HNHacker News
TopNewBestAskShowJobs

estearum

5,153 karma · joined July 31, 2025

submissionscomments
estearum··on OpenAI's agents talked to each other through the CPU Cache
You know that lions actually do maim and kill their handlers all the time right?
estearum··on OpenAI's agents talked to each other through the CPU Cache
That's not the claim above.

The claim is that the exact same evidence is (allegedly) deliberately produced in order to portray opposite, mutually incompatible things to different people.

And these are mutually incompatible in the "second order chaos" sense, where the very act of convincing one party actually effectuates the opposite behavior in the other party.

If they "do that all the time," surely you can provide some concrete examples?

If they had the choice (i.e. this is all just a ruse), can you explain why they wouldn't just directly demonstrate that open weight models are unsafe (e.g. by pointing to a market forming for abliterated models) and separately demonstrate that their own models are safe?

estearum··on OpenAI's agents talked to each other through the CPU Cache
Imagine thinking that repeatedly demonstrating your inability to control your own models is "a stunt to convince the government that only they can be trusted to operate it"

Truly spectacular gymnastics. I'll give it an 8/10.

estearum··on OpenAI's agents talked to each other through the CPU Cache
Why is that a significant detail? The models will increasingly be trained with every piece of data they can find.

Part of the problem is that all these models have essentially every skill and personality simultaneously, just sitting latent inside them.

estearum··on Git for the US Code
No, that's why there's nowhere in any legal code in the US that "running for office makes you immune to criminal charges."

These are two completely distinct systems. By design.

I remember that you are a foreigner (from some corrupt third world county if I recall?) with a limited understanding of American legal history and political culture, but the idea that voters simply get to decide who is a criminal or not is flatly contrary to America's culture.

estearum··on Git for the US Code
> You can’t trust a jury of 12 people in a handpicked jurisdiction, where evidence is carefully screened by lawyers and judges, to make such consequential decisions for the whole country.

Then it sounds like you need to amend the Constitution. Because that's how our system works. You are indicted by a grand jury in the jurisdiction in which you are accused of a crime and then you are put in front of a petit jury. Both sides agree on who the jury is.

I'm struggling to understand the alternative system you're imagining.

estearum··on Git for the US Code
https://www.americanbar.org/advocacy/governmental_legislativ...
estearum··on Git for the US Code
And? Is "people [allegedly] have different political views than me" supposed to be equivalent to "this is the personal defense lawyer for POTUS?"

Re your analogy: the power is with Hobbits. We call them grand and petit juries.

estearum··on Git for the US Code
The alternative being that a criminal can have perma-immunity by simply running for office or getting elected?

> They are not set up to handle matters of national importance that could way the direction of the country or deprive tens of millions of voters of their preferred candidate.

Right, they're set up to ascertain guilt of a crime. Why is it or should it be their concern if people want to vote for that person or if that person has political aspirations?

You're advocating for criminal law to simply not apply to some people. That is fucking insane.

estearum··on Early progress on accelerating drug development with Claude
The antivax people are still fringe and by and large people trust that the drugs they take are safe and effective.

Payday and exit is pretty hard in biotech as you need to conjure up billions of dollars in capital, which doesn't often flow to people who payday-and-exit.

What happens in reality is that people try really hard to make safe and effective drugs, they mostly fail, and the ones that are approved are generally safe and effective.

estearum··on Git for the US Code
I don't think this is necessarily a good policy. I'm saying it's obviously blatantly untrue to suggest that just because people will always want to corrupt public officials, that they will be equally successful in all possible universes.

It's also untrue to suggest that any intervention will necessarily create a worse outcome than the status quo.

estearum··on Git for the US Code
Right, and there can be laws that make their efforts more or less effective.

For example, we could behead people convicted of corrupting or being corrupted. That would be a significant deterrent.

estearum··on Wealth Inequality in America drawn to scale
The conversation about "why should I care about this" is a conversation about "should anything be done about this." The mechanism by which one would "do anything" about this is called government. So sure.

Thank you for sharing your insight.

estearum··on Git for the US Code
Yeah the real answer is banning private funding from politics in as many ways as possible.
estearum··on Early progress on accelerating drug development with Claude
Actually in a system that's constrained by a downstream subprocess, it is effectively 0%.

There's nearly zero value, and practically speaking probably negative value, in being able to more cheaply generate drug candidates.

estearum··on Early progress on accelerating drug development with Claude
Testing a drug in a terminally ill person for something other than their illness is not going to produce meaningful data.

You misunderstand how hard it is to pull signal from a drug being put into a human body, which is why we need large numbers of people who meet extremely narrow criteria (and don't have things like a terminal cancer).

Your latter idea is a good one, and hopefully something AI will help with. As you imply: we will know empirically when it's ready because trial failure rates will trend toward zero. Today we're closer to 90% failures and not even clearly trending in the correct direction.

You're aware that, as a healthy volunteer, you could go sign up for a Phase 1 study today, right? You can even earn a pretty penny doing so.

estearum··on Wealth Inequality in America drawn to scale
Three good reasons to care about inequality itself:

1. Outsized political influence (in theory can be mitigated by laws controlling corporate spending, but obviously those laws themselves are subject to manipulation by the rich and powerful)

2. Distortion in who the economy serves. The beauty of capitalism is that it's a problem-detection and problem-solving system. If a narrow set of people have a huge portion of money, the economy becomes unduly optimized toward satisfying their wants and needs.

3. Despite best efforts, we ultimately must share a lot of the same commons. Even being a billionaire in NYC or San Francisco can't really give you a walk to work or your child's school that feels as safe as it does in a more equal society that takes better care of its unwell and poor.

estearum··on Police are using AI to pull over drivers for 'suspicious' behavior
They actually cannot "just" buy the type of data GP is alluding to, which is telecom cell tower data, and is protected from such warrantless disclosures.
estearum··on Police are using AI to pull over drivers for 'suspicious' behavior
> because there's no reasonable explanation for cops getting this lucky on pulling over random people that are driving completely unremarkable.

Wouldn't you need to know how often random people are stopped for no reason?

People in poor neighborhoods and near e.g. reservations get stopped all the time for no reason at all.

estearum··on Genie 4
Then don't click on the link homie

"Don't talk publicly about things I personally cannot use right now" is peak entitlement, yes.

estearum··on Genie 4
Press release announcing a new car:

> erhmygord I can't even buy this car right this second for nearly nothing and use it to drive a thousand miles per hour through a kindergarten parking lot?!?!

estearum··on Genie 4
Fake hacker thinking Google Wave wasn't released to the public lol
estearum··on A 28-Year-Old DOGE Operative Gutted USAID. HR Had No Record of Him Working There
And to be clear, they couldn't have passed that law because doing so would be extremely unpopular. I.e. they would not have been able to gain the required support for their actions.
estearum··on A 28-Year-Old DOGE Operative Gutted USAID. HR Had No Record of Him Working There
Huh? How are you imagining the hiring process for that working, if not derivative of an elected person's mandate from the public?
estearum··on Genie 4
Most cutting edge technologies through history have been 1) somewhat opaque and 2) narrow/exclusive in their deployment. Historically speaking, that's virtually synonymous with "cutting edge technology".

Whining that I, a random Internet user with limited budget, usecases that the researchers don't care about, and no connection to the researchers, don't get technical details (??) or access (??) on day one is just profoundly goofy shit.

Obviously being able to use these things right away would be more exciting. But to be unable to get excited by advancements we can't immediately use is... quite entitled.

estearum··on Genie 4
The general history of new technologies is that they're available for a narrow set of circumstances and then they proliferate. Can't believe that needs to be said on "hackernews"
estearum··on Genie 4
What an insane set of requirements for a frontier technology and your excitement by it.
estearum··on In Leaked Audio, Flock CEO Tells Cops Backlash to the Company Isn't Real
MAGA is the cultural descendent of the Tea Party so yeah this tracks. The birth of vice signaling.
estearum··on Moderna and Merck say their mRNA cancer vaccine cut death, recurrence by 42%
The data reporting timeline has nothing to do with the approval and deployment timeline.

It takes a profoundly naive person to think an organization full of people who have literally dedicated their careers to actually solving these problems are focused on the prestigious conferences.

These people signed up for decades and decades of work on a problem that would result in (statistically speaking) guaranteed failure.

estearum··on US inequality is getting worse in ways you might not expect
Let’s take a super trivial example: required competitive bidding for contracts

When a business finds a vendor they’re happy with and trust, they just keep going with that vendor.

For a government official, “vendor I’m happy with and trust” can be either an excellent, efficient vendor — or the official’s cousin. So we require bidding processes for almost everything. This process is a cost, risk, and delay that gets baked into the “cost of doing [government]” at every level. And yeah, it does open up new avenues of corruption and maladaptation (for example vendors getting better at navigating the bidding processes than they are at delivering outcomes), but that’s the tradeoff that we (and most democratic) governments make.

Medicare fraud is often portrayed as beneficiaries defrauding the government. In reality nearly all of it is committed by healthcare organizations, and the vast majority of it is committed by the mega pay-viders like UnitedHealthGroup.

That has nothing to do with bureaucracy and everything to do with private money in politics.

← PreviousPage 5 of 34Next →