timr

The reason is that it's propaganda.

That also isn't even close to what the word means.

https://www.merriam-webster.com/dictionary/propaganda

Everyone loves the experts that repeat what they like to hear. The ones that don't are obviously quacks.

Or neither option in your false dichotomy has any relation to the quality of this paper?

This article is nonsense. It takes observational data from a survey, averages it, doesn’t control for obvious medical factors that more strongly influence the observed data than climate (e.g. illnesses), and then asserts that the correlation is causative.

It is not science. It’s just propaganda.

Whatbaoutism is not a valid excuse.

It’s not “whataboutism”. It’s a basic control in a paper that takes population data and asserts a causal link between a complex, personal biological variable and a global phenomenon.

There are probably dozens of factors that matter a lot more to individual blood gas levels than global average co2. At a minimum, I’d expect to see controls for obvious medical factors before taking this argument seriously.

This is such a bad paper. They take the NHANES data, average it for all participants, don’t bother controlling for things that have far more direct relevance to individual bicarbonate levels (e.g. diabetes, antacid use), and just assert that an observed correlation is causative.

This is a very hot take. Being a founder might confer you some career advantage [1] if you're one of the few to make it to series A and you grow a large team, but if you're like the vast majority of founders and never make it past seed (or realistically, to seed), your network is the alpha and omega of any subsequent job search. It's a bit like saying that being an NBA starter is a good way to get a job as a basketball coach.

Point being: don't start a startup if your goal is to get a job. Just get a job.

[1] Note that I'm not arguing about experience -- you can gain a lot of experience as a startup founder, but that experience is rarely directly marketable. Also, most startup founders are completely clueless when they start, so "a lot of experience" is a relative term.

I know that this is a tall order, but consider that this might not actually be “objective”, and that instead, you’re substituting your opinion for facts.

(You don’t need to enumerate all of the things you believe justify the statement. I’m well familiar with them. I’m just trying to make you understand that there’s a layer of subjectivity here.)

They've made it do exactly what they need, exactly how they need it.

…until it breaks, or the bot goes off and does something ridiculous that you can’t understand without domain expertise. Which always happens.

I’m sympathetic to your argument, but I also strongly believe that most of these so-called “exactly what I need” projects will be abandoned within a fairly short time. That’s totally fine, but it’s not where most of the effort in professional software lies.

It's maybe more interesting if you understand the input, but it's really not that important at the very beginning. Hearing the language being spoken will train your ears to hear the sounds and the words, even if you only understand a fraction of the content. Later on, it's probably more important that you actually understand the majority of what you're hearing, but even at a high level, it can be valuable to zone out on a stream of incomprehensible audio content. I like to do this with news podcasts while cooking.

I got from zero to low N3 via a year of immersion, and I still remember the moment when I started to "hear words" in real-world conversation for the first time. It was very much a 0->1 transition, and happened almost entirely without my active participation.

Don't sweat it too much, is my point. Just listen to stuff.

Isn't it assumed that the AI agent is allowed to read your files in the directory you launch the harness?

Yes. There's very little story here. Maybe Grok is being like 10% more aggressive than other providers in how they assemble context (more likely: it was faster to ship this way), but any provider has the ability to do the same thing, and will happily do it if it helps improve results. Authors acknowledge this openly, but it's buried:

"Cloud AI tools send context; this is normal." True, and conceded: any cloud coding agent must send code to its server to act on it. The novel deltas here are (a) a secrets file (e.g. .env) is transmitted unredacted, (b) the content is persisted to a named GCS bucket, not just processed transiently, and (c) the upload mechanism is not surfaced in the CLI's setup materials (§7) and on by default.

This is the entire controversial portion of the finding, in a single paragraph.

As far as the .env thing goes, you shouldn't be putting unencrypted .env files in the accessible path of any LLM. If you do, you're asking for trouble. It would obviously be better if Grok identified secrets and ignored them, but this is not a behavior you should rely on.

What's your mechanism for doing this?

Right now, I'm using Anki to learn Japanese vocabulary from N5 to N3[1]. I know the words and the example sentences well enough to read N3-level text, provided I know the grammar. But when it comes to listening, I struggle to understand even N4-level spoken Japanese.

You need to listen to people, ideally in actual conversations (I am conversational in Japanese, somewhere around N2).

Spaced repetition won’t help you here. The fundamental problem is that you can’t hear the words, which is not a matter of memorization. The only thing that will fix it is exposure to lots and lots of native input.

Fortunately, it is easy to find native listening input today! If you’re a beginner and not in a classroom setting, you should be spending at least half your time listening to native material. The best option is a native tutor and/or friend(s), but failing that, get thee to Youtube and Netflix or even Instagram. There are tons of good teachers on those platforms, posting content for free. Nihongo no mori, Takoyaki_senseiyade, nihongo_camp, and miyu_to_nihongo are just a few I follow on instagram who tend to post serious learning content, mainly in Japanese.

Also, spaced repetition is more valuable at the higher levels, not less. You do need to spend more time reading and listening to native input, but the volume of things to learn grows exponentially, so efficiency is essential. At N5/N4, I could have used paper flashcards and been fine.

You do you man.

No. That’s not my position, it’s a ridiculous metaphor here, and even if it weren’t, the logic doesn’t follow. I am not obligated to stop using a product because someone associated with the product did a bad thing once. Literally no product passes that test.

Yes, your rational and not-at-all ridiculously extreme rhetoric is definitely convincing me of your position.

Also, I can’t “shift the goalposts”. I didn’t set them up. Please make a note of it.

I’m not sure if you’re trying to say that it’s left or right, but it’s perhaps relevant to point out that that article is about the political bias of all major models. And right there in the subhead, they explicitly say that the models have a left bias.

That's not a "classic". It's just a silly, overwrought straw-man [1] that gets passed around by people who think they're being clever, and/or don't understand the topic with any degree of nuance.

Nobody reasonable suggests that RCTs must be used for every problem. When the effect size is enormous, or there's obviously no alternative (as in the parachute example), it's silly to suggest the use of an experiment.

On the other hand, if you want to compare some subtle difference between two competing parachute designs, for example, then perhaps an RCT would actually be appropriate. This is much closer to what is actually being measured in real clinical trials.

[1] To be completely fair to the authors, it was intended as a joke, not an argument. The real joke ended up being that people took it seriously.

I guess if you’re so completely committed to left-wing ideology that you’re willing to overlook ridiculous factual inaccuracies, it’s…fine.

https://www.currentaffairs.org/news/the-worst-magazine-in-am...

I explicitly chose a criticism from someone with a wildly different political ideology than my own, just to show that this is not a partisan take. I disagree with almost every opinion expressed in that link, yet agree completely with the validity of the examples given. The Atlantic routinely makes statements contrary to fact, and seemingly has no qualms about lying or exaggerating to make an attention-grabbing point. It’s coverage during Covid was particularly atrocious.

In medicine, observational evidence is actually better and far more ethical than the RCT. (Which simply dooms the terminally ill to fake treatment.)

This is just nonsense. First, everyone in a trial is informed of the situation. It's not "unethical" unless you lie about it. If you participate in a trial, you do so knowing that you might not get the experimental drug. It's a selfless, honorable thing to do, and we shouldn't be framing it as some kind of scam.

Second, we don't give terminally ill people "fake treatment" (placebo trials). We give them current standard of care. Giving someone a placebo trial doesn't prove anything that would change clinical practice, because you want to know if the drug works better than what is out there today. Rarely is that standard of care "nothing", and this (bad controls) is actually a primary reason that a lot of drug company trials are rejected by the FDA.

If I didn't see the Wall Street Journal editorial board repeating the same garbage in defense of patent medicines, I'd write you off as simply having a sophomoric understanding of how trials work. I'm convinced that someone is driving this absurd narrative.

Do you mean that you don't believe that humans learning a language have the problem the parent described? Because I do, and everyone I've ever met while learning does as well.

Do you mean that you don't believe the problem exists in general, because here's another example: if you give a song title, I can easily hum the opening. If you give me the opening, I cannot reliably name the song.

One thing I've never seen discussed on this topic (possible I just missed it, I only read popular accounts) is whether speaking multiple languages is a proxy for higher sociability / stronger social ties.

Yes. This is exactly what you should be asking with this kind of stuff. The research is hopelessly confounded by social status traits that correlate with wealth.

And before anyone says it, the abstract claiming "adjustment for linguistic, physical, and sociopolitical exposome factors" is fine, but it's essentially impossible to control for something as pervasive as the effects of wealth, without randomization. There are also factors -- like the culture you grew up in -- that are equally difficult to control. For example, if your dataset has only X multi-lingual Americans, and 1000x multi-lingual Western Europeans, no amount of statistical massage will correct for the imbalance.

No, that’s called pharmaceutical development. That’s the business.

We don’t generally fund Merck’s R&D with federal money. You’ll note the following critical detail from the article:

That will impact 22 projects being led by major pharmaceutical companies, including Pfizer and Moderna, for vaccines against bird flu and other viruses, HHS said.

We’ve gone so far round the bend with partisanship that straight-up corporate welfare has become a left-wing cause.

…Which hasn’t changed.

Also, for the record: very few (no?) vaccines are “mandated” by the federal government. Recommendations are made, and state and local governments do this, mainly through school districts.

Various agencies and the military will, of course, mandate things for their own staff.

The facts are true. Blaming Trump is the innovation.

For the record, I don’t care who gets blamed. I just think it’s a hilarious twist of partisan rhetoric.

No. Pharmaceutical companies love vaccines. They’re relatively easy to make, they’re indemnified against harms, they cannot be generic, and they are wildly profitable. And on top of all of that, they often get mandated by schools, ensuring a captive market.

If the government never funded another study for vaccines, ever, pharma companies would continue to pump them out.

Trump one had a sane cabinet that largely controlled his wilder impulses.

This is absurdly revisionist. The first administration’s cabinet/staff was a reality show and a merry go round of people like Anthony Scaramucci and Ryan Zinke. If anything “controlled” it, it was just the chaos of incompetence.

As far as loyalty goes, I suppose it’s worth reminding you that Kennedy was a Democrat, who ran in the Democratic presidential primary, and routinely criticized Trump.

Your understanding is incorrect. All research is unprofitable, by definition. Vaccines are wildly profitable.

Well, I have to say that this is the most innovative leap of partisan politics I’ve seen so far this year!

Most left-wing critics are still struggling with admitting that Anthony Fauci really did provide funding to EcoHealth, despite ample documentation.

You’re splitting hairs.