sarker
ketman hetman
No bio...
User ID: 636
sarker
ketman hetman
12hr ago
Billionaires and spooks are the ones to worry about with AI takeover, they have their hands on the buttons.
I'm informed that they already have their hands on the buttons. Certainly the middle aged women are not the ones with their fingers on the buttons.
An economic system that produces and requires a bunch of middle-aged women in office jobs, a bunch of teachers and bureaucrats and officials and journalists
In what sense are they required?
sarker
ketman hetman
16hr ago
There aren’t any examples of people building “Disneyland” for people unrelated to them, particularly people who do nothing to benefit those paying for it.
You are aware of the sums being spent on the homeless in West Coast cities, right?
sarker
ketman hetman
18hr ago
High income people are for the most part the upper middle class.
It's intuitively bizarre to say that people who out-earn 99% of people are part of the middle class. Data bears out our intuitions - half of the top 1% of wealth holders are in the top 1% by income, with the rest of the top 1% of income earners being made of the next 9% of wealth holders. You're welcome to define the top 10% of the country as the "upper middle class" but this is pretty nonsensical.
sarker
ketman hetman
21hr ago
The status quo ante is impossible because we destroyed their military and nuclear program.
Yeah, I'd say it's a little early to call this one. Casualties appear to amount to 5% of the IRGC strength. 5% is not nothing, but I wouldn't call it destruction. The status of the nuclear program is simply unclear at this point.
sarker
ketman hetman
21hr ago
Indeed, the Ayatollah is dead, however it's not clear that merely replacing Ayatollah Khamenei with Ayatollah Khamenei counts as a success. Or at least, I have never heard anyone describe such a thing as a success scenario before the war required it.
sarker
ketman hetman
21hr ago
You'll notice that the comment about the war being a success if Iran has no military was made after I had responded to the original comment that boils down to "the status quo ante would be a win"
sarker
ketman hetman
21hr ago
The top 1% of income earned pay 40% of income tax which goes to fund Medicare, Medicaid, social security, and EBT. I doubt that Al Capone spent as much of his income on soup kitchens as top 1% income earners spend on those things. Yet this is not enough.
sarker
ketman hetman
22hr ago
Because prosecuting a war at some cost of blood and treasure only to end up with the status quo ante is a failure.
sarker
ketman hetman
22hr ago
I find it a little amusing that most of these would constitute things being worse than the status quo ante. It seems that if tomorrow Donald announces that the war is over and everyone goes home and we end up where we were before all this mess except Iran spent some missiles and replaced the Ayatollah you'd consider this a smashing success.
sarker
ketman hetman
22hr ago
Like most Shagbark posts, this is post-hoc cope to justify what he was going to do anyway. Really, Shaggy, women don't like wagies? But if they did, you'd become one? It's too late, he has already depicted me as the screaming soyjak and himself as the handsome Aryan Chad.
The reality is that among my coworkers even the very physically, shall we say, ungifted don't seem to have problems finding wives and having one or two kids.
I can't speak to the ease of running a business, but 20% of small businesses close within a year. Presumably most of those are run by idiots, but the odds still don't seem great.
sarker
ketman hetman
23hr ago
Next I can say “nothing can ever convince me”
Is this the case, then?
Is it possible for you to admit that the war is going well?
Sure. For example:
- Iran hands over the uranium and pinkie promises not to make a nuke
- the Ayatollah steps down and the Iranians make a new thank you USA
- Iran renounces support for the Houthis, Hamas, and Hezbollah in word and in deed
Those are the kind of things that would make me say the war was a success. Now it's your turn!
sarker
ketman hetman
1d ago
what would it take for you to admit this war was a mistake?
Why won't you answer this simple question?
It has now been 29 days since you said that the war will basically be wrapped up in three weeks.
Is it even possible for you to admit that the war was a mistake, or do you believe that everything "achieved" so far has already made it worth it regardless of how it ends? If so, you should just say so, so that we can all update accordingly.
sarker
ketman hetman
1d ago
Perhaps, but there's plenty of deserts where people historically didn't circumcise, and the Polynesians circumcised despite no lack of water (at least, water good enough for washing sand).
The huge numbers you see are annualized projections that aren't representative of any actual revenue.
Of course they are representative of some actual revenue. It's an annualized extrapolation of the past month of revenue. I don't deny that it doesn't mean that they made that $14B in the past year (but neither is anyone claiming it is), but it's not a random number as you seem to be suggesting. Huge growth in monthly income is real growth.
sarker
ketman hetman
2d ago
This would be interesting if the primary purpose of LLMs was performing well on benchmarks. The benchmark is a measure, which may be flawed for various reasons. I think everyone who isn't a grifter understands this.
In the real world, I've never heard of anyone who says a model is good because it scores well on benchmarks, or choose one model over another due to its performance on benchmarks. From Zvi:
Benchmarks have never been less useful for telling us which models are best.
They are good for giving a general sense of the landscape. They definitely paint a picture. But if you’re comparing top models, like GPT-5.4 against Opus 4.6 against Gemini 3.1 Pro, you have to use the models, talk to the models, get reports from those who have and form a gestalt. The reports will contradict each other and you have to work through that. There’s no other way.
sarker
ketman hetman
2d ago
To be clear, you can use Claude code with the $20 a month plan. You can even use something like opencode and a cheaper model paid by the token on openrouter. Or you can even run a local model like Gemma for ~free once you have the hardware.
Despite the name, it isn't really coding specific. I presume "Claude cowork" is largely the same as Claude code but with a name that doesn't scare the hos.
I do recognize the frustration that Claude can't look up the API details through the web interface. Perhaps there's some security considerations and they didn't want to have the . model call arbitrary endpoints.
As far as profitability goes, we shall only know when these companies IPO or go bust, but Anthropic revenue growth is massive.
sarker
ketman hetman
2d ago
After running for tens of minutes and spitting out a bunch of technical data about the API, it gave me this message:
This is hilarious, but a useful illustration of how people come to think of LLMs as useless. Claude in an agentic harness (e.g. Claude code) easily researches new APIs and figures out how to get information out of them. For example, I was trying to get historical heat index information and CC presented several possible data sources, I asked if it could use a different one, and it looked at it and figured out how to call the API and extract the relevant info.
sarker
ketman hetman
2d ago
More hires should be able to get that task done in a set amount of time,
Similarly to how more women should be able to birth a child in a set amount of time.
sarker
ketman hetman
2d ago
They're pretty rare even in my area. I think I might only have seen one in my life.
sarker
ketman hetman
2d ago
How old is the startup? How much funding have they gotten? Most startup equity is worth $0 in the final analysis.
sarker
ketman hetman
2d ago
They did pay for themselves, they paid taxes for 40-50 years (some began working at 15 and didn't stop till 65!) to "buy into" the system.
No, they paid for the people receiving benefits when they were working. They didn't pay for themselves. Now workers are paying for them.
sarker
ketman hetman
3d ago
It's like you didn't even bother to read the user's manual that's on the pole.
Not saying it would or wouldn't work well for this, but even Opus 4.6 is $5 per million input tokens. You probably wouldn't want to feed in all the transcripts at once though.
More options
Context Copy link