Hacker News Reader

rflrobon Facial expression software confirms bronze medalists happier than silver

In this case, however, there is evidently prior research using humans judging the faces that shows similar results. There definitely are all kinds of biases that could creep in from using facial recognition software, but a well designed study will attempt to quantify and control for those biases.

I definitely agree that the idea that software eliminates bias is laughable. Swaps it for another, hopefully less extreme set of biases doesn’t draw as many clicks, though.

rflrobon Lachlan Morton completes 5,510km Alt Tour, beating peloton to Paris by five days

As I heard it, I’d you want to live to 80, you should eat right, exercise regularly, and not smoke. If you want to live to 100, you need to have the right grandparents and can live however you like.

rflrobon This Chemical Does Not Exist

This chemical could probably be made but doesn’t exist seems most accurate.

https://www.chemistryworld.com/opinion/chemical-space-is-big...

rflrobon Useful and useless code comments

I find the choice of the phrase “¿Como te llamas?” fascinating as evidence against “don’t do translation word for word in your head. ” Taken literally, “how (como) you (te) are called (llamas)” has no words in common with “what is your name”. The comments in the code would allow someone with a good Spanish vocabulary to actually translate into fluent English.

rflrobon Need Something Sorted? Sleep on It

My first CS teacher talked about what he called Bang Sort, which is also O(n):

1. For each entry in your list, cut a straw of length proportional to the value to be sorted

2. Take all your straws in a bundle

3. Bang them gently on a flat table

4. Draw out the straws in order of length, each operation of which can be done in O(1) time

rflrobon CRISPR Editing in Primates

This is really program dependent. The NSF actually publishes statistics on all kinds of aspects of degree programs. Life Sciences (which doesn't have the reputation as a super speedy science program) has a median time in PhD program of around 5.5 years. Obviously the distribution is going to be skewed, but some fields have had a push towards really reducing the time to graduation.

I'm all for doing serious introspection before (and during!) a PhD, but the comment is way too short to set off many alarm bells for me.

https://ncses.nsf.gov/pubs/nsf21308/data-tables#group7

rflrobon Texas wants to charge Tesla and other EV owners annual fees

https://www.ev-volumes.com/country/usa/

Teslas make up the plurality, and often majority, of electric vehicles sold in the US.

rflrobon “They introduce kernel bugs on purpose”

But Kanjie Lu, Pakki’s advisor, was one of the authors. The claim that “ You, and your group, have publicly admitted to sending known-buggy patches” may not be totally accurate (or it might be—Pakki could be on other papers I’m not aware of), but it’s not totally inaccurate either. Most academic work is variations on a theme, so it’s reasonable to be suspect of things from Lu’s group.

rflrobon mRNA vaccines – a new era in vaccinology (2018)

We actually know the primary sequence (i.e. order of amino acids) for every gene in the body. It’s trivially easy (a senior undergrad in bioinformatics ought to be able to do it with little or no handholding) to look for genes with even a vaguely similar sequence. Protein folding is a hard problem, but the space of possible solutions is large enough that it’s unlikely that an unrelated sequence would have a similar enough structure to be recognized by the same antibodies.

Most variation between populations is not so much what genes are there, but how at what level they are expressed, point mutations, and short repeats (all of which are unlikely to create totally new structures). For there to be a population that had spike protein expressed, there would have to have been some sort of horizontal gene transfer into the germline (which is relatively protected from mutations for precisely this reason) at some point in the past. There’s reasonable atlases of human genetic variation (see, for example, the 1000 genomes project), so if there was some large population of people with any obviously new viral sequence, we’d see it.

Plus, if a person somehow, miraculously, had a protein that looked like Spike, their immune system should already be tuned to ignore it. (I have a PhD in biology, but immunology was not something I have much expertise in, so take this with a grain of NaCl).

rflrobon Covid vaccination linked to 'substantial reduction' in hospital admissions

I'm not sure that a simple sliding correlation really captures how treatments, protocols, and behaviors have changed over time. Leaving aside the winter holidays case peak (which is much more multi-modal than the others), I see two peaks:

* A peak of cases around Apr 11, followed by a peak of hospitalizations on Apr 22, with a peak of deaths also around Apr 22.

* A peak of cases around Jul 22, followed by a peak of hospitalizations around July 26, followed by a peak of deaths around August 4.

If I were going to do a more detailed analysis, I would want to try breaking out individual states/counties (subject to some reasonable population minimum), such that multiple distinct trends nationally don't interfere with each other in the data.

rflrobon I am a heroin user. I do not have a drug problem

> I'm not saying our drug laws are just or can't be reformed but the fact that heroin is illegal prevents a LOT of people from using it. How many people will then use, aided by these pronouncements that "Heroin isn't actually that addictive!" and ruin their lives?

Do you know a lot of people who have expressed to you that they would use heroin recreationally if only it were legal? My gut is that the number isn’t large.

My impression of the evidence from places that have tried decriminalization (eg Portugal) is that the number of people who use and then are able to get their lives back on the rails is higher than the extra people who use because it’s decriminalized.

rflrobon How to solve the Secret Santa Problem using graph theory

I think it can as well, unless you require constraints as in the article.

rflrobon How to solve the Secret Santa Problem using graph theory

Maybe I’m not clever enough, but wouldn’t the naive implementation of the naive algorithm be O(n^2) operations? Depending on whether you have a linked list or an array, traversing to or removing a random element takes O(n) time, which you need to repeat n times.

rflrobon SARS-CoV-2 501Y.V2 escapes neutralization by South African COVID-19 donor plasma

It is a different variant, but both the 501Y.V2 ("South African lineage" tested in OP) and B.1.1.7 ("UK lineage" tested in the Pfizer study you linked) have the N501Y spike mutation, which (I gather) is the residue of particular concern.

rflrobon SARS-CoV-2 501Y.V2 escapes neutralization by South African COVID-19 donor plasma

Given that most of the vaccines were produced "at risk" (i.e. before the trials showed that they were effective), I doubt that if the FDA had approved these vaccines on the day they were designed, we would have significantly more doses in people's arms today.

rflrobon Stromatolites: A Billion-Year-Old Living Organism

How “old” is a single-celled organism? I don’t know the precise details for Cyanobacteria, but it could be a relatively equal split (unlike budding, a perfectly reasonable alternative). In that case, are both offspring of a cell division the age of the “parent”? If so, why stop at the time since one generation? “But wait”, you may complain, “they are constantly exchanging molecules with their environment!” So is the “age” of the organism the point at which none (or equivalently half or any other fraction you choose) of the atoms inside the organism were inside the organism previously? That implies you could watch a single, discrete cell not dividing, and it’s “age” would be constantly fluctuating, not simply monotonically increasing. It’s entirely possible that age is not a meaningful, well-defined concept here.

But less philosophically, when people refer to “living fossils”, they generally mean that the organisms haven’t changed much in form or physiology in a long time. There’s no reason to believe that they aren’t subject to the same pressure to adapt to changing conditions via natural selection, but those adaptations aren’t readily visible.

rflrobon BioNTech CEO applies Covid-19 vaccine's mRNA tech to multiple sclerosis

I'm sure they would, but bear in mind that the costs are likely going to be much higher for an MS vaccine than for nCov2, due to the economies of scale that the latter has. Insurance companies will still probably love a one-time $20,000 (or whatever number they can justify) treatment compared to ongoing costs.

rflrobon Slack is down

I'm still logged in on mobile and can communicate with people from my team, but cannot log in from desktop. With so few people able to connect, it's also unclear whether Slack is eating my messages or there's just no one to respond. So I'd certainly rank that as "trouble using slack" rather than "the system is completely down".

rflrobon Science Is Getting Less Bang for Its Buck (2018)

I always thought that if I ran my own lab, I would have rotation projects be to replicate a recent result from the lab. Then, a lot of the infrastructure and support would still be there, so it would be pretty clear if the fault lay with the experiment. Plus, it would reinforce the notion among trainees that the point of science is to be replicated, and that the hard part of doing something novel is figuring out what to do, not actually doing the work.

rflrobon No one who got Moderna's vaccine in trial developed severe Covid-19

There’s a lot of reason to suspect that, once the pipeline is set up and validated, tailored RNA will be a lot cheaper than CAR-T. cell culture is hard, whereas ordering a gene length DNA construct is “only” a few hundred bucks[1]. From there, getting it to mRNA is relatively straightforward. Doing this all medical grade and at scale will be a huge challenge, but should be significantly easier than mass producing CAR-Ts.

[1] https://www.genewiz.com/Public/Services/Gene-Synthesis/Turbo...

rflrobon Every possible pun of the form ‘You put the X in Y’

I noticed there's a few cases where a th sound puns with a t sound. For example, you put the OUGHT in ROTHMEYER. Intentional due to loose matching? Uncertainty in pronunciations? Or is this a bug where the θ phoneme is represented as the digraph th?

rflrobon SARS-CoV-2 positivity rates associated with circulating Vitamin D levels

Or even that people who take supplements are more likely to get a “just in case” Covid test than someone who doesn’t supplement.

rflrobon SARS-CoV-2 positivity rates associated with circulating Vitamin D levels

A properly designed, well powered, double blind randomized controlled trial does a much better job at establishing causation than a retrospective correlation study, though.

rflrobon Journalists aren’t the enemy of the people. But we’re not your friends either

Ronan Farrow’s investigation of Harvey Weinstein was nearly squashed by NBC, where he worked at the time. He had to take it to The New Yorker.

rflrobon New 3D depigmentation technique enables high-res imaging of intact organisms

I’m looking for a detailed checklist that a reasonably well trained scientist in the field could follow. This should include equipment they used, and guidance on where variables should be adjusted for sample size.

See https://www.protocols.io/view/human-colon-tissue-clearing-an... and compare to the amount of detail even in the supplemental methods.

Again, I think the concision and “refer to previous methods” approach is mostly* fine for discovery papers, but if the method is the thing, I want all the gory details in one place.

* but also, academics hate chasing down chains of references, some of which are in journals their university doesn’t subscribe to.

rflrobon New 3D depigmentation technique enables high-res imaging of intact organisms

It should be considered malpractice to publish an article about a new technique and not include a detailed protocol for how to do it. It’s understandable if the article is trying to show some effect, develops the new technique as a means to that end, and then publishes the result they found, but in cases where the article is “we can now do this in whatever organism we want”, not telling people how they can do it themselves holds science back.

rflrobon Trump administration establishes $75M quantum computing centers

To zeroth order, it works out to roughly 30 small research groups for 5 years.

At a place like Berkeley, where I did my PhD (but not in anything quantum related), a PhD student and a postdoc both cost a professor something like $75k/year in salary and benefits (money not spent on postdoc salary basically goes to student tuition). Add in the professors salary as well (which is often substantially supported by grants), figure $150k/year for the prof. So in people costs alone, let’s say 1 professor and 2 trainees, that’s about $300k/year, or $1.5M/5 years. Multiply by 30 professors supported, and you’re at $45M, which is 60% of the $75M. And yeah, this is assuming that experiments are free.

Now is 20-30 professors a lot? If sufficiently narrow and esteemed, absolutely. There were 29 people at the 1927 Solvay Conference. Give them an extra couple months where they aren’t sweating a grant, and that’s a lot of time to think.

rflrobon Research database explores racial bias in police shootings (2019)

Shootings are both extreme enough and infrequent enough (though still all too frequent) that one could plausibly assemble a nearly complete national dataset, plus they are more likely to get covered in the news/on social media. I’m not familiar with the literature, but I’d bet there are dozens of high quality (though more limited) studies on the racial disparities in the interactions you mentioned for each study on Police killings.

rflrobon Ask HN: Tips for students before starting their first job

The thing is, most romantic relationships fail. If it’s someone you met online or even a friend of a friend, you likely won’t see them that often after you break up, which is decidedly not the case for a colleague. So while it might work out with a colleague, and the odds aren’t even that bad, more caution is warranted.

Of course, it’s probably not terrible advice to tell everyone in a breakup to treat their soon-to-be ex as if they will one day be your manager. It’s just not likely that everyone will be able to follow through.

rflrobon Let's hear scientists with different Covid-19 views, not attack them

Reading the tweet thread linked under "personal attacks and general disparaging comments" (https://twitter.com/CT_Bergstrom/status/1253822258692284416), I see much more a criticism of poor science journalism in the WSJ than the poor science in the article itself (aside from one swipe as a "methodological piece of shit"). I'm all for rigorous debate of different scientific ideas, but the WSJ Opinion page is not where people should be marshaling allies to their scientific ideas that are regarded as tenuous by the majority of the community.