uniqueuid

It's always nice to see products that cater to the best users can be instead of the worst.

Personally, AI for writing is in the same corner as the other pathologies you've listed (popularity counts etc), so it's not for me. But some folks will see that differently.

The latest 2.7 seems to have a lot of CVEs (did not verify that manually though)

https://www.cvedetails.com/version-list/10210/18230/10/Pytho...

IIRC some commercial distros maintain patches for 2.7 but then you're paying for being 15 years behind the future.

I'm with you regarding the argument, but want to nitpick:

"dismissing" a politician sounds like an easy fix but we probably don't want hyper-polarized dismissal wars where politicians are "shot down" immediately after being elected. That's why there are other mechanisms such as not re-electing, public shaming, transparency fora etc. ... we need to work on strengthening those, the accountability and transparency.

Thanks for the context!

To me this was the most informative comment in the thread because it offers some effect size comparison.

Why not both? :)

Ok here is the crucial part of the paper:

It's a difference in differences design, using individual-level test scores and de-seasonalized data (p. 13). Their wording is:

> Y_igst is the outcome of interest for student i in grade g in school s in time period t, HighAct_s is an indicator for high pre-ban smartphone activity schools, D_t is a series of time period dummies (t = 0 indicates the first period after the ban took effect), δ_s is school fixed effects, and θ_g is grade fixed effects. In this setting, β_t are the parameters of interest, reflecting the difference in the outcome of interest between treatment and comparison schools for each period, with the period before the ban serving as the omitted category, holding grade level constant.

To me some modeling choices seem a bit heavy-handed, but I'm not an economist and could not do better.

You have to admit that it's quite clever how they approximate phone use:

> Our identification strategy relies upon our ability to calculate school-specific measures of smartphone activity that we can attribute to students, rather than adults in the building. To do so, we use detailed smartphone activity data from Advan between January 2023 and December 2024 that we link to LUSD schools using point-of-interest coordinates.13 In particular, we focus on the average number of unique smartphone visits (pings) between 9am and 1pm on school days (a common time frame that elementary, middle, and high schools in LUSD are all in session during school days) in the last two months of the 2022-23 school year (right before the ban took effect) and the first two months of the 2023-24 and 2024-25 school years.14 To disentangle student activity from the smartphone activity of teachers/staff, we subtract the average number of unique smartphone visits between 9am and 1pm on teacher workdays (in the same school year) from the same average on regular school days.

To be fair, all those details are in the paper. And a 1-2 percent increase does not seem low to me for such a measure.

Sure, most of us stay in system 1 (heuristic) most of the time.

But I think it's wrong to assume most people are incapable of serious, thorough thinking. Parents around the world correctly dose medication for their kids all the time, and they mostly do this completely fine.

The key is that people are clever when they both can and want to, and some communication regarding drugs is not well-designed to alert them to want at the right time.

Paracetamol is the most common cause of liver failure in the US. Its toxicity threshold is somewhere around 4g/day for an adult.

I remember that my wife once bought an over the counter cold drug in Italy that had > 1g per pill.

So we should be aware that it's very easy to overdose this particular drug.

More info: https://www.ncbi.nlm.nih.gov/books/NBK441917/

Reading this I could not help but think of compliance and treatment safety for self-managed dosing.

It's evident, for example, that drugs such as Paracetamol (Tylenol for you Americans) should be dosed by body weight in children. To make life simpler for parents, they are given age and/or weight brackets, sometimes along with upper thresholds (e.g. mg/day).

This of course means that lighter children are comparatively over-dosed and heavier children under-dosed compared to a median.

The problem is - I think this works pretty well as a safeguard against dangerous over-dosing (i.e. liver toxicity etc.).

Now how would we turn that advice into a gradual dosing recommendation? We can use mg/kg body weight as is done e.g. in antibiotics. But that carries the potentially fatal risk of miscalculation, and some parents might intentionally overdose over a wrong risk perception.

What we would need is something like an exponential risk curve, indicating a "safe zone" and a "danger zone" while highlighting some critical threshold. This again would need to be age/weight-specific.

Do we think parents would be deterred from giving a kid too high of a paracetamol dose? I'm not so sure, especially over time.

So in the end, I think that in some cases (especially with self-administered dosing) round numbers and sharp thresholds may work well to mitigate fatal risks, even while increasing nonfatal risks.

OTOH, posit funds a lot of development of important packages in the tidyverse and does a lot of community work etc.

So if maintaining RStudio is so much of a burden that it impedes the rest of their work, I don't think it's a bad idea to reduce the amount of work spent trying to compete with VSCode when that's an increasingly tough sell.

I'm not a fan of VSCode personally, but would probably be happy with a tmux setup with a console for R and some minimal output viewer, so people like me should be able to cobble something together that's a workable alternative to Posit.

It has always (or at least for decades) been.

Short answer: no

Tape is really complicated and physically challenging, and there are no incentives for people investing insane amounts of time for something that has almost no fan base. See the blog post about why you don’t want tape from some time ago.

Edit: https://blog.benjojo.co.uk/post/lto-tape-backups-for-linux-n...

You're right, it seems both the UI and the old about:profiles page do use the same underlying implementation, but the UI does not pick up any profiles added through the about: page. If you create a new profile from the UI, that will show up in about (after a restart).

Profiles have always been great, but it's kind of unfortunate that this feature seems to be locked behind a sign-in (the link in the article describes the UI as being in the profile menu).

I mean, I've been using about:profiles for ages, but it would definitely be nice to have a bit more polish (e.g. every now and again I forget that a newly created profile is automatically promoted to default)

[edit] well seems I have to eat my words - there's a switch in about:config named "browser.profiles.enabled" that toggles a profiles menu item with some UI that apparently has existed for years. Nice!

I'm not sure everyone got my hint that the proposal is obviously very bad,

(1) because ivy league also produces a lot of work that's not so great (i.e. wrong (looking at you, Ariely) or un-ambitious) and

(2) because from time to time, some really important work comes out of surprising places.

I don't think we have a good verdict on the Orthega hypothesis yet, but I'm not a professional meta scientist.

That said, your proposal seems like a really good idea, I like it! Except I'd apply it to individuals and/or labs.

If your goal is exclusively reducing strain of overloaded editors, then that's just a side effect that you might tolerate :)

I got that suggestion recently talking to a colleague from a prestigious university.

Her suggestion was simple: Kick out all non-ivy league and most international researchers. Then you have a working reputation system.

Make of that what you will ...

You will effectively want a 48GB card or more for quantized versions, otherwise you won't have meaningful space left for the KV cache. Blackwell and above is generally a good idea to get faster hardware support for 4b (some recent models took some time to ship for older architectures, gpt-oss IIRC).

It is immensely frustrating to me that to this day, after a solid century of advances in sciences and mathematical literacy, we are still implicitly stuck in the mental model of averages.

Reality is fucking far away from averages and we know it. "The economy is doing great/terrible" is an almost worthless indicator unless the person you're talking about actually has business relations into every corner.

Yes, there are interdependencies, but they do not justify that we pretend numbers are so expensive we can only print two of them (mean, sd) at a time. Let's finally stop drinking information through a 2 mile straw and instead show high resolution 2d data at least.

[edit] this is of course not a criticism of parent or OP, it's a systemic problem that we all are guilty of.

Since it seems unclear what they do and why it matters:

Clockss seems to be an organization designed to make sure scientific content does not disappear library of Alexandria-style.

The most important task here is being legally safe, which is why they emphasize ivy league credentials, distributed nature, audits and so on. Technically it's not really difficult (except perhaps for dealing with publisher captchas heh).

They are legally safe because of this mechanism:

> Digital content is stored in the CLOCKSS archive with no user access unless a “trigger” event occurs.

All in all I think it's absolutely necessary.

And that relates to whole journals, not single papers, apparently.

https://clockss.org/triggered-content/

An interesting read.

This kind of problem is exactly what statistics is designed to do, and it makes me a bit sad that we are left with a bit of a shoulder shrug. It's absolutely possible to do a much better job at disentangling possible causes here with something as simple as a multilevel regression. (Although ok, proper causal inference would be more work).

If you look at the concrete laws, they are less spectacular.

For example, the concept of privacy protecting against media coverage is actually weaker for politicians (when in official duty) than for ordinary citizens (Allgemeines Persönlichkeitsrecht).

And libel only applies to statements of facts. I.e. you can't (easily) be prosecuted for opinions, just for making harmful false claims.

Nitpick:

1. Censorship in German constitutional law is only defined as the state pre-screening before publication. That's a very narrow area and rarely applies. Most people from an US legal tradition will consider censorship to include other things such as mandating removal of certain content after the fact, but that's different legal branches with different mechanisms (i.e. libel).

2. What Schulz is talking about in the second link definitely is state censorship (blocking a TV station), but it's not implemented by Germany but on the EU level. (Germany is still involved - complicated matter).

Finally we should appreciate that the US government's opinion on censorship seems to have pivoted quite a lot, so I would expect free speech maximalism to not remain a very popular position on the government level (even though many people may still support it, either naïvely or with robust arguments).

Heat pumps account for 2/3 of new heating installations in Germany [1]. Modern buildings with effective insulation seem to make them quite viable, but that hinges on the availability of attractive electricity prices.

The second factor is that carbon-based fuels may become more expensive over time, so perhaps electricity costs “just” needs to remain stable to become attractive.

[1] https://www.zeit.de/wirtschaft/energiemonitor-strompreis-gas...

It’s important to put this in context.

Lp(a) is a largely distinct risk factor from “ordinary” cholesterol and cannot be changed by diet or exercise. Survey papers show practically no effective treatment (statins help all cause mortality in patients but do not lower lp(a)). There are two (iirc) ongoing trials for new, effective drugs. But those are not available yet and will probably be prohibitively expensive, going by the advertisements that the companies run.

So yeah, get an Lp(a) test once (it doesn’t vary too much over time) and reduce your other risk factors, but don’t put too much hope into an easy solution to this specific cause yet.

edit: found the two papers that were a good read:

Kamstrup, P. R. (2021). Lipoprotein(a) and Cardiovascular Disease. Clinical Chemistry, 67(1), 154–166. https://doi.org/10.1093/clinchem/hvaa247

Schwartz, G. G., & Ballantyne, C. M. (2022). Existing and emerging strategies to lower Lipoprotein(a). Atherosclerosis, 349, 110–122. https://doi.org/10.1016/j.atherosclerosis.2022.04.020

There are many more weird and complex architectures in models for video understanding.

For example, beyond video->text->llm and video->embedding in llm, you can also have an llm controlling/guiding a separate video extractor.

See this paper for a pretty thorough overview.

Tang, Y., Bi, J., Xu, S., Song, L., Liang, S., Wang, T., Zhang, D., An, J., Lin, J., Zhu, R., Vosoughi, A., Huang, C., Zhang, Z., Liu, P., Feng, M., Zheng, F., Zhang, J., Luo, P., Luo, J., & Xu, C. (2025). Video Understanding with Large Language Models: A Survey (No. arXiv:2312.17432). arXiv. https://doi.org/10.48550/arXiv.2312.17432

For what it's worth, I have an i9-13900K paired with the largest air cooler available at the time (a be quiet! Dark Rock 5 IIRC), and it's incapable of sufficiently cooling that CPU.

The 13900k draws more than 200W initially and thermal throttles after a minute at most, even in an air conditioned room.

I don't think that thermal problems should be pushed to end user to this degree.