HNHacker News
TopNewBestAskShowJobs

throw10920

4,183 karma · joined October 15, 2021

put your email in your HN profile!

quiet.can8525@fastmail.com

Marketers: you do NOT have permission to add this email to any databases or use for any advertising and solicitation whatsoever - personal correspondence only

submissionscomments
throw10920··on GLM-5.3-Flash
> Uh, actively trying to hack an embedded device that runs Linux over the network, specifically an IP security camera, could be considered red teaming, no?

No. Vendors (and their model guardrails) do, indeed, treat those as separate from pentesting non-embedded infrastructure, and that is because they are very different activities.

And, if you actually read my comment, it says "red team my own computers". That's categorically different from pentesting an IP camera.

> You never mentioned which exact activities you were getting flagged on and getting refused.

Because further details than those I've provided aren't relevant, and it's clearly different from what you're doing.

Your experience isn't relevant to my situation.

throw10920··on GLM-5.3-Flash
> Maybe the system prompt you're injecting is making it refuse?

No, this has nothing to do with my harness. I use one of the most popular open-source harnesses available.

> I found that GLM-5.2 was pretty happy helping me reverse engineer/hack devices.

This is a completely different category of things than what I'm getting refusals on, so I'm not sure why you're bringing it up.

throw10920··on GLM-5.3-Flash
> I want to use these models to red team my own computers.

Exactly what I was trying to use it for! ):

I'm in the same boat - I haven't heard of a way to get around it aside from either self-hosting (GLM-5.2? good luck) or "self-hosting" (paying bucks per hour to Vast) an abliterated model.

throw10920··on GLM-5.3-Flash
And there's a reason for that: the US government does not compel model trainers to train their models to paint them in a favorable light, while the PRC does.
throw10920··on GLM-5.3-Flash
> In fact, I don't think I've ever even had a prompt refused.

I very much have. I've gotten GLM-5.2 refusals for extremely benign security testing on my own infrastructure of the same flavor that people were getting (wrongly) flagged for on Fable during the initial release.

throw10920··on Field measurements of neighborhood-scale air temperature impacts of data centers
> Primary production of the planet, i.e. total carbon fixation, is broadly not influenced in this manner.

Plants account for roughly half of global carbon fixation. "Broadly not influenced" is straight-up false. Your follow-on arguments are consequently false as a result.

> My point is that it is tiny in scale even at best.

Such as this one.

throw10920··on Field measurements of neighborhood-scale air temperature impacts of data centers
> No, it inherently incorporates all natural processes that capture carbon.

It literally cannot, because it's not parameterized on the amount of additional plants being planted.

> Artificial, sure, if you want to point at them I believe current effots might reduce the required reductions from 99.9% to 99.8%.

Goalpost moving (i.e. fixing a particular value in hypothetical future assumptions that was not fixed or agreed-upon previously in order to support an otherwise-false point), which solidly sets the original claim as false.

> However, I wrote "emissions", and the overwhelming majority of CCUS systems are reditecting what would otherwise be emissions directly to capture, i.e. what I wrote would include that as reducing "emissions".

Technically compatible with your statement but not how the vast majority of individuals would interpret it as you originally wrote it.

To be clear: I'm not arguing that we don't need to substantially reduce greenhouse gas emissions or that carbon capture is necessary. I'm arguing that last point, that "we need to reduce emissions by 99.9%" is extremely misleading and much more harmful than hurtful for those ends.

throw10920··on Field measurements of neighborhood-scale air temperature impacts of data centers
> These bad faith "but you exist and do unrelated things" fallacies are a.) bad faith b.) make people have ai people more.

> Bringing uo AC while you defend datacenters making global warming worst is just a cherry on top.of it. Just underlines anti-human fck you all ideology of it all.

You are badly breaking the HN guidelines in a dozen different ways. You should refamiliarize yourself with them:

https://news.ycombinator.com/newsguidelines.html

throw10920··on Field measurements of neighborhood-scale air temperature impacts of data centers
> It is based on the mean lifetime of a CO2 pulse being order-of a thousand years, which is itself approximation of what IPCC says; actual decay rate behavious is not even as simple as exponential decay.

> At 1000 years exactly, steady-state requires 99.9% reduction in the long term.

So, that's ignoring carbon capture by means natural or artificial, and therefore is a totally irrelevant number, both to this conversation, and to most others.

throw10920··on Universal health coverage could save $1T and 114k lives a year: study
> Basically the only organization that's motivated to keep costs as low as possible is the government.

Unfortunately, this isn't true for the government either.

Source: personal experience of me and literally dozens of people that I know. I've briefly worked with my country's government, and in that time I personally experienced and got dozens of stories along the lines of "the government spent tens of thousands of dollars of aggregate government employees' time because a single employee booked a hotel that was less than one dollar above the approved rate while traveling".

There's a good reason for this, of course: bureaucracies' policies are mostly "scar tissue" from high-profile cases where a bad actor did something they shouldn't have but wasn't specifically against policy, and then a policy was written for that case and stands for the rest of time. And, bureaucracies are risk-averse, especially democratic governments, whose leaders are elected based on optics almost as much as policy.

But it doesn't change the facts. Not only are large bureaucracies inefficient, but government bureaucracies specifically are incentivized to optimize for optics and structure rather than improving that inefficiency, and anyone who has actually worked for a large government can tell you that.

throw10920··on Field measurements of neighborhood-scale air temperature impacts of data centers
> We (humanity worldwide) need to reduce greenhouse gas emissions by 99.9%.

Citation needed.

throw10920··on Memory prices climb 500% in 12 months
That was 24 years ago. Half the people commenting on this board aren't even that old. How many of the shareholders on the boards or executives at the helms of those companies do you think are still there?
throw10920··on Cursor launches Origin, GitHub alternative
> We don't get to use technology for technologies sake, and pretend it doesn't have social and political consequences or connections.

Nobody is """pretending""" that in this thread.

The reality that is being asserted is that (1) politics doesn't have to be on-topic for every forum, (2) it's very explicitly not the focus of HN, and (3) even if it were on-topic the guidelines are very clear about ways to engage in a civil fashion and a very large number of comments on Musk-related posts are extremely clearly not doing this and breaking the guidelines.

Hence, downvoting and flagging are the correct things to do.

throw10920··on Cursor launches Origin, GitHub alternative
Just downvote and flag and move on. "Who hurt you" just encourages bad actors to behave worse (and isn't particularly good for itself). Repeatedly flag-killing their comments tends to teach a better lesson about what conduct is acceptable.
throw10920··on Cursor launches Origin, GitHub alternative
I'm going through and flagging the guideline-breaking comments (e.g. https://news.ycombinator.com/item?id=49348639) - I'd encourage you to do that, too, to help preserve the quality of HN.
throw10920··on Israel creates fake think tank in likely attempt to dupe AI chatbots
You're welcome! There's nothing quite like finding the phrase that labels a concept you've been aware of for a long time.
throw10920··on Israel creates fake think tank in likely attempt to dupe AI chatbots
Use-mention distinction is lost on a lot of people. It's a useful test for whether someone is reacting to a word (e.g. "damn") out of pure emotion and conditioning, or whether they're acting rationally.
throw10920··on Why Target Common Lisp for Code Generation?
I don't understand. Are you saying that you think that writing Common Lisp is akin to using a debugger in other programming languages?
throw10920··on Grok 4.6 scores 61 on the Artificial Analysis Intelligence Index
> As are you! Your comment was snarky and mean. Clear violations.

This is factually incorrect. I did not say anything "snarky" or "mean". I stated facts and pointed out that the poster was breaking the guidelines.

throw10920··on Grok 4.6 scores 61 on the Artificial Analysis Intelligence Index
> So you’re asserting that the ends justify the means or?

No, they're not. You're making up things and pretending that they said them.

> Please do better.

You're badly breaking the HN guidelines. Please review them: https://news.ycombinator.com/newsguidelines.html

throw10920··on Grok 4.6 scores 61 on the Artificial Analysis Intelligence Index
> xAI sub: $30 when other starts at $20, pro like sub for $300 where other charge $200.

This is meaningless because you're not controlling for amount of subsidized usage or model quality.

throw10920··on Grok 4.6
I like Grok, but I don't think that it's quite Fable-tier. It's good, but I think the position that it occupies on the Pareto frontier is a little more toward the "cheap" side and a little less toward the "intelligence" side.
throw10920··on DeepSeek V4 Pro 0813
You're doing great. Don't let insanely low-effort (negative-effort, as in making others dumber rather than having no effect?) comments like from the above throwaway affect your actions.
throw10920··on Grok 4.6
> Anyone else find it weird how within 2 months of Fable releasing all the major labs suddenly had Fable-level models?

Frontier model release cycles generally take around 6-8 months anyway. OpenAI and xAI (or however you spell it, branding almost as bad as X/itter) were probably working on their next generation of models already, and Anthropic just beat them 2 months to this release.

You also say "near-concurrent release of the same jump" - but 2 months isn't "near-concurrent", it's a full quarter of the normal release cycle.

I don't think that the other explanations you gave are implausible, though - for both human circulation and distillation, you can apply those during a training and development run (with reduced effectiveness). Reasonable to imagine those as bumping them up another few points to bring competitors from "a little below Fable" to "around Fable".

throw10920··on I made tinnitus my friend, then it disappeared [video]
> Turns out the process to straighten your neck - even after like three decades - with just some casual invest everyday only takes about a month or two

Could you provide some detail? Even YouTube links would be helpful.

throw10920··on Os8088: A powerful Mac-like OS for the IBM XT, 286, 386
OK, if you're so confident, surely you've gathered data that empirically shows that "virtually everyone on HN is using AI to write code and simultaneously dismisses all interesting new software as written by AI", because that's a very easy claim to prove or disprove.

Can we see your results and methodology?

throw10920··on US Military's cyber command unit grapples with cluster of deaths by suicide
> “Women belong in the kitchen”

They never said anything remotely like that in their comment. You're actively lying about their words in addition to badly breaking the HN guidelines. You should re-read them:

https://news.ycombinator.com/newsguidelines.html

throw10920··on DeepSeek V4 Flash 0731
> This kind of takes makes me cringe. Why don't you go back to LinkedIn?

This is not appropriate for HN. Please review the guidelines: https://news.ycombinator.com/newsguidelines.html

throw10920··on Show HN: I spent 2 years designing a mechanical Magic Keyboard
> I think the point is that it isn’t really a legitimate criticism of the device

Tacking on emotional manipulation does not somehow make your point (more) valid.

> Not having a fingerprint sensor was not a design choice by the keyboard maker. It is a constraint of the operating system that peripheral makers have no ability to work around.

What? Numerous other sibling posts have pointed out that this is not only possible, but it's already been done. Claiming it's "a constraint of the operating system" is false.

https://news.ycombinator.com/item?id=49199063

throw10920··on UEFA and its national associations will not participate in FIFA competitions
> Whataboutism is a word invented to deflect any criticism of western hypocrisy.

Factually false. It's a logical fallacy, because you're attempting to manipulate the audience without actually responding to the claim.

> Like antisemitism, its overuse has rendered it toothless.

Also factually false. "Antisemitism" has no agreed-upon definition. "Whataboutism" has a very concrete and well agreed-upon definition: deflecting a an argument by responding with a counter-accusation that is intended to mirror the perceived accusation.

Your whole comment as manipulative, though, so it's not surprising that you would get the definition wrong.

← PreviousPage 2 of 34Next →