hardware with software.
"Only measure that which you are comfortable turning into a target."
This is an interesting thought but I disagree and certainly logically wouldn't call it a corollary.
As someone managing a system, one approach is to have a balanced set of measurements and to hold some of them back so they can't be gamed by those being measured. I.e. have more measurememts than targets so you can detect alignment drift.
It's pretty distressing to watch the desperation and the struggle of a baby screaming and a mom trying but unable to feed her baby.
Then the quote sort of turns from clever into a sign of a bit of real-life ignorance.
That said, I remember trying to weigh the hype at the time of the announcement reading/skimming the papers Anthropic published, recognizing that bugcount alone wasn't super-relevant but also remember being impressed by an NFS bug and a kernel bug that struck me as relevant at the time. So where did that NFS issue show up in GKH's list you showed so nicely above?
It turns out, AFAICT, it's not on his list, but the reasons are perhaps interesting to others so I will post here. It turns out there were two NFS issues this past year conflated a bit in my memory:
* The Linux CVE-2026-31402 NFS heap overflow that could allow unauthenticated memory reads over the network isn't in that list of 79, presumably because it was found by Claude Code, not Mythos months earlier. (I am guessing it's not his "malicious network packet into the middle of the stack" and is a stronger attack being a remote attack.)
* And the CVE-2026-4747 NFS stack buffer overflow that allowed gaining full unauthenticated remote root access didn't show up in GKH's list of 79 because despite being Mythos-caught, it wasn't Linux, it was FreeBSD.
I guess this does match my memory now that I think about it, that there weren't any smoking Linux guns caught by Mythos.
* (I guess there was also a longstanding 27-year old OpenBSD TCP SACK-handling stack integer overflow than enabled remote crashes / Denial of Service found by Mythos.)
There is definitely Mythos hype, but just because it hit the BSD code base more than the GKH-managed Linux code base doesn't mean it was inappropriate to raise eyebrows from Mythos, in particular since "attacks only get better".
If I download a bunch of music from a torrent without a license, even if I don't listen to it, I'm liable, but if OpenAI or other LLMs gets content by some other unlicensed means (Anna's Archive, t), they are somehow not liable for the copy they made however temporary (but not so temporary if they leave it around to train a second model)?
And/or derivative works? And/or contributory copyright infringement when they regurgitate that copyrighted text when given certain prompts?
I get there is some nuance to copyright law, (four prongs), some utility to the outcome, and some legal (but not plausible) deniability. But there is no way they have clean hands on the copyright front for at least some actions they have taken. If there were, it would be in all their marketing and they would be pushing regulators to bind their competitors, onshore or offshore, more tightly in this regard.
I was around when search engines took advantage of similar ambiguities in copyright that took many many years to get litigated for similar reasons.
Protecting against an AI convincing the human to let it out of the box is an interesting challenge. Even a dual-keyed system to do so only requires convincing two people, although I suppose more elaborate containment mechanisms can be devised. N-keys is democracy and that doesn't seem immune.
Two random thoughts:
* This box we want to keep AIs in reminds me a bit of the story of Pandora's box
* I'm also reminded of the original alignment problem in Genesis 3 where one creature convinces another to take an unaligned action.
Containment is a pretty fundamental tricky security problem actually once persuasion is part of the threat model.
I'd only add one more observation.
Sometimes the root cause is poor table design (or in analytic/OLAP use cases poor ETL design without proper data validation checks or handling) where uniqueness is not enforced and that is the root cause that should be fixed if at all possible. A "first normal form" violation in the database design so to speak.
If that root cause is not addressed, then SELECT DISTINCT is more often necessary and the SELECT DISTINCT disease to be safe culture and behavior in the code base on top of the database just spreads.
The rise of the Internet made channel sales handoffs a weak point in the business model. I bet this missed sale went to a reseller who didn't respond and Sun didn't have any controls or visibility on that failure.
Misdiagnosis from a therapist is not new, but AI use by the therapist is.
People have had itching ears to hear what they want to hear for a long time, and recently we have all seen in the press stories of AI-only therapy sycophanticly reinforcing delusions, but it is so strange to see a professional therapist human-washing such stuff.
But it makes sense; therapists are not yet savvy to all the weaknesses of LLM analysis and framing bias and they too can be tempted themselves to ask an AI and just pass outputs upwards like various work colleagues I've had...
Still, I was quite disturbed about the unprofessionalism of the therapist when I heard about it. I think we'll see this come up more in the press. like we've seen with lawyers. But not as soon because lawyers have a validation function called a judge.
Although the more paranoid AWS customers who turned on (and pay for) S3 cross region replication or similar cross region DR for other services would be fine.
I did it with production Scala apps over a decade ago. I even built sproc TDD test suites.
They've gotten a little better in recent years.
The lost business value and competitiveness caused by outsourcing IT overseas to unmotivated parties under Carr's premise is hard to put your finger on but I have seen the aftermath and it's pretty massive.
Just think, isn't the whole point of a company, of incorporating, is that the liability shifts from the employee to the firm?
It is 100% fair to hold the company responsible, and doubly so when the questionable thing the employee is doing is something that A) benefits the company, and B) is on company time (aand with company resources.) I know you weren't making a legal-minutia point but even the high level principles involved are and should be the opposite of what you suggest.
Companies may want to have their cake and eat it too (have employees do their dirty work but then push the blame on the employee) but that's not how it works and it's generally not a society you'd want to live in the more we head in that direction.