HNHacker News
TopNewBestAskShowJobs

croissants

1,228 karma · joined October 27, 2019

submissionscomments
croissants··on The Illustrated Guide to a PhD
My fairly generic PhD advice, as somebody who did one, graduated a bit early, and is happy about the process and where it's taken me:

* Choose your advisor with care. This is not very easy as an applicant looking at professors' websites, but if you are admitted, any good school will probably have an in-person or virtual admitted students day where you can talk to current students out of faculty earshot. Take advantage of these times to ask about your potential advisor. A truly bad advisor will probably produce at least one person who will warn you about them. If you can't do this in person, try to get a quick phone/video chat -- something off-record where they can be honest. I was always happy to do these for my advisor, because I liked him and wanted him to get more good students. Conversely, I know people who were warned off specific advisors during these events, for good reasons. A bit of subjectivity: a good advisor at a decent school is usually better than a bad advisor at a good school.

* The financial niceness of doing a PhD in field X seems to correlate pretty well with the current job market value of a masters in field X, at least partially for reasons of leverage -- if you can leave and transition into a cushier job, advisors have to provide a bit more value. Computer science scores highly on this metric.

* There is a ton of negativity about PhDs in places like HN. This isn't unjustified: doing a PhD with a bad advisor can be a very bad experience. At the same time, I think "person who had a bad PhD experience" is also "person who writes comments on the internet" with higher probability than "person who had a good PhD experience".

croissants··on Meta's memo to employees rolling back DEI programs
Do you have a link to this analysis? I'm curious what "condition on race" actually means.
croissants··on Yemeni Coffee Shops in Texas
I think the chain goes:

Homelessness increases --> any place where you can just hang out without paying attracts more homeless people --> there are too many homeless people hanging out for hanging out to remain good for business --> cheap hangouts close --> expensive hangouts like co-working spaces become viable.

croissants··on Jimmy Carter has died
Having not been born during his presidency, whenever I think of Jimmy Carter, I think of his appearance in Travels in Georgia [1], a John McPhee essay that's mostly about following a couple of workers for the Georgia Natural Areas council as they travel through and collect information about Natural Areas in the early 1970s. Jimmy Carter, then-governor, pops up at the end to be taken in a canoe down a river that the couple of workers want protected. Having basically nobody around to impress, Carter is nonetheless affable, open, curious, and generally good-hearted. The piece ends at the governor's mansion, like this:

A ball sat on the pavement. Before going in, we shot baskets for a while. "The river is just great," the Governor said, laying one in. "And it ought to be kept the way it is. It's almost heartbreaking to feel that the river is in danger of destruction. I guess I'll write a letter to all the landowners and say, 'If you'll use some self-restraint, it'll decrease the amount of legal restraint put on you in the future.' I don't think people want to incur the permanent wrath of the governor or the legislature."

"I've tried to talk to property owners," Carol said. "To get them to register their land with the Natural Areas Council. But they wouldn't even talk to me."

The Governor said, "To be blunt about it, Carol, why would they?"

The Governor had the ball and was dribbling in place, as if contemplating a property owner in front of him, one-on-one. He went to the basket, shot, and missed. Carol got the rebound and fed the ball to Sam. He shot. He missed, too.

[1] https://www.newyorker.com/magazine/1973/04/28/travels-in-geo...

croissants··on Should more of us be moving to live near friends?
This is in the category of "things I believe I read from a reputable source, but which I can no longer definitively attribute", but my understanding is that one contributor to Tokyo's lovely and specific density is that its zoning laws are pretty relaxed, and on top of that they are not very strictly enforced. This is maybe less workable in a society with higher variability in people's judgment of what is ok.
croissants··on More than 140 Kenya Facebook moderators sue after diagnoses of PTSD
A return to less centralized web publishing would also be bad for the many creators who lack the technical expertise or interest to jump through all the hoops required for building and hosting your own website. Maybe this seems like a pretty small friction to the median HN user, but I don't think it's true for creators in general, as evidenced by the enormous increase in both the number and sophistication of online creators over the past couple of decades.

Is that increase worth traumatizing moderators? I have no idea. But I frequently see this sentiment on HN about the old internet being better, framed as criticism of big internet companies, when it really seems to be at least in part criticism of how the median internet user has changed -- and the solution, coincidentally, would at least partially reverse that change.

croissants··on UnitedHealth's Effort to Deny Coverage for a Patient's Care (2023)
Thanks. Do you have some sense of the breakdown in actual expenses incurred by the basic research and subsequent industrial development? It may qualitatively be true that many governments sponsor basic research that is later productized, but if the second bit costs 1000x more than the first bit, it's not exactly a fairly shared cost.
croissants··on UnitedHealth's Effort to Deny Coverage for a Patient's Care (2023)
Where are you getting 10X? If I look at, say, Merck, their "sales, general, and administrative expenses" [1] trends at even or slightly higher than their R&D expenses [2]. Abbvie's ratio is around 2x [3, 4].

AFAIK, SG&A is the number that groups criticizing pharmaceutical companies look at, even though it includes a lot of money spent on things like paying lawyers, which any pharmaceutical company that actually wants to make back the billions spent on developing drugs will have to do.

[1] https://www.statista.com/statistics/269499/merck-and-co-sgan...

[2] https://www.statista.com/statistics/282738/expenditure-on-re...

[3] https://www.macrotrends.net/stocks/charts/ABBV/abbvie/sellin...

[4] https://www.statista.com/statistics/417068/r-and-d-expenditu...

croissants··on UnitedHealth's Effort to Deny Coverage for a Patient's Care (2023)
Do you have evidence for "most new treatment research is government funded (and by governments of other countries)"? My understanding is that the opposite is true: big pharmaceutical companies foot most of the bill for developing new drugs, then US-based insurers pay high costs, and most "other countries" enjoy much lower prices because US healthcare is effectively subsidizing their drug prices.
croissants··on Protecting undersea internet cables is a tech nightmare
I'm not sure what chatbots you're using, but the big free ones at this point can and, in my experience, typically do give valid code samples for a surprising range of tasks. I'm skeptical about LLMs as the ultimate future of AI, but they aren't useless.
croissants··on Protecting undersea internet cables is a tech nightmare
> For some reason people on HN imagine that

It's an easy simplifying heuristic to suppose that countries are all self-interested and nobody actually cares about "norms" or "war crimes" or "not filling low-orbit space with satellite fragments", because you get to forget about a lot of context and motivations and just put on your game theory hat and go, but like most simplifying heuristics that make things cleaner to think about, the conclusions this leads you to should be taken with a grain of salt

croissants··on Protecting undersea internet cables is a tech nightmare
This is the thing that actually worries me the most about AI. Right now there are meaningful practical obstacles to, say, manufacturing ricin or 3-D printing a working gun or figuring out how to kill somebody with a drone, and apparently these practical obstacles are enough to deter a nutso but insufficiently committed/intelligent/methodical would-be assassin, but if you have a chatbot that will happily talk you through the whole process, maybe that equilibrium changes.
croissants··on The Soul of an Old Machine: Revisiting the Timeless von Neumann Architecture
Another great Kidder book is "House", which applies the same perceptive profiling to the people involved in building a single house -- the couple, the architect, and the construction workers. There's an argument that a lot of the best work is done not in pursuit of an external reward but because doing good work is itself rewarding, and one way of viewing these books is as immersive explorations of how that plays out.
croissants··on How long til we're all on Ozempic?
Tangential, but

> 1: a CDC meta-review said that 26% of surveyed US trans students attempted suicide this year, N=20,103 surveyed, ~660 of which were trans. https://archive.ph/0H81G

This paper also finds that 5% of cisgender male and 11% of cisgender female students (out of ~8k surveyed for each) attempted suicide in the past year. It's kind of strange, because the age <=18 suicide rate (of "completed" attempts) is much smaller, approximately 1 in 5,000 to 1 in 10,000 [1].

[1] https://www.cdc.gov/nchs/data/vsrr/vsrr024.pdf, Figure 3

croissants··on Success depends on emotion management
> It doesn't say anything about the inverse.

Interpreted as a literal propositional statement, it's saying "if you have a good attitude, then you will have success". Therefore if you do not have success, it must be because you did not have a good attitude, which is your fault. It's the contrapositive, right?

croissants··on Covid-19 Revealed a United and Monolithic Media/Tech Complex
> Aspects of society that conventional wisdom holds to be outside or above the influence of tech and media, things like government, we suddenly realize are downstream from a single, monolithic technological communications machine that spans every screen and has its own biases.

> Covid-19 revealed that the Media/Tech Complex is capable of acting as a single unified entity if necessary. That makes them more powerful than the Roman empire ever was.

This article is long on ominous assertions about the incredible power of the "Media/Tech Complex" and short on actual evidence of that power. The main piece of evidence seems to be: look how they all said similar things at the beginning of the pandemic! But this seems like a weak example -- you may as well say getting a tornado warning on your phone and on the local TV news is evidence of the same thing.

The claim that Media/Tech is "monolithic" is also pretty weird to me. There are popular articles from big media companies tut-tutting "Big Tech" every week. They don't seem to be good friends.

croissants··on Police misconduct settlements
> we're not much different from a 3rd world failed state like Venezuela

In 2016, Venezuela's murder rate was well over 50 per 100,000, and the USA's murder rate was slightly over 5 per 100,000 [1]. This gap is pretty consistent.

[1] https://dataunodc.un.org/crime/intentional-homicide-victims

croissants··on Balloonomania
Another amusing event from the life of the balloonist Jean-Pierre Blanchard mentioned in the article: in 1793, Blanchard arranged the first American balloon flight, starting in Philadelphia. George Washington was interested enough to attend and write Blanchard a passport saying, roughly, "look, this guy is French and traveling in this unpredictable way, if you find him then know that I, George Washington, think he's cool, and you should help him get back to Philadelphia", which apparently worked [1].

[1] https://www.mountvernon.org/library/digitalhistory/digital-e...

croissants··on Crypto-Anarchism
> I'm not sure a group of people could hold strong for this long, given the price rises.

Isn't it possible that whatever person or group of people comprise Satoshi just decided to throw away the key, making willpower irrelevant?

croissants··on Crypto-Anarchism
A few years ago I read a book called Maximum City. The premise is "let's look at the lives of several people [including a powerful police officer, a dancer, low-level criminals, high-level criminals, very poor people just getting by and, remarkably, a young Hrithik Roshan] in circa-2000 Mumbai". I also happened at the same time to be within hearing distance of a few discussions that touched on crypto-anarchism.

The thought I remember having is that crypto-anarchism seemed to require being embedded in a well-functioning state, but that this assumption was kind of just waved away. I had this thought because so many of the people in Maximum City are just at the mercy of the slightly more powerful people around them -- paying protection money, not getting paid for work, paying bribes -- and there's not a well-functioning government that can help them. It seemed like said people would love a lot less anonymity ("hey! look! this guy is threatening to beat me up if I don't pay him every month! that's illegal!"), and would be well-served by tech infrastructure that makes them more visible.

croissants··on The Art of Reading More Effectively and Efficiently
The blog post's author lists a Ferriss book as one of his "3 favorite books" in a video on his channel, so it's more like "guilt by endorsement".
croissants··on The Art of Reading More Effectively and Efficiently
I like the idea of "synoptic" reading to learn about a concrete topic. It makes sense when different works offer different versions of what happens and why.

But it seems weird to do it with fiction. If I'm reading fiction, I want to get into the world created by the author. I think repeatedly stepping in and out to compare it to other works, or analyze its construction, or think about a GoodReads review, all of that is going to impede that process. So I usually just get into the book's world, finish it, and then step back for 30 minutes a day or two later and write a few paragraphs of my thoughts.

This little essay often includes connections to other books, and articulating thoughts on a book right after reading it contributes a ton to my appreciation and understanding. It makes it a lot easier to pitch the book to somebody else, for example (easier than saying "it's...it's just so good, just read it"). And for me this process works best if I refrain from doing it too consciously during the reading process. The experience of immersing your unconscious in something is pretty unique.

That said, I don't finish every book I start, and I don't write comments unless I feel like there's something to work out. But there's usually something to work out!

Unrelated pedantic comment: TFA attributes this quote to Edgar Allan Poe

> Marking a book is literally an experience of your differences or agreements with the author. It is the highest respect you can pay him.

But it doesn't sound like him, and the internet suggests it's actually a quote from this other Adler guy also referenced in TFA?

croissants··on Interest surges in top colleges, while struggling ones scrape for applicants
> From my own personal experience

In the other direction, my personal experience is that a pure math undergraduate helped me to structure assumptions, evidence, and conclusions in a way that I thought was sorely missing in most of my humanities classes. But I doubt either one of our personal anecdotes is all that convincing to a third party.

> More and more, and this thread re-enforces it, I think the STEM vs. humanities argument is veiled misogyny. White boys studying science = good, brown girls studying poetry = bad.

If you're going to jump to this uncharitable a view of my argument, it probably makes sense for us to just stop here.

croissants··on Interest surges in top colleges, while struggling ones scrape for applicants
It seems like this elides the difference between the goals of humanities education and the actual practice and outcomes of humanities education. I honestly don't know if humanities education confers the ability to "read closely, think clearly and critically, and synthesize information" at a rate that's so much higher than STEM education that it makes sense to divert tens of thousands of dollars to a humanities major instead of a STEM one.

Just as some naive first cut at answering this, I looked up GRE scores by intended graduate major [1]. "Physical sciences" (including math) majors average a 151 Verbal and "Humanities and Arts" average a...156, which seems pretty close, and even closer if you squint to try and account for the fact that the physical sciences skews English-as-a-second-language more than the humanities?

Of course, the GRE isn't a perfect proxy for the ability to "read closely, think clearly and critically, and synthesize information", but...graduate schools seem ok with it?

[1] https://www.ets.org/s/gre/pdf/gre_guide_table4.pdf

croissants··on Margaret Mitchell fired from Google
> Somehow Microsoft Research manages to be a much more open place than Google and has researchers publish critical pieces all the time.

Can you give examples of these critical pieces? I can't think of any, but MSR publishes a lot of stuff...

croissants··on Margaret Mitchell fired from Google
> This is super common in other professions sometimes charged with enforcing ethical and legal corporate behavior, including law and accounting. It's not a new idea.

I'd be interested in reading about guidelines for people tasked with "enforcing ethical and legal corporate behavior" in other industries, do you have any references?

croissants··on Who are the culture heroes of today?
Remember The Last Dance, the documentary about Michael Jordan's last season with the Chicago Bulls? Remember that weird talk show clip where the host describes Jordan as close to a living god? Now, Michael Jordan was an awesome athlete, but I think his mythos owes a lot to the specific media landscape at the time.

That is, media was extensive enough for a person to be a truly global star, but for the most part it only went one way. The media gave you an image of Michael and, for the most part, the image was pretty close to "living god". In contrast, modern media is intensely participatory. There are vicious comment sections and viral Tweets dunking on every person. Nobody seems to stay above it for very long.

There's a famous, maybe apocryphal quote where John Updike explains Ted Williams' decision to mostly ignore fans as "gods do not answer letters". But at this point, fame seems to require answering letters. I think people need to maintain some kind of distance to be the sort of heroes described in this article, but the culture we have seems pretty opposed to this kind of distance.

croissants··on Dismantling Racism in Mathematics Instruction [pdf]
> Supported by the Gates Foundation, so likely to be widely disseminated in schools.

Is it? The financial support seems to be $3 million in 2017 [1]. That year the Gates foundation spent roughly $380 million on "U.S. education" [2]. It seems hard to imagine that they plan on widely implementing the results of every single grant?

[1] https://www.gatesfoundation.org/How-We-Work/Quick-Links/Gran...

[2] https://www.gatesfoundation.org/Who-We-Are/Resources-and-Med...

croissants··on Who Should Stop Unethical A.I.?
> There's nothing that prevents a think tank from working with domain experts

One issue here is that the pool of domain experts depends on the different incentives faced by different researchers. If the top machine learning conferences all decide to reject papers about predictive policing, then this will reduce the incentive for machine learning researchers (who mostly want to publish in these conferences) to work in the area at all.

Maybe critics of predictive policing think this is good, or think that it's better for the field to restrict itself to the few computer science researchers who are prepared to seriously devote themselves to it? But there don't seem to be very many of them. And, as someone who thinks there is a lot of room for police departments to in general do a better job --- and I think most of the people arguing against predictive policing would believe the same! --- it's disappointing to me that the end result of this argument seems to be disincentivizing technical research on it.

croissants··on Who Should Stop Unethical A.I.?
I'm with you that a lot of mass media writing about AI is silly, but I think this is article is not in that category. For example, it doesn't "aver [Speech2Face] is transphobic", that's from a statement by Alex Hanna, and that statement is immediately followed by comments from other people questioning the statement, and the piece in general is pretty even-keeled about giving space to criticisms and responses.

I think the article paints a good picture of the machine learning research community figuring out how to grapple with the growing number of people who want to probe its ethics, without portraying anybody as a villain.

← PreviousPage 2 of 9Next →