eaglefield

Impressive!

I think the most I experienced at the physics department in Aarhus was 70ish students. 200 sounds like a big undertaking.

At the price per student it probably makes sense to run some voluntary trial exams during the semester. This would give students a chance to get acquainted to the format, help them check their understanding and if the voice is very intimidating allow them to get used to that as well.

As an aside, I'm surprised oral exams aren't possible at 36 students. I feel like I've taken plenty of courses with more participants and oral exams. But the break even point is probably very different from country to country.

The solution of differential equations by separation of variables in physics is also notated in an abusive way. You have some differential equation

dy/dx = g(x)h(y)

You separate the variables by some quick manipulations

dy/h(y) = g(x) dx

And then you have a small step in some coordinate on both sides. So by integrating both sides

\int 1/h(y) dy = \int g(x) dx

you find a solution to your differential equation. Obviously there's a real formal procedure underneath it with also some safeguards. For example you're supposed to check that h(y) doesn't equal 0 at any point. But the happy path in physics is often done without worrying about all that.

Anubis is not meant to stop single endpoints from scraping. It's meant to make it harder for massive AI scrapers. The problematic ones evade rate limiting by using many different ip addresses, and make scraping cheaper on themselves by running headless. Anubis is specifically built to make that kind of scraping harder as i understand it.

Later in the article it seems they expect the EU and the US to sign a trade agreement soon-ish, whereafter they can start importing again. So i think the plan is just to wait it out.

I just did today's word and I thought this was pretty fun. M⁻¹M stumped me for a while, but once I got it, it gave me a good chuckle for how obvious it was in hindsight. I do work in physics, so that gives me a head-start on recognizing all these formula. From the other comments here it seems enjoyment is very correlated to amount of these formula one already knows.

Anyways looking forward to checking this the next couple of days

I don't think I would classify all stages of fetal development as persons, though obviously that line is crossed at some point during pregnancy. However, I read Judith Thomsons "A Defense of Abortion" at some point and she succesfully convinced me fetal personhood doesn't actually matter that much to the morality of abortion.

It can be found on the internet[0], and is only 13 pages. Alternatively I think the wikipedia article outlines her three main arguments quite well[1], and is a quicker read. The most known part is her violinist thought experiment, which can be read to only allow abortion in cases of rape. However, the subsequent parts convinced me that, abortion is moral in general. So long as reasonable steps have been taken to avoid pregnancy.

[0] https://rintintin.colorado.edu/~vancecd/phil215/Thomson.pdf

[1] https://en.wikipedia.org/wiki/A_Defense_of_Abortion

Ah that's embarrassing. I had wanted to link to this article since it contained some context. https://nypost.com/2022/06/30/twitter-suspends-jordan-peters...

Thanks for pointing it out, will fix.

> I mean: you can disagree with Jordan Peterson (I'm certainly not a religious person, for a start) but what did he do that warranted the old Twitter censoring him?

This tweet[0] was the kickoff point as I recall. I don't really think that getting banned pending deleting the tweet is that egregious. It's one thing to have non-woke opinions and argue for them. I don't follow Jordan Peterson, but I imagine he had posted many times on transgender issues, before he posted the offending tweet.

The issue seems more to have been specifically calling out specific persons, "criminal physician", Elliot Page, in a way that could be construed as inciting to violence

[0] https://nypost.com/2022/06/30/twitter-suspends-jordan-peters...

The github repository is a godot project. Godot is a game engine. The quickest way is to download godot. Clone the repository and open the folder as a project in godot.

My impression is that the arrest has to do with telegrams failure to cooperate with law enforcement. Hardware manufactorers don't generally have the ability to monitor and moderate what their product is used for. I guess one could imagine a hardware manufactorer being arrested for refusing to turn over a log of sales or similar.

Obviously god has no such restriction, but he's probably outside French jurisdiction, and i don't think they have an extradition deal in place.

I think that's an overly uncharitable read on this approach. Lots of tasks that have difficult thoughts, that need to be thought before they can be completed, also have phases in which work just has to be done. I'm in the middle of collaborating on an article for submission to a physics journal. I wouldn't term it filler work, but most of the complex thoughts on the problem have been thought through and the work right now is creating a coherent story that goes over our results. An outline method would work fine for this part of the project.

As for the spy novel, i think the outlining is actually quite similar to how Sylvester Stallone described his writing process[0]. You wouldn't fill the outline with generic beats, you would put in your basic plan for the story.

[0] https://www.youtube.com/watch?v=v_xqfkVNwEU

It's related, but as i recall the reason that short pulses cut better has to do with heat transfer. Most heat transfer in metal is due to the vibration in the electrons, being much lighter than the atomic lattice they move in. Very short pulses, means that the heating in the beam path happens faster, than the electrons can transfer it out to the rest of the metal. In long pulses once the metal is vaporised, a lot of the surrounding material has melted. This molten metal cools into jagged structures and that leaves the edge weird. All this doesn't happen if the pulse is short enough. It vaporises metal before the surrounding structure has a chance to heat up.

Here's a figure showing the quite stark difference. https://www.researchgate.net/figure/SEM-image-of-the-substra...

Here's a figure comparing the hole left in a metal surface with normal and femto-second pulses. The difference is quite stark.

https://www.researchgate.net/figure/SEM-image-of-the-substra...

Could be that fiber doesn't have any resale value. Whereas copper theft is a somewhat common issue.

That's fair. I get the wrong answer on the gpt3 and gpt4-o models, but there's always some uncertainty involved in these gaps. When I appended "Consider your answer carefully." to my prompt, it answered as though the goal was for only the sheep to get across, and that the farmer had to get back to the original shore.

IMO the word intelligence doesn't seem a good description for the thing LLMs possess, they perform pretty well on some tests measuring it. But they're also sometimes wrong in ways intelligent beings (humans) never are. Like this[0] riddle about boats and farmers i stumbled over recently:

> A farmer stands on the side of a river with a sheep. There is a boat on the riverbank that has room for exactly one person and one sheep. How can the farmer get across with the sheep in the fewest number of trips?

It's obviously riffing on the classic wolf sheep lettuce riddle, but I don't think that's gonna fool any humans into answering anything but the obvious. ChatGPT-4o on the other hand thinks it'll take three trips.

They perform a good approximation of intelligence most of the time but the fact that their error pattern is so distinct from humans in some ways, suggests that we probably shouldn't attribute intelligence to them. At least in a human sense of the word.

I think they're talking about linearity as transformations with a linearity score close to 1. They defined that linearity score a little higher up. Such that composing many almost linear transformations will create a total transformation that is very nonlinear.

Obviously it'll be up to some court to decide, but I don't think openai did themselves any favours in a possible voice protection lawsuit, when Sam Altmann tweeted out "her". That makes it seem very much like the imitation was intentional.

Additionally I think you're understating the similarity to the Midler v Ford case. It has a similar pattern where they first contacted the "original", then hired an impersonator. The song wasn't at issue, they had a license to that part.

I think this part of some of the court documents[0] is especially relevant. Hedwig was the sound alike hired.

> Hedwig was told by Young & Rubicam that "they wanted someone who could sound like Bette Midler's recording of [Do You Want To Dance]." She was asked to make a "demo" tape of the song if she was interested. She made an a capella demo and got the job.

> At the direction of Young & Rubicam, Hedwig then made a record for the commercial. The Midler record of "Do You Want To Dance" was first played to her. She was told to "sound as much as possible like the Bette Midler record," leaving out only a few "aahs" unsuitable for the commercial. Hedwig imitated Midler to the best of her ability.

> After the commercial was aired Midler was told by "a number of people" that it "sounded exactly" like her record of "Do You Want To Dance." Hedwig was told by "many personal friends" that they thought it was Midler singing the commercial. Ken Fritz, a personal manager in the entertainment business not associated with Midler, declares by affidavit that he heard the commercial on more than one occasion and thought Midler was doing the singing.

> Neither the name nor the picture of Midler was used in the commercial; Young & Rubicam had a license from the copyright holder to use the song. At issue in this case is only the protection of Midler's voice.

So the fact that us random internet commentors did not recognize her voice doesn't seem to matter in cases like these. It's enough that the sound alike had been told to mimic the original voice, and that people familiar with the voice be fooled.

[0] https://law.justia.com/cases/federal/appellate-courts/F2/849...

Theres no smoothing involved, it's three distinct domains with different function definitions. Top part is log(x) from [a,\infty), bottom part is -log(-x) from (-\infty,-a] and the middle part is cx from (-a,a). With c chosen to ensure continuity in the transform.

In the example from matplotlib I linked in the earlier comment they call out that the symlog transform has a discontinous gradient at the a's, and that the asinh transform can be used instead if that's a problem.

Edit: On reflection it's probably not entirely correct to talk about it as choosing an appropriate c. Since that transform seems to kinda break apart around a=1. Simpler to consider it a matter of plotting on a logarithmic scale down to some value. Then continuing the plot on a linear scale until you reach the negative value on the other side and then plotting on a negative logarithmic scale (-log(|x|)).

Symlog is symmetric log. It's not arcsinh, instead it's a normal logarithmic scale, that turns into a linear scale around zero (for example in the range [-1,1]) and then it turns into a negative logarithm on the other side of zero. So similar but slightly different.

https://matplotlib.org/stable/gallery/scales/symlog_demo.htm...

for an example.

I don't think your second part is really true at least for the titanic. You can look up the brick list on the Internet[1] and the vast majority of pieces exist in other sets. Some of them may only exist in that colour in the titanic set but at least to me they're all still recognisably lego bricks. Unless your standard only includes the right angle pieces, but that applies as well to anything produced by lego in the last maybe 30 years. The titanic uses black lego minifigure hotdogs for the anchor, which if anything I think is a testament to the creative piece use at play in these.

[1] https://www.bricklink.com/catalogItemInv.asp?S=10294-1

I wonder if this is a space where apple (and other webcam providers) would benefit from surfacing some API to the software. To allow an instance to instance control over these confetti features. So context the software is used in, can decide whether the call is a serious business call or family catchup where silliness is allowed. Eg it would be off by default in your orgs Microsoft teams, but on when you skyped friends and family.

On the other hand, it sounds like from the article, that the small gaffes from this technology sometimes help break up what can otherwise be a very impersonal experience. All else being equal, a lot the sensory impressions in a face to face meeting must disappear, in projecting it down into a 2d videofeed. So exaggerating what is left, even if it misses sometimes, maybe makes for a better experience.

I'm no lawyer, but it seems like it ends and begins at wanting to sell in the european market. It doesn't seem that different from mandating that cars obey certain emission standards, contain digital radios etc. Or how food packaging contain nutritional information. Mandating 2-year guarantees for sold goods. There's quite a lot of legislation on specific requirements on sold goods.

One can argue whether this specific legislation is wise, but legally i don't think there's any limit to what the EU can mandate for goods sold in their market.

Yes, moving towards any object at speed in special relativity will make it seems contracted (shorter along the axis you're moving.). So the observer in the garage will see a contracted ladder. Meanwhile an observer seated on the ladder would see a non contracted ladder, but the garage would be contracted. The trouble of the paradox is of course that the garage does not get larger to allow in the "larger" non contracted ladder. This seemingly breaks what is called the principle of relativity, which is that the laws of physics are the same in all inertial frames of reference. The simultaneity they're discussing is the notion that events that seem simultaneous in one frame of reference, may not seem simultaneous in another frame of reference.

So where the first door would close and the second open at the same time to the garage observer. The second door would open much before the first door closes to the ladder observer.

Since ASL doesn't have a massive corpus of scrabeable training data (at least I would think). It's probably on the harder side when it comes to creating machine translators.

I'm running silverblue but running my containers through distrobox. Both toolbox and distrobox are running on podman under the hood, so it's the same technology as far as I understand. However, distrobox has some interesting features relevant to this idea of development isolation. One is that it has an assemble feature[1] built-in. Where you can feed it a recipe file and it will build or rebuild containers accordingly. The other is that it allows setting a custom home directory for the container, among other host/container isolating options[2].

Perfomance wise my containers take a couple MB of rams and no perceptible CPU usage when not in use. At least as far as I can tell.

[1] https://github.com/89luca89/distrobox/blob/main/docs/usage/d...

[2] https://github.com/89luca89/distrobox/blob/main/docs/usage/d...