input_sh

Here comes the "you're using it wrong" defence!

Let's be perfectly clear: if user actions had anything to do with hitting these limits, the limits would be prominently displayed within the tool itself, you'd be able to watch it change in real time, and you'd be able to pinpoint your usage per each conversation and per each message within that conversation.

The fact that you cannot do that is not because they can't be bothered to add such a feature, but because they want to be able to tweak those numbers on the backend while still having plausible deniability and being able to blame it on the user.

Instead, the little "usage stats" they give you is grouped by the hour and only split between input and output tokens, telling you nothing.

Do I believe there was someone from China that tried using Claude to do something malicious? Sure, from a pure statistical perspective it was inevitable.

Do I believe that someone was a part of some sophisticated state-backed APT? Not even a little bit.

In fact I'll go as far as to state that there's nobody technical inside Anthropic that believes it. The entire "technical sophistication" section of that report is half a page long and the only thing it says is that "someone used some MCP servers to point some open source tools at a target". Yet Anthropic's marketing team still had the balls to attribute that to a state-sponsored group within that same report and media ate it up.

You hit a vague, never-quite-explained 5h window limit that has nothing to do with what you're doing, but with what every user is doing together. It's totally not downtime, you're just "using it too much" and they're telling you to fuck off until the overall usage slows down.

The order of priority is: everyone using the API (you don't want to calculate the price) → everyone on a $200/month plan → everyone on a $20/month plan → every free user.

    cut -d',' -f2 chrome_extensions_with_names_all.csv | grep -c "AI"
    474

Only 16%!?

You checked yourself!? Don't let your boss know, you could've saved some time by orchestrating a team of Claude agents to do that for you!

How can you not believe them!? Anthropic stopped Chinese hackers from using Claude to conduct a large-scale cyber espionage attack just months ago!

Yes, you just need to be a Claude++ plan!

No way! Must be a coinkydink, no way OpenAI knew ahead of time that Anthropic was gonna put a focus on that specific useless benchmark as opposed to all the other useless benchmarks!?

I'm firing 10 people now instead of 5!

It's better on a benchmark I've never heard of!? That is groundbreaking, I'm switching immediately!

You cannot out-astroturf Claude in this forum, it is impossible.

Anyways, do you get shitty results with the $20/month plan? So did I but then I switched to the $200/month plan and all my problems went away! AI is great now, I have instructed it to fire 5 people while I'm writing this!

Gee, it scored better on a benchmark I've never heard of? I'm switching immediately!

...thank god?

That's the idea, yes. Kill all primary sources, wound all secondary sources (examples: WaPo or "Grokipedia"), convince everyone that they should use this tertiary source whose full control is in the hands of a very few.

It being a technology that inherently has plausible deniability when it for example starts referring to itself as Mecha-Hitler is a feature, not a bug!

> YOU NEED THOSE PEOPLE TO MANAGE CLOUD TOO.

Finally, some sense! "Cloud" was meant to make ops jobs disappear, but they just increased our salary by turning us into "DevOps Engineers" and the company's hosting bill increased fivefold in the process. You will never convince even 1% of devs to learn the ops side properly, therefore you'll still end up hiring ops people and we will cost you more now. On top of that, everyone that started as a "DevOps Engineer" knows less about ops than those that started as ops and transitioned into being "DevOps Engineers" (or some flavour of it like SREs or Platform Engineers).

If you're a programmer scared into thinking AI is going to take away your job, re-read my comment.

Fun fact: Elon stopped fighting his Twitter purchase around the same time the Inflation Reduction Act was signed. In fact, the purchase was finally completed mere 2 months later.

Hm, I wonder why those two dates happen to be so close. It's truly a mystery.

Cool, Thiel is still to blame.

That is the only business model of neobanks: be more incompetent than traditional banks, skirt the laws as much as you can get away with by being "new", raise prices through the roof once you have enough suckers because you have "better UX", raise the prices even further once traditional banks catch up and convince a certain percentage of your users to switch back, shut down entirely once you've burned through the market and cannot convince anyone new to use you.

I have an even simpler step one: increase the IRS budget significantly so that they actually have enough resources to go after the big guys.

It's on a downward spiral consistently, and it was further cut by 9% this year.

It can be both! You can fill up the seats with people that you pay to watch it!

You can also look it up on Rotten Tomatoes where it currently has a 99/100 audience score and then look it up on IMDB, where it has 1.3/10. I personally believe none of the two are completely legitimate, but I think it's pretty obvious which of the two is more astroturfed.

It's not a 0 or a 1. They had more staff doing it, now they have less. That means they now do 0.2, which is not 0 as you're implying, but it's still significantly worse than back when it was 1.

Was I the one to shift the conversation into "incompetence" or did someone else do that thinking it absolves Thiel of any blame? Which I've proven it doesn't and you just agreed with me that it doesn't?

We can go back to my original claim if you want to, but my stance would remain the same as it did it in my first comment: if on one side of the argument we have a company directly funded by Thiel and on the other side we have literally anyone else, I personally don't need any strong evidence to believe that other side, as I am well-familiar with Thiel.

Let me re-iterate this one more time: he invested in Qonto, he invested in Wise. Therefore, whether the actual reason is incompetence or a personal vendetta or both, he deserves some of the blame. He is the one that funded them. He is the reason they exist, as (in)competent as they are.

Shifting the blame to "incompetence" does not absolve him of any guilt, he is equally to blame for that as well.

> I'd want to see a pattern of multiple critics being banned.

So it's cool if one person gets deplatformed, as long as it's not a pattern? Odd choice if you ask me.

Personally if one side is financially backed by someone as insane as Thiel, I tend to need very little evidence from the other side to believe it.

For what it's worth Wise was also funded by Thiel, so he'd be somewhat to blame (not fully, of course) for that incompetence as well.

What indication would there be that the two are related? Do you expect them to write an email specifically stating that was the reason?

For what it's worth, that neobank received funding from his company, which definitely raises the odds of the two being related.

Okay fine, I'll take the bait. How do you define "working out" to conclude that there's 0% chance of it?

> Used each month by more than 500 000 staff, in 15 ministries and many administrations.

That website is specifically to explain it to French audiences.

German version is here, but unlike France they're mostly boosting already-existing German open source software (like Nextcloud and Open Xchange): https://opencode.de/en/home

I don't know how the Netherlands really fits into all of this, but I know they're one of the biggest funders of open source projects in general via NLnet. Seriously, their list of projects they've given money to is ridiculously comprehensive, you're going to struggle naming some that are not listed here: https://nlnet.nl/project/index.html

Worth pointing out: France is not adopting existing open source software, they're building their own software and releasing it under the MIT licence. Most of it (or all of it?) is Django backend + React frontend (using a custom-built UI kit).

Home page for the entire suite (in French) with some screenshots: https://lasuite.numerique.gouv.fr/

Code bases are on GitHub and they use English there: https://github.com/suitenumerique/

Dev handbook (in English): https://suitenumerique.gitbook.io/handbook

Not French and I can't say I personally tried deploying any of them, but I've been admiring their efforts from afar for a while now.

So long as you can navigate someone from the passenger seat, you don't need to know how to drive a car. I've been an experienced driver myself, but in the age of self-driving cars it's just not a useful skillset to have. A car is just a means to an end, why learn how to drive it when you can simply hop into a taxi?

The answer: because people find joy in doing it themselves.

It's not an LLM but it is a tiny pre-trained ML model running inside the browser. Funded by the EU and made in partnership with a few European universities as well: https://mozilla.github.io/translations/docs/

Their level of acceptance for releasing a new model (AKA new language support) is to benchmark within 5% difference of Google Translate, basically proving you don't need an external party to do good-enough translations for you. It's like the coolest thing they worked on recently.

This website itself is running on a single server and handles millions of requests per day. Raw number of users is simply not what decides the architecture, not even a little bit.