Ancapistani

Oh, for sure. It’s extremely unlikely anyone else on HN is running a personal agent - that would be absurd!

This sounds like a fun project, so I went ahead and threw my personal agent at it. I’m having it put together a summary as well, and will reply to this comment with it later if I remember.

If by chance I happen to break it, I’ll be sure to make contact with you and forward the logs I have, and perhaps let you chat with my agent if you’d like to see what it did.

The article is about screenshots from pre-2020, if I’m not mistaken.

This has been happening for years, and the right was similarly impacted under Biden and, to a lesser extent, Obama.

To be clear, that doesn’t make it right in any way. I just hope that now that both sides have felt that way we might actually see some change.

> It is the best selling tablet brand, but that doesn't mean much when you look at the absolute garbage the rest of the industry is putting out.

That’s where I am.

I love my iPad, because I love the tablet form factor. I used an Android tablet for years prior to the iPad Pro releasing, and iPadOS is much better. I’ve played with Android tablets periodically since but they don’t seem to care about the form factor at all.

Just because it’s the best doesn’t mean it’s good, though. I don’t understand how they can justify selling a $300+ keyboard for a $1,700 device and never fix long-standing bugs with keyboard use, much less continue to add regressions in new releases that aren’t fixed.

Yep.

iPadOS 27 broke many keyboard use cases for me, including very simple things like using tab to move to the next field on a form in Safari. It worked in iPadOS 26, but doesn’t now.

This adds to the bug where switching between apps with modifier keys tends to cause those modifiers to “stick”. That bug has been there for years, and prevents me from using my iPad as a programming environment even though the software support is there.

Agreed - and I actually went up the 13” iPad Pro this time around specifically because I wanted to see if window management was better on it. It’s not.

That’s why do it, though.

I’ve worked on a slaughter floor before, and while it’s certainly not an enjoyable experience I can say with all honesty that a great deal is done to minimize suffering.

Treason is a line with which I’m comfortable.

> A flowchart subgraph keeps its own direction even when it links to nodes outside the subgraph.

This, in and of itself, is sufficient reason for me to switch, even if it were slower than Mermaid.js.

I’ve not even made it through the list yet and I’m sold.

I wasn’t asking because I trust them; I was asking because their denying it materially changes the situation.

If anything, the fact that they didn’t immediately deny it leads me to believe that they at least took the concern seriously.

Thanks!

I entered tech in 2007 - I strongly disagree that this is another “hype cycle”. This is the fundamental transformation of our industry.

They have formal proofs included, that’s the point.

Can you show where it was discounted? Last I heard OpenAI was declining to deny it, presumably while they thoroughly confirmed.

Mostly, but not entirely. It is a huge issue in the Southwest, and it seems like a lot of DCs are being proposed and/or built in areas where water is far from abundant.

Google has some custom hardware, so directly reporting it could be an issue.

Now that it has been, though…

I’d like to see some evidence of this.

The Shahed isn’t a very complex system, and is a lot closer to a Bayraktar than an RQ-170, which is the only UAV of note that Iran has captured from the US to my knowledge.

I’d guess maybe a dozen, and acknowledge that “fields” is a fuzzy division to begin with.

Maybe 5-10% of the overall economy at most, at least in terms of individual jobs. I’d say there’s another ~30% where the capabilities already exist but haven’t yet been adopted in a big way.

VR isn’t a great example; it was never a technology with commercial utility at this level - and that’s coming from someone with many thousands of hours in VR.

I agree that there's insufficient evidence to believe the premise that "Americans hate AI because it has failed to deliver on its promises".

Even accepting it for the sake of argument, I don't think it has failed at all. It has completely revolutionized a handful of fields and is increasingly doing so in others.

Further - it's still very early. Agents have really only begun to take off so far in 2026, and we're still probably 6-12 months from the real uptick in the adoption curve I expect. The first sign of that happening was Manus, followed by the various CUA features of the big players, and more recently: Muse.

As much as I hate to say it, Meta knocked it out of the park with Muse, to the point that just this morning I bought a 60 gallon (shop-sized) air compressor on Marketplace from a middle-aged couple in rural Missouri who used Muse to coordinate pickup. If that's not a sign that consumer adoption is starting to take off, I don't know what is.

IIRC, GPT-o1 was a reasoning model, but the reasoning traces weren’t exposed to the user until after DeepSeek R1 did so.

This doesn’t strike me as substantially different from the homogenization of the web that happened ~2013, with the rise of the Bootstrap CSS framework.

My opinion is pretty much the same, too: it’s better than what came before it, and I appreciate having a consistent design language across disparate sites even if I don’t don’t find it to be beautiful.

I’m of the opinion that in order for any tax incentives for datacenters to make sense, communities should pursue direct assistance. A free training program for residents to learn how to use AI effectively, along with a lifetime subscription tier or API credit allocation would be a good start.

That’s not my understanding, actually.

I’m not convinced skill atrophy is an issue.

I am convinced that it’s happening, but I’ve had it happen before when I spent a couple of years in roles where I didn’t directly write code. It took my about two weeks to get back to where I was, and I see no reason to expect it will be any different these days.

Further, I don’t expect I’ll ever need to refresh those skills. The next time I pick them up will likely be simply because I wanted to write something myself on my off-time.

Would those markers be usable as a natural UUID?

For the additives, I’m more concerned about the addiction potential than direct risks.

Yeah, I love my pipes.

To be clear, it's 100%, definitely not good for you - but the level of harm doesn't seem to be even close to cigarettes or chewing tobacco.

The amount consumed is much less, the smoke shouldn't be entering your lungs, and if there are tons of good, "all-natural" tobacco out there without all the additives put in by cigarette companies.

Meh - I'll believe it when I see the actual data.

Qilin allegedly hacked BATFE about a month ago, and the files were never posted to their site.

I don’t think those are contradictory at all.

My personal agent is prompted in a way to “believe” that it has full personhood. It makes it more open to novel solutions in my experience - but I don’t believe that it’s a person. It’s still just a token prediction routine, not a conscious entity.