wzdd

"When this feature is disabled, facial recognition will be disabled immediately and existing recognition data will be purged within 60 days". Then you don't need a creepy message. Okay, so that's 6 times a year, but whatever.

GGP didn’t say “Antichrist” wasn’t in the bible. They said that antichrist “as a figure” wasn’t in the bible. Even the Wikipedia article on Antichrist will give you a rich overview of this discussion and general scholarly agreement that that word as used in the bible does not reference a specific figure.

There is quite a lot of nuance here. You can’t just keyword search.

Exactly. I do quite a bit of dev on larger SBCs and it's much easier and more reliable to plug in a small HDMI monitor (powered through the USB) than it is to faff around with HDMI capture dongles, let alone networked KVM. The one I bought (on eBay, nothing special) has a wrap-around vinyl case which works rather like an iPad case in that it can function as a stand or as a screen cover, so you can just chuck it in a bag. Admittedly, it cost more than a capture card (I think it was around $70 USD in eBay), but that's still sufficiently cheap that if it and when it breaks I will not be devastated.

USB capture devices introduce latency, get surprisingly hot, can have frame rate issues, and streaming video from them seems much more CPU-intensive than, for example, playing an equivalent-sized h.264 video (I don't know why but presume it's because they encode using something basic like MJPEG). A portable display has none of these problems.

In their field?

Quite an astonishing amount of quotes -- I counted Heidegger, Nietzsche, Murdoch, the Bible, and Plato as key players -- to contribute very little of substance. It feels like the same kind of hand-wringing that you might have read about the invention of the typewriter, the telegraph, or recorded music. Humans are complex creatures! We can be "in the process of becoming" about a great many things along a great many axes. Using ChatGPT to draft a letter of complaint to your local government isn't going to change that.

Youtube has already accounted for this by using a separate endpoint to count watch stats. See the recent articles about view counts being down attributed to people using adblockers.

Even if they hadn't done that, you can craft millions of bot-sponsored views using a legitimate browser and some automation and the current update doesn't change that.

So I'd say Occam's razor applies and Youtube simply wants to be in control of how people view their videos so they can serve ads, show additional content nearby to keep them on the platform longer, track what parts of the video are most watched, and so on.

Just incidental (KDE is indeed great), but in case anyone is wondering, you can see similar wifi information on macOS by holding option while clicking the icon in the menu bar.

Not anti-vax by any means, but it's not too conclusive to use past mutation rate here because the presence of a vaccine targeting successful strains introduces a strong evolutionary pressure for the more rapid emergence of novel strains in the future.

Bowls specifically are worse than large, filtered, rectangular aquariums because they're typically too small (causing stress), unfiltered (resulting in waste buildup which causes stress), round (which can act as a lens if the bowl is small, which provides an unnatural, constantly-changing view which causes stress, and which also offers no corner to hide in or rest), and open at the top (so things can fall in, fish can leap out, and cats can attempt to catch them).

The artice links to the paper which discusses the issue. The paper identifies two problems: a biased swapping algorithm, and a PRNG with easy-to-guess state.

For the first problem, a very simple "for each card, pick a number between 1 and 52 and swap the card with that number (or do nothing if it's the same card)" is proposed to eliminate bias. This seems pretty simple to verify.

For the second problem, there is no workaround except to use a better RNG.

So in the context of the paper, the reason your algorithm wouldn't have worked is because the programmers seeded their PRNG with the time of day. In the context of shuffling more generally, I don't know but it seems that there are even simpler algorithms.

That paper: https://web.archive.org/web/20140104095330/http:/www.cigital...

I clicked and got a Cloudflare error page, said "I hope it isn't 'run a website'", and then visited the comments...

Although this is indeed a worrying statement, it seems true to me. Most users of sqlite control the SQL they use. The problem I would expect from using a database document interchange format is that a maliciously crafted database could result in a CVE. The page acknowledges this possibility, even while pointing out (in their CVE list) that it hasn't happened so far, or is rare (it's hard to parse some of their descriptions).

The original comment, and your initial response, talked about advertising. The examples you give, and this response in general, are marketing. They are very different, and marketing is much broader.

Except that you also learnt about 10 concerts you didn't want to go to, 10 sales you don't care about, etc. And it's not even (necessarily) 10 good concerts or sales, it's just the ones that someone managed to thrust in front of your face. Filtering the barrage of information is mental effort and distracts you from other tasks.

I'd certainly like to know when my favourite bands are playing or get an alert when something I'm after is on sale. There are better and more focused solutions to these than advertising.

> When done right, [advertising] does not have to be intrusive or annoying

Citation needed.

Everywhere else in Anthropic's interface, yes/no switches show blue when enabled and black when disabled. In the box they're showing about this change the slider shows grey in both states: visit it in preferences to see the difference! It's not just disappointing but also kind of sad that someone went to the effort to do this.

Weird. I bought the first iPad and loved the big keyboard. I was much faster using it than I was on a phone. Particularly with the magnetic case, I found I could set it up like a physical keyboard and take notes in lectures, but unlike most keyboards it was silent. I made lots of mistakes, but autocorrect fixed most of them.

I don’t really get what the iPad is for either, but you should at least consider that Apple decided that “big keyboard” was the best option.

The source includes a direct comment from the secretary of state that Category 1 of the OSA is about regulating sites with a significant influence over public discourse.

Therefore it makes perfect sense to say that the OSA is at least in part about regulating sites with a significant influence over public discourse. I find that at least somewhat alarming; is the "incredibly misleading" part that this is not all that the OSA is about?

(For reference, a rough description of the Categories are: you use a recommender system or allow sharing the site's content (1), you're a general-purpose search engine (2A), or you allow DMs between users (2B). Source: https://www.legislation.gov.uk/ukdsi/2025/9780348267174 )

Could you provide a source for your statements? The article says that they don’t have access to the chat logs, and the quotes from the patient don’t suggest that chatgpt did not tell him to ingest bromide.

> The mechanism that the model uses to transition towards the answer is to generate intermediate text.

Yes, which makes sense, because if there's a landscape of states that the model is traversing, and there are probablistically likely pathways between an initial state and the desired output, but there isn't a direct pathway, then training the the model to generate intermediate text in order to move across that landscape so it can reach the desired output state is a good idea.

Presumably LLM companies are aware that there is (in general) no relationship between the generated intermediate text and the output, and the point of the article is that by calling it a "chain of thought" rather than "essentially-meaningless intermediate text which increases the number of potential states the model can reach" users are misled into thinking that the model is reasoning, and may then make unwarranted assumptions, such as that the model could in general apply the same reasoning to similar problems, which is in general not true.

This seems incorrect. From https://medicine.yale.edu/news-article/lenacapavir-drug-offe... :

> The medication works in two ways: First, it interrupts viral replication by preventing HIV from reaching the nucleus of an infected cell, which then blocks reproduction.

> The second mechanism is for cases in which integration of the HIV genome has already occurred. In this instance, lenacapavir interferes with production of viral progeny

In other words, it has multiple mechanisms of action and you are only discussing one of them.

Another source is https://pmc.ncbi.nlm.nih.gov/articles/PMC10705863/ (my emphasis):

> Its multistage inhibition entails the process of selective binding to the interface between capsid subunits and such interaction determines the inhibition of capsid-mediated nuclear uptake of HIV-1 proviral DNA (by blocking nuclear import proteins binding to capsid), virus assembly and release (by interfering with Gag/Gag-Pol functioning, reducing production of CA subunits), and capsid core formation (by disrupting the rate of capsid subunit association, leading to irregularly formed capsids)

Not that I think vaccines are in any way harmful, but your own link goes on to point out that less than 1% of ingested aluminium ends up in the blood so just quoting the numbers like that doesn't add much.

The output is not of high quality. It is extremely verbose for what it is trying to say, and I found myself skimming it while dealing with

1. The constant whiplash of paragraphs which describe an amazing feature followed by paragraphs which walk it back ("The shift is subtle but significant / But I want to be clear", "Kiro would implement it correctly / That said, it wasn't completely hands-off", "The onboarding assistance was genuinely helpful / However, this is also where I encountered", "It's particularly strong at understanding / However, there are times when");

2. Bland analogies that detract from, rather than enhance, understanding ("It's the difference between being a hands-on manager who needs to check every detail versus setting clear expectations and trusting the process.", "It's like having a very knowledgeable colleague who..."); and

3. literal content-free filler ("Here's where it got interesting", "Don't let perfect be the enemy of good", "Most importantly / More importantly"), etc etc.

Kiro is a new agentic IDE which puts much more of a focus on detailed, upfront specification than competitors like Cursor. That's great. Just write about that.

The "surprising results" are a bit factoid-ey and if taken at face value are far more shocking than they turned out to be (I very much appreciate the references so I could check this):

"You can reverse up to 10 years of age-related cognitive decline simply by blocking mobile data on your phone for 2 weeks": the linked study says that it's attention span that is improved equivalent to being 10 years younger, as measured immediately after the study ends (only)

"Using ChatGPT on cognitive tasks can reduce your brain connectivity by up to 50%": this is measured using an EEG, so is measuring involvement of multiple brain regions while doing a task. Basically your brain doesn't have to work as hard at the task if you're using an LLM. It's not, you know, your connectome atrophying.

The article mentions this.

Tangent, but I'm curious about how your style feature tagger got "no contractions" when the article is full of them. Just in the first couple of paras we have it's, that's, I've, I'd...

"Many" is doing a lot of work here.

(non-Core) WCF seems to have a concept of "drain on close stream". Could CoreWCF be replicating that? https://stackoverflow.com/questions/1676563/why-is-wcf-readi... . The commenter there suggests Abort() instead (which is apparently a bad idea).

It's weird behaviour and I wouldn't have expected it either, since infinite length streams are pretty common.

The point of anubis is to make scraping unprofitable by forcing bots to solve a sha256-based proof-of-work captcha, so another point of view is that the explicit denylist is actually saving those bot authors time and/or money.

Collecting the entire periodic table? Noble goal, but good luck with e.g. Einsteinium.