> One could specify a smallest effect size of interest and compare the plausibility of seeing the reported p-value under that distribution compared to the null distribution. 6 Maier and Lakens (2022) suggest you could do this exercise when planning a test in order to justify your choice of alpha-level
Huh, I’d never thought to do that before. You pretty much have to choose a smallest effect size of interest in order to do a power analysis in the first place, to figure out how many samples to collect, so this is a neat next step to then base significance level off of it.
In a perfect world everybody would be putting careful thought into their desired (acceptable) type I and type II error rates as part of the experimental design process before they ever collected any data.
Given rampant incentive misalignments (the goal in academic research is often to publish something as much as—or more than—to discover truth), having fixed significance levels as standards across whole fields may be superior in practice.
The real problem is that you very often don't have any idea about what your data are going to look like before you collect them; type 1/2 errors depend a lot on how big the sources of variance in your data are. Even a really simple case -- e.g. do students randomly assigned to AM vs PM sessions of a class score better on exams? -- has a lot of unknown parameters: variance of exam scores, variance in baseline student ability, variance of rate of change in score across the semester, can you approximate scores as gaussian or do you need beta, ordinal, or some other model, etc.
Usually you have to go collect data first, then analyze it, then (in an ideal world where science is well-incentivized) replicate your own analysis in a second wave of data collection doing everything exactly the same. Psychology has actually gotten to a point where this is mostly how it works; many other fields have not.
He doesn’t mention it in this post, but in another post he talked about the toll of needing to frequently attend meetings in the middle of the night in his time zone.
Whatever his reasons for leaving, I hope that he finds a better balance in his new role.
This was the takeaway I had taking to a colleague about his time at Intel - they're a genuinely global company with engineering teams in practically all time zones who are still expected to collaborate with each other. No matter what time of day the meeting was scheduled for, it was the middle of the night for somebody, and no, just working on written docs async for everything didn't cut it, and they couldn't just fly people out all the time. So that's apparently just part of what it means to take a job at Intel these days.
Thanks for the mention. The philosophy behind Dusk is also eerily relevant to Chuck's problem at the moment. To quote my own manifest[1]:
When you operate a system, there is no problem that can arise that will make you powerless. Sure, you can have a hardware failure that hopelessly breaks your system, but at least you'll be able to identify that failure and know for sure that there is no software solution or workaround. That's control.
In this situation, of course Windows is to blame. But it could also happen with Linux, even if it's to a much much lesser degree.
If an update breaks your software in a way that is obscure enough to break only your software, then nobody else will fix your problem, and the system as a whole is too complex for you to dive in, making you powerless.
It looks like the initial post that this was a response to ended up flagged.
I don’t mean to accuse you of anything, especially since the signs are relatively subtle here, but this post and the initial one both show signs of having been edited by AI.
As a cultural matter, HN prefers you to not do that. It would be much better to write your posts using your own voice.
The article covers a variety of different approaches to dealing with high cocoa prices, but the Amsterdam brownie in the title is using a more heavily alkalized cocoa powder to maintain a similar taste while using less cocoa:
> The former gets its punch from using more heavily “dutched,” or alkalized, cocoa. It’s also what made that magical brownie taste so chocolatey.
If you buy dutch chocolate for baking you are told that this is actually significantly less flavorful, but a darker color, and useful for eg dark baked goods, when you're mostly trying to create a certain color shade without adding a ton of chocolate flavor.
It's inaccurate to say that dutched cocoa is less flavorful. Chocolate flavor is composed of multiple notes / experiences. Dutched cocoa has less acidic or fruity notes, but more dark rich, fudgy notes. Different, not less flavorful.
Do you mean a red flag for the quality of the article, or for the actions of the department? "Department of War" is currently a real name for the department:
> On September 5, 2025, President Donald Trump signed an executive order authorizing "Department of War" and "secretary of war" as secondary titles to the main titles of "Department of Defense" and "secretary of defense." The terms must be accommodated by federal agencies and are permitted in executive branch communications, ceremonial settings, and non-statutory documents. However, only an act of Congress can legally and formally change the department's name and secretary's title, so "Department of Defense" and "secretary of defense" remain legally official.[10][11] Trump described his rebranding as an effort to project a stronger and more bellicose name and said the "defense" names were "woke".[12]
No, it's not. As your quote says, the Department of Defense was created by Congress; the President has no authority whatsoever to rename it or designate a secondary name for it. Writing the words "executive order" on a document doesn't make it legally effective.
Any citizen, of course, can use whatever fake names they'd like for people or places or government organizations. It's a free country. But I don't see any reason to choose this particular fake name except for the purpose of delivering propaganda to your readers.
I'm 99 percent sure that executive orders are not at all legally binding. They're just ways of setting out policy. But the president does not have the authority to override congress with an executive order. An executive order can say whatever the president wishes.
It's difficult for me to evaluate the code quality, since I know C++ well but I don't know the conventions around using the R bindings. But from what I am able to tell the quality seems good. Really impressive for something vibe-coded.
I know that HN culturally frowns on video links and unexplained references, so to be explicit:
The seeping of the internet into the real world is an important theme of the anime Serial Experiments Lain, which is excellent, and if you generally like anime and resonate with the kind of stuff that people are bringing up in this subthread, then I recommend giving it a watch.
> One could specify a smallest effect size of interest and compare the plausibility of seeing the reported p-value under that distribution compared to the null distribution. 6 Maier and Lakens (2022) suggest you could do this exercise when planning a test in order to justify your choice of alpha-level
Huh, I’d never thought to do that before. You pretty much have to choose a smallest effect size of interest in order to do a power analysis in the first place, to figure out how many samples to collect, so this is a neat next step to then base significance level off of it.
reply