It's USB-C, "you'll never know" is the answer, often it's "kinda working, but extra slow". I consider having a cheap USB power meter a necessity at this point.
I also found out the hard way that many modern powerbanks will shut themselves off if the charged device doesn't draw enough power, which means many smartwatches will simply not get charged by default. Supposedly that behaviour can be turned off by pressing/holding the power button the right way, but telling non-tech people "with this specific device you just have to remember to first do this arcane ritual before it works" is usually not received well.
We're having this discussion on an article about Anthropic changing their privacy policy. If you don't believe Anthropic will follow their privacy policy, then a change to the privacy policy should mean nothing to you.
Well, yes and no - it gives them more plausible deniability ("oh, this particular piece just ended up in the training set by accident") if they get caught when compared to the previous ToS.
If it were a lie, why take the PR hit of telling the truth about starting to train on user data but lying about the specifics? It'd be much simpler to just lie about not training on user data at all.
If your threat model is to unconditionally not trust the companies, what they're saying is irrelevant. Which is fair enough, you probably should not be using a service you don't trust at all. But there's not much of a discussion to be had when you can just assert that everything they say is a lie.
> Meta downloaded copyrighted content and trained their models on it, OpenAI did the same.
> Uber developed Greyball to cheat the officials and break the law.
These seem like randomly chosen generic grievances, not examples of companies making promises in their privacy policy (or similar) and breaking them. Am I missing some connection?
> These seem like randomly chosen generic grievances, not examples of companies making promises in their privacy policy (or similar) and breaking them. Am I missing some connection?
My point is that whenever we send our data to a third party, we can assume it could be abused, either unintentionally (by a hack, mistake etc.) or intentionally, because these companies are corrupted to the core and have a very relaxed attitude to obeying the law in general as these random examples show.
It's all PR. Some people won't read the details and just assume it will train on all data. Some people might complain and they tell it was a bug or a minor slip. And moving forward, after a few months, nobody will remember it was ever different. And some might vaguely remember them saying something about it at some point or something like that.
> Meta downloaded copyrighted content and trained their models on it, OpenAI did the same
Where did these companies claim they didn’t do this?
Even websites can be covered by copyright. It has always been known that they trained on copyrighted content. The output is considered derivative and therefore it’s not illegal.
> The output is considered derivative and therefore it’s not illegal.
Well, this is what they claim. In practice, this is untrue on several levels. First, earlier OpenAI models were able to quote verbatim, and they were maimed later not to do that. Second, there were several lawsuits against OpenAI and not all of them ended. And finally, assuming that courts decide what they did was legal would mean everyone can legally download and use a copy of Libgen (part of "Books3") whereas the courts around the world are doing the opposite and are blocking access to Libgen country by country. So unless you set double standards, something is not right here. Even Meta employees torrenting Lingen knew that so let's not pretend we buy this rhetoric.
None. And even if it's the nicest goody two shoes company in the history of capitalism, the NSA will have your data and then there'll be a breach and then Russian cyber criminals will have it too.
At this point I'm with you on the zero trust: we should be shouting loud and clear to everyone, if you put data into a web browser or app, that data will at some point be sold for profit without any say so from you.
I mean you really sell short where your data is going to be taken from. Browsers and apps are just the start, your TV is selling your data. Your car is selling your data. The places you shop are selling your data.
Reading this comment gave me a flash of vertigo as I realized how deep down the rabbit hole of "crazy dude that only pays in cash" I'd fallen.
I don't own a car and only take public transit or bike. I fill my transit card with cash. I buy food in cash from the farmer's morning market. My tv isn't connected to the Internet, it's connected to a raspberry pi which is connected to my home lab running jellyfin and a YouTube archiving software. I de Googled and use an old used phone and foss apps.
It's all happened so gradually I didn't even realize how far I'd gone!
What is their gross margins? Is the loss just because they have to keep competing at an ultra competitive race?
On a software focused forum like HN, I’m surprised people still don’t understand the grow at all cost until you are the top 2 or 3 left model. There has been dozens of examples of tech companies losing money for years just to become highly profitable after.
People are still not getting that big tech is investing like their lives depend on it is because they are. GenAI can render the core businesses of big tech obsolete.
> There has been dozens of examples of tech companies losing money for years just to become highly profitable after.
Sure, but there have been thousands of tech companies that lost money year over year and went bust. Odds are that any given AI company will end up losing a lot of money. Maybe the asymmetric potential payoff is worth the risk in certain cases, but it's not crazy to be skeptical about Anthropic or any other hot company.
Sure, but there have been thousands of tech companies that lost money year over year and went bust. Odds are that any given AI company will end up losing a lot of money. Maybe the asymmetric potential payoff is worth the risk in certain cases, but it's not crazy to be skeptical about Anthropic or any other hot company.
That's the game. It's called the power law in VC investing. It's worked.
That depends on the regulators being asleep on the wheel again and letting monopolies happen. At the minimum here in Europe they'd automatically fall under Digital Services Act. According to OpenAI they have 41.3 million average monthly active recipients in the European Union. Only a few million more and they fall under DSA.
“Recently clicked” wasn’t the criterium, it was “showed activity in late 2024”. So nothing that anybody has done this year – including this archiving – will affect which links Google keep alive.
I’ve learned a lot more from talking directly to people, watching videos, participating in forums, and doing things than I ever have from books. This is especially true now. I’m not sure the 1913 take was wrong.
For the past year or so, whenever I’ve wanted to learn something from a book, I’ve downloaded the Kindle edition, converted to plain text, given it to Claude, and asked it to tutor me chapter by chapter.
You can also interpret the Dr Dre quote an abbreviation of, “I’m gonna try (to change the course of hip hop again) and change the course of hip hop again.”
In this form “try and” means you will try to do something and that you will succeed. Some of the articles tests make more sense in this light; Of course you wouldn’t reorder the trying and the succeeding because that’s the order the events will happen.
This ignores the fact that “try and” developed concurrently with “try to” and possibly before. So it wasn’t originally an abbreviation for a phrase that was yet to be established.
"I have tried and finished my homework" is correct to my ear (possibly because I'm Canadian), but it means successful completion as opposed to "I have tried to finish my homework" implying I didn't get around to it.
For me, if someone says "try and see for yourself", it implies possibility of failure or something new.
If outcome is considered in terms of success or failure then try implies non-zero probability of failure.
If outcome isn't considered in terms of success or failure then "try this flavour of ice-cream" is experience and try this outfit(fits or doesn't) is a test you can't fail.
Philosophically, it is as master Yoda said: Do or do not, there is no try.
I don’t think that’s anything like the meaning of “I’ll try and go to the store tomorrow”. There’s no implication that anyone is trying to stop me.
Also, your abbreviation analysis would still leave a syntactic mystery, as that sort of ellipsis doesn’t seem to follow any general attested pattern of ellipsis in English.
That example would be something like 'I'll try to go to the store tomorrow and see if I can' along the lines GP suggests. 'stop me' only came from the specific example they were using.
You can actually construct this using regular VP ellipsis (or possibly Right Node Raising?) in English, but it sounds weird and doesn’t convey the same meaning. So I don’t think so.
“I’ll try to ___ and see if I can go to the store tomorrow”. [where ___ is the VP ‘go to the store’]
Then you have the various syntactic facts mentioned in the article , such as the possibility of wh-extraction. This isn’t possible in an analogous ellipsis construction:
“What did you try and eat?”
* ”What did you try to and see if you can eat?”
There’s also an interesting tense restriction which suggests that there’s no independent elided clause:
Your examples are not ringing bells for me as a native speaker. The linguistics terms may or may not be confounding, but are too unfamiliar for me to discuss.
If you check the parent comment, the 'most' applied to the fraction of mysteries, not the the fraction of instances of the construction that the analysis is supposed to apply to. But anyway, this isn't an exception. The overwhelming generalization is that "try and do X" means the same as "try to do X". This holds for imperatives like the OP's example just as much as for my examples. There's very little difference between the to/and variants of any of the following:
Try and/to do it quietly
Try and/to be a little more polite.
Try and/to hand your homework in on time.
I agree that in some specific cases there are slightly different shades of meaning. However, this doesn't seem to be a very systematic phenomenon, or one that obviously justifies the assumption that "try and" is an elliptical expression of a complex multi-clausal construction.
I also like how several linguists attempt to call out this usage as wrong:
> deemed prescriptively incorrect (Routledge 1864:579 in D. Ross 2013a:120; Partridge 1947:338, Crews et al. 1989:656 in Brook & Tagliamonte 2016:320).
Linguists don’t say varieties are right or wrong (even though they might have private aesthetic opinions like everyone else). That would be like a biologist saying dogs are the correct version of mammals and cats are wrong and/or don’t exist.
Just because someone has a degree and a job in the field doesn't mean they understand how science works. Prescriptive biology is even sillier than prescriptive linguistics.
Maybe in the USA. But bodies like the French Academy and many similar ones in other countries very much do issue opinions on correct grammatical (and vocabulary) usage. And members of these institutions are very much considered the top of their field in their countries.
The people they’re citing are either authors of usage guides or linguists who are simply noting that the usage has been deemed incorrect by some of the former.
These are not linguists doing that. No self-respecting linguists will waste time doing prescriptivism. These are two linguistic articles about this constructs that are quoting amateur language usage manuals. The oldest one is a boys magazine[1] published in 1864 discussing "the Queen's English"[2]. The newest one (Crews et al.) seems to be an obscure usage manual for writers[3].
As demonstrated here, "try and" is older and more "original" than "try to", if not contemporary with it. Any other reason why would "try to" be more "correct" cannot even make sense as anything more than a purely uneducated opinion. When you dig deep into most examples of perspectivism you'll usually run into the same story too. "Incorrect" forms often predate the "correct" forms and are often employed by respected writers (such as Shakespeare and Jane Austen). And even if they don't, there isn't really any scientific ground to brand one form as incorrect.
Linguists do not generally engage in linguist prescriptivism. As far as I'm concerned (and I believe most linguists would agree), this is stylistic opinion at best and pseudoscience at worst. Still, it's not linguists can do anything to stop amateurs from publishing prescriptive language usage manuals, so you'll always have people who claim that "try and" or "ain't" or "me and my friend went for a walk" are incorrect.
[2] Yes, this is Edmund Routledge whose father is the namesake of the present scholarly publisher, but they were just publishing popular books back in the 19th century.
If a modern linguist call any usage as wrong, I would ask for his diploma and check if I have to close his university, because clearly they shouldn't teach linguistics 101, let alone bring someone towards a PhD. Linguistics is descriptive, not prescriptive.
Then linguistics is worthless. Descriptivism can't actually tell you anything; all someone can say is "yep, you sure are using that word that way". Fortunately prescriptivists still exist despite people's best efforts to give it a bad name.
Linguistics can explain why and how the language evolve, who caused this evolution. I don't know any modern linguist (as in doctor in linguistics) who ever wrote 'you should say this and not that', because all of them knows that languages change. The only 'prescriptivists' are bad philosophers and English majors.
How can you learn about how people spoke and the patterns that dictate how that changed over time if all you care about is what is considered technically correct at the moment?
This is a good intuition. The construction is actually sometimes jokingly called the "Try And"-C, where "C" stands for Complementizer, a thing that introduces and subordinates a clause.
I think this capture’s the essence better than anything else, “try and” simply behaving as “try and see if I can” (or whatever fits instead of “I” here)
reply