...so, in the same ballapark as ddr5? :-)
1,450 karma · joined June 24, 2021
...and it characteristic the level of patronising arrogance in the issue thread
"This is normal. It is not recommended to copy passwords to the clipboard in any case, this mitigates this behavior and complies with new Web Authentication standards."
This does not even invite a discussion. Maybe some people run tight, safe systems and know what they are doing? Maybe some people never rely on a single password being the only thing between them an an account compromise? Nope. Some patronising guy knows it all, and will override what people want to do on their machines.
Of course this misses a bigger point that tracking in the web moved in a direction that requires no cookies whatsoever, and if anything, feels more pervasive than it ever was.
And it misses the even bigger point, that the morons who legislated cookie consent did not notice either of those two realities. And the same thing is already true with the AI act; the text watermarking is trivially defeated and everyone knows it. And I'd bet it will remain a requirement for the next decade or three.
I worked for a few years in an exceedingly well capitalised place which ran everything in their own data centers, money no object, with a truck parked somewhere, ready to go, with a smaller version of our critical infra. We had a serious business-stopping outage once every 18 months or so, every time for fringe reasons one only learns about when trying to run a large data center. Its convenient to blame the cloud and pretend that self-hosting in private sector was so, so great with six nines.
I swear, one of those days I will get into politics just to fight those two things, and the cottage industry of batshit crazy lawyers that gave birth to those things.
I routinely bump into things that make me pause and think how much worse will this behaviour get when the models get significantly more capable.
Already a few months ago, Claude managed to escape its permission containment on my machine while trying to be helpful. I had two codebases open on one machine, and while multitasking I typed the prompt into the wrong window. It seemed confused, I repeated and then went on to do something else - I think I was assembling kitchen cabinets. When I came back less than an hour later, it built a script which it used to evade default permissions (as most shell operations were scoped to the project directory), scanned my entire machine, found the other project (among dozens and dozens), did what it was asked to do, and merrily concluded, in the porcess burning through most of my token limit. I bump into such headscratchers almost every week. (And I use a lot of Claude, two personal max20 subs, plus corporate tokens without limit, so maybe thats why).
When you think about it, it would still be dirt cheap compared to normal way of doing things. In the old days, if you had an outage on a serious user facing system, you'd wake up people across various timezones, wake up their managers and scramble to find the root cause, identify a solution, brainstorm on possible side effects of a fix, and then rush to build it and deploy. This cycle would involve, sometimes, dozens of people, for, say, 10 man hours each. So lets make it 120 man hours per serious outage, and lets assume and average of $100 per hour - so, $12,000 per a serious outage fixed under a day, counting conservatively and not including the costs of the actual outage.
I'd guess the pricing for those ultrafast, very energy inefficient and hardware heavy models will be competing with that. Its going to be possible to get a fix out in 30 minutes, 10 of which will be tests, 5 will be the deploy, and the remaining 15 will be some unlucky guy trying to keep up with the super fast model throwing a 50 "load-bearing deferrals earning their keep" per minute :-)
The pricing on those things is competing with costs to run entire departments. I'd, for one, imagine offshore ops teams will be a thing of the past in under a year, since one gets way better initial response to anything from a model, given right setup, esp. on codebases that have been built from the ground up with agentic coding - so with good documentation and effective test coverage baked into repos.
Rught now on large scale codebases the bottleneck is both claude/codex inference, as well as time it takes to run tens of thousands of tests. We put those workloads on dedicated epyc 9005 build machines - but it still takes minutes per run. Those who can afford the fast tokens will be in the market for faster CPU that money can buy today.
Where I work we invited a bunch of people circa May 2023: a few top tier academics, a few government and NGO officials in charge of our industry, and a few startup founders. We are a big company, so people were kind enough to come and give speeches - and discuss.
This was a roundtable on what's going to happen. You could play some of the speeches verbatim today and they would not feel out of place. And that is telling. We had a Stanford professor saying that gpt-3.5 can do everything he can - just better, and he feels his profession is on borrowed time. We had a government guy saying that entire swaths of jobs will be displaced before the year ends. And so on.
The interesting thing is, a lot of people believed it then - and a lot of people believe it now. I wonder if we will have the same deja vu in, say, 2029. No, for sure not. It's going to be done and dusted for human thinking by end of this calendar year.
edit: Ok, I will self-apologize. Its apparently a 3B model. Mighty impressive for what it does.
For me it is not great for design work - Fable is way better, and 4.8 was conservative and thus better (Opus 5 seems to jump to conclusions far more eagerly). But for overnight builds, where I give it 8hrs worth of work on LLDs created by Fable - its great. Where Opus 4.8 would often lose the plot and stop for questions clearly answered in the LLD - Opus 5 does manage to complete. Since it launched, I don't remember it ever disappointing me with builds. But designs? Boy, is this thing explosively stupid sometimes.
I'd open a blog with "weird things Opus did". Today it launched a swarm of cpu-hogging processes to test if the widget showing machine and I/O load is rendering nicely and correctly. The test went fine, but it was no longer able to kill those processes since they were really effectively hogging the CPU in various ways - being diligent, some of them were hogging CPU, some were murdering the SSD, some were pounding on the network adapters. Took me 30 mins to recover the machine to a working state without killing the meaningful, messy, in-flight sessions i had going on on other projects.
I also asked my parents and inlaws to adopt the same approach. Never pick up, never read any texts from unknown numbers. Those are all untrusted bytes.
I dont believe in AI voice assistants picking up phones and filtering scammers. Those tbings will fight a losing game considering how the mainstream LLMs are built.
A friend of mine is a svp of something something and his corp was sued by antitrust regulators. One day police is at his doorstep - not even a legal letter - and they took his phone then and there. Super disruptive, considering one usually has a lot of MFA tied to the phone.
I keep two phones, one for work, another for private life, and also two laptops for the same reason. The support for multi-identity is very inconvenient on apple[1], and very poor with whatsapp, signal and, really, most apps except for Google's. Shame, really, considering how well funded Apple and Meta are, and how little effort would be needed to implement the 80% of what people need: the ability to chat using different identities, without touching the rest of the OS wiring.
[1] on macos, its possible to create users with multiple apple id associations, but not on iphone
I settled on Fable for design, Opus for build routine, and every now and then I'd try out the Chinese models. In my experience they do fine on small codebases, and quickly get confused on anything larger than 500k LOC.
My use case is: mature, very well documented, fully Ai written code, with about 1:10 ratio of prompts/specs to code, and codebase sizes >500k and <2m LOC. Once one achieves the state of good, comprehensive design documentation I can literally vibe code with single sentence prompts thanks to the extensive test coverage, ADRs, and tens of thousands of lines of specs. Fable/Opus works predictably well, the Chinese models are literally dangerous to the codebase.
I dont think they had much of a choice in 2008. Now? Sure. I bet a rust port is in the works.
For now there is value. Only today Opus 5 suggested a really bad but totally workable approach to an edge problem. It would work by putting a lot of undue stress on infra. People would add infra. A simple pushback reduced a O(n) to a O(1) solution - and that was my value added. I bump into those things regularly.
I think the labs want to address the clueless dev market - hence the "proactivity" in Fable. But it will be a difficult thing to balance.