jules

Ah I see, so you represent all closures as two machine words, in a representation that the GC can understand?

What's the technique to unbox closures and ADTs?

If you want your sort to replace the one in the stdlib I imagine you have to adhere to the strict rules.

These are college students taking the same course, so the effects you hint at are reduced compared to getting a sample from the general population.

That's fine, selecting a beautiful location takes a certain amount of skill. Not as much as designing a beautiful building, but still. The issue is that this approach doesn't scale. We can't all live in a forest on top of a river. At some point, you have to build a city where you have lots of the given architecture. You can have a city center made entirely out of classical architecture, or such a city center with an interesting sprinkle of modern architecture. But if you build it entirely out of modern architecture, it will not be interesting, nor beautiful, nor hospitable.

This is a good test to apply to architectural styles. People often cite Fallingwater (https://upload.wikimedia.org/wikipedia/commons/b/bc/Fallingw...) as a triumph of modern architecture. In my view, it only really looks good because of its surroundings. If you'd erase the trees and river, it wouldn't look good anymore, especially if you zoom in to the rounded off smooth brown slabs. If, on the other hand, you erased the house and kept the trees and river, it would still be a beautiful scene. If you replaced the modern house by any old watermill constructed by 18th century peasants, it would look just as good as what a famous modern architect was able to achieve. Therefore, I claim people's aesthetic senses are primarily responding to the trees and water, and not to the product of the architect.

A similar phenomenon occurs when architects place a modern building in a scene of beautiful classical architecture, sometimes even connecting the new building to the old. The "beauty" of this type of modern architecture is parasitic on pre-existing beauty and cannot stand on its own.

Would you want to stay there again if they triple charge you and do not want to fix it? That's called theft. Would you want to work with a business that is nonchalant about stealing from you? I bet the hotel would not be so nonchalant about you stealing 2x of the cost of the hotel room from them.

I made the mistake of not using a credit card to pay for an airline ticket. They canceled my flight at the last moment with no reason given, did not want to provide alternative transportation, and only offered to refund 80 euros of "credits" tied to that airline (far less than what I paid for the flight!). I never want to use that airline again, and would happily chargeback.

Nice paper!

It's not clear that even a super smart compiler can do this. The best schedule depends on the latency of instructions. This is a problem because we can't know statically whether a particular memory load is in L1/L2/L3/DRAM/etc., as this can vary for different executions of the same load instruction.

Nice slides! Of course, I should have asked whether it's on the pareto frontier. By the way, the author of pdqsort has a new sorting algorithm: https://www.youtube.com/watch?v=2y3IK1l6PI4

Great paper. Is that the fastest sort right now?

Except that persons C, who control the instances where many interesting people are, have banned instances run by people who are not like the Cs. Furthermore, I don't even want to bother with all this stuff, playing detective trying to figure out the character of the moderators of different instances, figuring out how to migrate, and figuring out which other instances they federate with. There is no beauty to be found here, just human nature being such that as a species we can't even accomplish a platform where we can read the text messages from people we want to follow.

His mastodon account has also been suspended on the server I'm on. Ugh, what is it with person C deciding that person B is not allowed to see what person A says. I dislike human nature sometimes. Or at least the personality type that is attracted to moderator positions.

The battery price graph tells a different story. Improvements have been slowing down for a long time, and have been almost zero the past couple of years and we've actually had a price increase this year. Switching to an entirely new technology is speculative at best, not all that dissimilar from putting all hope in fusion. Could work, but betting the future of the planet on it is silly.

Measuring cost by installed capacity is highly misleading because solar and wind need storage to be reliable sources. The cost of storage dwarfs the other costs, and battery prices have stagnated.

I haven't read this book, but I do try to mention a person's name. Not because of them, but because otherwise I will have forgotten their name in 2 minutes.

When eating a dish, I can usually identify the ingredients and spices by flavour, but different wines taste more or less the same to me. Some taste different, but I can't say that they taste significantly better than others. Spending $50 more on the food in a restaurant is going to make an enormously bigger difference than spending $50 more on wine. In fact, I find that not drinking wine at all usually makes for a better experience if the food is good, because the super strong flavour of wine destroys your ability to discern flavours on the next bite. Not all that different from putting hot sauce on all your food, in fact.

There are (short) computer programs where you input a mathematical proposition and a proof in a kind of proof programming language, and the program will then check if it's a valid proof. Saying that mathematical truth is a social construct is technically true but misses the point entirely.

I bet he would if his movie scene is pixel for pixel identical to the scene he watched.

I once simply pointed out how an equation in a climate science paper didn't make sense because there was a unit error (there was exp(x) where x was not unitless) => removed.

What's the advantage of being distributed?

Well, they apparently did donate to some woke-adjacent pseudoscience, which seems like a waste of money to me. Regardless, I agree with you that the question isn't where some $0.25 million went, but why they are spending $100 million and still asking for donations with that misleading message, while a few years ago they were spending a small fraction of that even though their hosting costs were actually higher back then. It increasingly looks like Wikimedia as an organisation has a parasitical relationship with Wikipedia, doing enough to keep its host alive with a small fraction of its budget, while benefiting from the work of volunteers and not even fixing longstanding issues with the software that the volunteers ask to be fixed. Even beyond fixing bugs, I'd be happy to add to their $100 million budget if they actually did useful things with it. I can easily think of 10 features that would improve Wikipedia. For instance, Wikipedia pages such as "list of countries by GDP per capita" (https://en.wikipedia.org/wiki/List_of_countries_by_GDP_(PPP)...) are useful, but the table UI is not that great and the map is a non-interactive PNG image. For $100 million, why can't we hover over the country and see its name and the exact GDP per capita number?

Here is the message wikipedia has on its donation page:

> We'll get straight to the point: Today we ask you to defend Wikipedia's independence.

> We're a non-profit that depends on donations to stay online and thriving, but 98% of our readers don't give; they simply look the other way. If everyone who reads Wikipedia gave just a little, we could keep Wikipedia thriving for years to come. The price of a cup of coffee is all we ask.

> When we made Wikipedia a non-profit, people told us we’d regret it. But if Wikipedia were to become commercial, it would be a great loss to the world.

> Wikipedia is a place to learn, not a place for advertising. The heart and soul of Wikipedia is a community of people working to bring you unlimited access to reliable, neutral information.

> We know that most people will ignore this message. But if Wikipedia is useful to you, please consider making a donation of €5, €20, €50 or whatever you can to protect and sustain Wikipedia.

Whether or not one likes the causes they give the money to, if they spend $100 million and they only use $2.4 million for hosting, and they also give money for political activism, then this is a misleading message, making it sound like they are on the cusp of not being able to cover the costs that keep the site online unless they start having ads on wikipedia.

That's magical thinking. Let's start with the obvious question: what kind of storage would that be? Remember, your argument against nuclear was price. Have you run the napkin math on your preferred choice of storage?

Yes, this argument gets repeated again and again and again, and when you ask about storage you never get a decent answer. And yet, the argument keeps getting repeated even by people who matter and make the decisions. When I started watching debates in the Dutch house of representatives I pretty much lost all hope. Of the 150 members, I'd say 145 are either unable or unwilling to think quantitatively and do simple napkin math. Not just regarding energy policy but in all areas. Yet they are verbally very strong, which makes the problem even worse. I suspect it's the same in most western countries.

If you think I'm exaggerating, remember when the Royal Statistical Society asked "if you spin a coin twice, what is the probability of getting two heads?" and 47% of Conservative MPs and 77% Labour MPs got the answer wrong. Many technical people cannot even fathom this, and tend to assume that people who cannot answer such questions must be severely mentally impaired in general. That is completely wrong; you have lots of people who are near geniuses verbally but incapable of or unwilling to do quantitative thinking.

> They just run regression on it and anything with p<00.5 that seemed fun was published.

How do you think science operates in actual practice? OkCupid at least has a sample large enough that you can commit all sorts of statistical sins and still produce halfway decent conclusions, unlike your run-off-the-mill n=10 study.

Did she log in from a different geographical location? That can matter.

That is not true. Wind output is highly correlated across the continent. Days in which wind output is very low happen regularly. In fact, every couple of years there is an entire week like that.

The advantage of nuclear is that it will work almost certainly, albeit possibly at a high cost. We should continue to invest into wind and solar where that makes economic sense. On the off chance that a magic energy storage technology comes along we can then phase out nuclear again. However, betting the future of an entire continent on the coming into existence of such a technology is wildly irresponsible.

What explains the big difference?