cldellow

I think I see where you're coming from, but to me, the analogy falls short.

I think it's because email has the postal system metaphor to work with: I put my letter in an envelope, the post office just forwards it. Similarly, Hotmail does not _need_ knowledge of the contents of my email to forward it to its destination. If you asked a user who had access to their email, they might reasonably expect only the recipient would.

With LLMs, it is quite clear to the user that you are interacting with _something_ that is inspecting your messages and reacting to them. If you asked the user who had access to their inputs, they'd probably say "Claude" or "Anthropic". Thus, when Claude rats them out for threatening violence, they probably wouldn't be that surprised.

To me, this feels much closer to someone phoning up Microsoft's customer support phone number and mentioning that they have a gun and know where Bill Gates lives, and that then resulting in a visit from the police. I do idly wonder if it was intentional that this user had this conversation with Claude, and not, say, with ChatGPT or Gemini.

No, I don't want to live in that world, and I would definitely read an article discussing how to ensure we don't reach that world.

That said, I think that world is quite a few steps away from the scenario being discussed, which is "when talking to X, X might report my conversation to the police".

You comment elsewhere that this might be a have vs have-not divide -- if you can afford to self-host a powerful LLM, you will have more liberties.

I think this isn't that trenchant of an observation for two reasons: (1) the wealthy have always had an advantage in exercising their rights over the poor, but more importantly, (2) the capability vs cost ratio of self-hosted LLMs continues to improve at an insanely rapid pace.

In my view, LLM companies are in a tough spot here.

I grew up in Dawson Creek, BC. An hour's drive away is Tumbler Ridge. In February 2026, someone did a mass shooting, killing 6 people and injuring 2 dozen more.

The shooter had previously discussed this with ChatGPT, but OpenAI did not report it to the authorities at the time.

In hindsight, that was an error.

Should the LLM companies be the ones determining if threats are real, or should they just default to letting the police investigate?

I'm fine with the latter. Again, I absolutely believe that these companies have immense power to do evil, I just don't think the linked article is strong evidence of it.

I would personally be fine with Anthropic disclosing the chat logs, but I can see why they'd prefer not to. No doubt that would have resulted in another complaint about them building a panopticon and sharing people's conversations verbatim with the authorities.

The person doesn't seem to deny having written the messages, or claim that they were flagrantly taken out of context. His response to the media was that he was "just fucking around" and that he didn't want to talk about it.

It's also worth mentioning that this happened in August 2026, after the February 2026 Tumbler Ridge BC shooting. The shooter in that scenario had discussed planning gun violence with ChatGPT, but ChatGPT didn't report it to the authorities, so no follow up happened. Presumably the LLM companies have decided to more aggressively report these things to the police as a result.

The editorializing in this article is quite something.

If I strip out the editorializing, I think the conclusions are:

1. Someone told Claude they had an AR-15 and the CEO of Anthropic was "in their sights". Anthropic reported this to the police.

2. Anthropic contracts with a security agency.

3. Anthropic is looking to build in-house capability for security.

Anthropic clearly has high-value targets (see, e.g. the person with the AR-15 from point 1). When I worked at Microsoft in the 2000s, they also had in-house security to investigate threats against facilities and executives.

I would believe Anthropic is doing evil things, but I don't think this article actually makes that claim, despite the bombast about "pre-crime".

According to https://www.taylorpress.net/article/10664,blueprint-data-cen..., the restriction on the deed was gone by the time the property was transferred to the city of Taylor. That seems to have been in 2003.

Is the idea that "when it benefits them" was... 23 years ago, and then they just sat on the land waiting for big tech to come along and want to buy it?

As mundane as it may sound, it seems most likely this was a clerical error made a long time ago. Maybe it can get unwound, but maybe not. If the people of this town are being screwed, it's by incompetence on someone's part 23 years ago, not by big tech.

https://www.taylorpress.net/article/10705,judge-rules-in-fav... has a bit more info.

I don't know how to pull the actual court documents without paying for them, but the article indicates the case was dismissed for lack of standing.

The plaintiffs tried to argue that as neighbours, they had an interest in the land usage being enforced. The court disagreed.

I presume the original family could bring a case? It doesn't seem like the 404 article or the Taylor Press article talked to them to see how _they_ feel about how their gift is being used.

Fair enough - my initial comment talked about quality, but I realized that was my own take on the situation.

I had revised the comment because I think I now understand Drew's chief complaint to be about the moral side of LLM usage, not the practical quality side of LLM usage.

He does use the word "slop" which implicates quality, but that's a single word in his essay, versus whole paragraphs about the moral questions of LLM usage and his stated reason that the fork was "to keep my conscience clear".

Buried (IMO) in the post is:

> sadly even Vim now comes under scrutiny in that effort as both Vim and NeoVim are relying on LLMs to develop the software.

...where he links to a comment in a closed issue where someone accuses a contributor of using an LLM to generate patches: https://github.com/vim/vim/issues/18800#issuecomment-3568099...

The tl;dr: Drew thinks Vim development has been tainted by LLM contributions, and is thus morally unsuitable to be used, and he will therefore be forking it.

Ha, what a coincidence. Just today I was reading a three year old Stackoverflow discussion about this [1].

It prompted Laurenz to submit the documentation patch that is cited in the article. In the discussion of the patch itself, people seem to conclude that it's a good improvement to the docs, but that the behaviour itself is a bit of a footgun. [2]

[1]: https://stackoverflow.com/questions/73951604/autovacuum-and-...

[2]: https://www.postgresql.org/message-id/Y8cQJIMFAe7QT73/%40mom...

> If he's correct

He's not.

According to [1], the White House claims Vietnam has a 90% tariff rate.

According to [2], 90.4% is the ratio of Vietnam's trade deficit with the US -- they have a deficit of $123.5B on $136.6B of exports.

The same math holds true for other countries, e.g. Japan's claimed 46% tariff rate is their deficit of $68.5B on $148.2B of exports. The EU's claimed 39% tariff rate is their deficit of $235.6B on $605.8B of exports.

Who knows, maaaaybe it just so happens that these countries magically have tariff rates that match the ratio of their trade deficits.

Or maybe, the reason Vietnam doesn't buy a lot of US stuff is because they're poor. The reason they sell the US a bunch of stuff is because their labour is cheap to Americans. (They do have tariffs, but they're nowhere near 90%: [3].)

America's government is not trustworthy. Assuming that what they say is truthful is a poor use of time.

[1]: https://x.com/WhiteHouse/status/1907533090559324204/photo/1

[2]: https://ustr.gov/countries-regions/southeast-asia-pacific/vi...

[3]: https://www.investmentmonitor.ai/news/vietnam-gives-us-tax-b...

Search engine rankings, both live and historical, have value to marketers.

Google makes it challenging to reliably crawl their rankings at scale, so there is a real cost to collecting this data.

As a result, there aren't good, open, public archives of rankings. There are paid services like Semrush, though.

I'm Canadian.

We definitely had a dim view of the Iraq invasion. At the time, I ran a Microsoft recruiting event on campus at University of Waterloo. Students made some widget using whatever framework du jour was popular in order to win a prize. The top-voted widget was one that showed how many Iraqi civilians had been killed by the US so far during the war. Still, we sort of understood: y'all had a legitimate grievance due to 9/11 (just, not, y'know, against Iraq...)

We had a dim view of the first Trump administration. Muslim ban? Trump's anti-vax horseshit during a pandemic? January 6?

And then Americans re-elected him. Whether it's the Nazi stuff, the tariff stuff, the annexation stuff, the trans stuff, the firing the blacks and women stuff, it's just exhausting chaos that was all predictable.

Now we have a dim view of Americans.

There is a benefit, in that I'm hopeful this will be the impetus for Canada to become a more serious country. Still, it's incredibly wasteful. If we retool to trade more with Europe and Asia, both Canada and the US will be poorer for it. But god, it will be worth it.

Yeah, the author seems to be writing for a white male audience in some regards.

My wife and I host bicycle tourers when they pass through our town. One was Thomas Meixner, an East German who started travelling the world on bike when the wall fell. He's visited something like 120 countries and biked 250,000 km.

My wife asked him if he thought a solo woman could do what he did in the places he did it. He tactfully changed the subject.

It seems to be pointing at something attempting to visualize JIRA activity, but with a lot of content-security-policy warnings?

Seems like someone screwed up some load balancer settings.

Canada has begun cracking down on this. There are now caps on the number of visas issued. Foreign students are no longer allowed to work full-time jobs by default while studying. This is especially relevant, because the Comprehensive Ranking System for permanent immigration gives points for years of full-time work experience in Canada, as well as for diplomas earned in Canada.

Unwinding it will be a bit messy: lots of post-secondary institutions have to figure out how run programs with a lot less funding, and what to do with capital projects that no longer make sense in light of greatly reduced enrollment, etc.

The damage done to the average Canadian's view of immigrants will take some time to fade away. But I suspect it will, with time, especially since our traditional immigration system really does just skim the "good" immigrants -- the ones with money and the skills to succeed.

The Lieutenant Governor of Texas, one month into the pandemic, was mulling how it might be best for society if we just carried on as-is and accept that a lot of old people would die:

> As a senior citizen, are you willing to take a chance on your survival, in exchange for keeping the America that all America loves for your children and grandchildren? And if that's the exchange, I'm all in.

The chief science advisor of the UK described his meetings with the PM of the UK in 2020:

> “He says his party ‘thinks the whole thing is pathetic and Covid is just nature’s way of dealing with old people – and I am not entirely sure I disagree with them. A lot of moderate people think it is a bit too much.’”

> Vallance’s diary also recounts how then chief whip Mark Spencer told a cabinet meeting in December 2020 that “we should let the old people get it and protect others”. He said that Johnson then added: “A lot of my backbenchers think that and I must say I agree with them”.

FWIW, I live in the region and disagree with OP's characterization of "serious decline" and "most people have left".

I went to school here from 2003-2008, moved away and moved back in 2011.

The area's population has increased by ~20% since 2012 (~the death of RIM, according to its stock price). In 2011, it got regional train service to Toronto. In 2019, it got a local light rail train.

The university area that the OP seems to be referring to is, IMO, more walkable and bikeable now than before. Some of the towers are mixed use, with ground floor retail.

The city is definitely quite different from the early 2000s, though.

Thanks, that's a handy video! Yes, mbtiles is based on SQLite, I was imprecise in my language.

When I said RDBMS, I meant those that have a client/server model. The versatiles docs talk about the complexity and surface area of database systems as a motivator for creating their own container format. From this I inferred they were referring to Postgres and PostGIS, which are used in the canonical OpenMapTiles implementation.

Watching that video, they do mention not liking the traditional Postgres/PostGIS approach due to its heavy weight. But they also say they disliked mbtiles due to its SQLite dependency, and that the versatiles format is inspired on/based on pmtiles. (Apologies if I'm missing nuance here, I was watching auto translated auto generated captions.)

I found https://github.com/versatiles-org/versatiles-rs/issues/24 which contrasts the versatiles format vs the pmtiles format. After reading it, I'm not personally convinced of the benefits of versatiles vs just throwing a CDN in front of a clustered pmtiles file, but perhaps I'm missing something.

It's a bit complicated, because I think the versatiles brand is used to describe multiple things:

- the schema of the map: what objects are available in each tile at different zoom levels. It sounds like versatiles uses the shortbread schema (contrast vs OpenMapTiles, protomaps)

- a container format: a way to pack multiple tiles into a single file. It sounds like they created their own format here (contrast vs mbtiles, pmtiles).

- the scripts/tooling to build everything

- the overall finished map product itself (contrast vs Google Maps, Stadio Maps, protomaps, OpenMapTiles, etc)

The versatile container format seems to require a custom HTTP server. But if you want, you could produce the versatiles map and store it in a pmtiles container. Or you could stick a caching proxy in front of their publicly available tile server at https://tiles.versatiles.org/tiles/osm/{z}/{x}/{y}

It would be interesting to hear them describe why they decided to create their own container format. The text that I have found seems to be contrasting it to RDBMS containers, but is silent about mbtiles/pmtiles.

Oof, just above that entry is this one:

> That may be awkward for someone who DM’d and will have a later last day (could be next year, even), but my priority is people staying, not people leaving. I’m hearing from people staying that they’re stretched thin right now, so we need to hire and ease that overload. It’s not a time for a bunch of people to leave at once.

Am I reading that right? Someone tried to take the voluntary layoff package, but Matt reneged on the terms, so now that person gets the joy of working under a cloud until Matt decides to honour the deal? It's small potatoes in light of the other stuff, but yikes.

I keep seeing people refer to this tweet. Can you share a link to it, please?

I don't use WP Engine or WordPress, so I don't have a side in this fight.

As an outsider, that context seems a bit dubious to me.

@photomatt has tweeted [5135]: "[...] Please let me know if any employee faces firing or retaliation for speaking up about their company's participation (or lack thereof) in WordPress. We'll make sure it's a big public deal and that you get support. [...]"

If this was true, I would think that @photomatt's twitter feed would be loudly boosting this disgruntled employee's story of WP Engine-imposed limits and subsequent retaliation. Yet @photomatt's twitter feeds seems silent to me. This makes me skeptical of this context.

[5135]: https://x.com/photomatt/status/1836862087320195174

Me and two of my siblings have had marriages to facilitate work visas, permanent residence and access to health care during a pregnancy. (Also because we liked the other person, but like, we would have kept liking them regardless of whether we got married.)

Most pregnancies take a while, two weeks is plenty of time. Don't even get me started on how long immigration bureaucracy takes -- two weeks is a drop in the bucket.

In Ontario, the Compulsory Automobile Insurance Act [1] requires that all owners/lessees of vehicles take out insurance policies. The Insurance Act [2] sets out that the minimum amount should be $200,000.

$200,000 is a much better floor than, for example, Ohio's $25,000. An Ohioan friend was injured by a motorist who had the minimum coverage. Her care cost more than that. The motorist who caused the injuries didn't have a lot of assets and she was unable to recover the excess from the motorist.

Still, there are some perhaps unintended downsides. Canadian rental car companies, as the owners of the vehicles, are obliged to provide $200,000 insurance as part of every contract. As a result, it seems there's not much market for them to sell excess liability insurance, and none do that I'm aware of. I, as someone who has plenty of assets to lose if I injured someone, would happily buy a higher liability insurance. Doubly so when I rent a car to travel to the US, since the terms of the contract are often "the rental car company will provide the minimum insurance required in the jurisdiction where the claim is incurred".

[1]: https://www.ontario.ca/laws/statute/90c25 [2]: https://www.ontario.ca/laws/statute/90i08

Cloudflare's not down for me on my non-Render hosted site.

My Render-hosted site is down, but Render's outage seemed to start an hour after Cloudflare's outage.

By all means, the OP should go. Eclipses are cool!

But they weren't asking if they should go, they were asking how to convince someone else to spend 30 hours travelling for 4 minutes of eclipse. My answer was that they should roll it into a larger trip, to amortize the cost.

I was giving them an opinion on how I'd feel in that person's shoes. And I say this as someone who has travelled for eclipses, and will travel for this one.

Honestly, you'd have a hard time convincing me that it was worth it unless it was part of some other trip.

If you're willing to travel 30 hours for a total solar eclipse, wait a few years and travel 10-20 hours instead.

They're not _that_ rare.

Google Sheets has an extensibility system: you write a React app that renders as a sidebar inside the sheets UI. This exposes a few additional API endpoints vs the vanilla sheets API. For example, you can determine the selected range, when a user has edited a value, changed sheets, etc. We use this in our add-on to help people select a range. Sounds like it might help for you, too!

The extensibility system also lets you show a modal dialog, which we use for previewing API calls to third party HTTP servers -- sounds like a similar thing could work well for you, too.

You could also look at supporting named ranges if you don't already, so people can refer to a range as `FinanceData` instead of `Sheet 1!A3:F90`

Some more colour is that the OSM Buildings project didn't even ask permission to use the OSM mark. If they were a community project, this was permissible. If they were a commercial project, they ought to have sought permission from the OSMF.

Given that, it seems pretty reasonable to me that the OSMF's stance was basically: your name is your problem, if you think someone is infringing on it, it's up to you to enforce it.

The OSMF runs on a shoestring budget - something like $400,000/year, vs Wikimedia Foundation, which runs on $170,000,000/year. It would be a real shame if a potential donor thought this particular instance was a reason not to donate to the OSMF.