Hacker News Reader

apathyon Sci-Hub as Necessary, Effective Civil Disobedience

I send her $5 a month and it's 1000x more valuable than HBO go or any of that sort of shit.

She is quite simply a hero, full stop.

apathyon Every neuron potentially has a different genome than those it's connected to

God damn it, that Science paper is piss weak even for a Science paper. It's yet another consortium advertisement. Meanwhile mapping of somatic mutations in blood progenitors has been happening for a decade.

apathyon Every neuron potentially has a different genome than those it's connected to

https://www.nature.com/articles/ncomms12484

the more closely you look, the more obvious it becomes that this is the rule rather than the exception. I would be somewhat surprised if children have functional mutations rampant between neurons, and I suspect that some fraction of this is artifactual. But I have no doubt (does anyone?) that some degree of somatic mosaicism is the rule. About the only cells that tend to hang around much longer than neurons are blood stem cells, and as soon as you look closely at those, it's all but unavoidable.

apathyon Empirical Bayes for multiple sample sizes

> Author here.

Alright, I'll put on my Reviewer Number 3 hat and say that I learned some neat things from your work, including that the National Swine Improvement Federation. I'll try and do a halfway decent job here.

> I would love if an expert could expand on the relationship between mixed effects and Empirical Bayes.

A real expert? Here you go:

http://statweb.stanford.edu/~ckirby/brad/LSI/monograph_CUP.p...

Read it, all of it, but particularly chapter 1, section 2.5, and chapters 8, 10, and 11. Why does testing, effect size estimation, and high-dimensional analysis have anything to do with anything? Because...

1) independence is largely a myth 2) you are likely to have multiple ratings per reviewer on your site, whether your generating distribution is nearly-continuous (0-10, mean-centered) or discrete (0/1, A/B/C). If you discard this, you are throwing away an enormous amount of information, and failing utterly to understand why a person would estimate not just the variance but the covariance even for a univariate response.

The second point is the one that matters.

Also, "empirical Bayes" is in modern parlance equivalent to "Bayes". What's the alternative? "Conjectural Bayes"? (Maybe I should quit while I'm ahead, pure frequentists may be lurking somewhere)

> I welcome your feedback though on how I could make this point clearer.

For starters, edit. Your post is too damned long.

Think about where you are getting diminishing returns and why. Is there ever a realistic situation where your ratings site would not keep track of who submitted the rating? (It's certainly not going to be an unbiased sample, if so; the ballot box will get stuffed) So if you have to keep track of who's voting, you automatically have information to decompose the covariance matrix, and everything else logically follows.

A univariate response with a multivariate predictor (say, rating ~ movie*rater) can have multiple sources of variance, and estimating these from small samples is hard. When you use a James-Stein estimator, you trade variance for bias. You're shrinking towards movie-specific variance estimates, but you almost certainly have enough information to shrink towards movie-centric and rater-centric estimates of fixed and random effects, tempered by the number of ratings per movie and the number of ratings per rater. (Obviously you should not have more than one rating per movie per rater, else your sample cannot be unbiased).

I think you will return to this and write a much crisper, more concise, and more useful summary once this sinks in. I could be wrong. But you'll have learned something deeply useful even if I am. I do not think you can lose by it.

apathyon Empirical Bayes for multiple sample sizes

Thank you. As a statistician, the fact that mixed effects models (e.g. does this rater tend to rate high?) are overlooked is, IMHO, a death sentence. Too much nomenclature, too early (link to the table within the text, please, and omit needless words), and with too little attention paid to the value of an external citation.

Also, MCMC for ratings? Surely you jest. If the author had touched on mixed models, then maybe it would make sense. But given the sample sizes involved here, and the noise in the variance estimates, I recommend that the author investigate mixed models tout suite if they do in fact care about the sources of shared and unshared effects on variance. Because that is what mixed models do.

apathyon “A closure is a poor man’s object; an object is a poor man’s closure” (2003)

The visual here is awesome. Although all good programmers are antiheroes in my world, so I read that as a mustache twirling Guy Steele merrily evading the bumbling cops. Practically the whole point of Real Programming is to evade the limits of The Man. (Cf Mel: http://www.catb.org/jargon/html/story-of-mel.html )

apathyon CRISPR eliminates HIV-1 infection in live animals

under Bayh-Dole I would expect them to get 33%-33%-33% or 60%-40%, something like that? That's how it works for mine.

apathyon “A closure is a poor man’s object; an object is a poor man’s closure” (2003)

I did work with a guy named Max Kalashnikov at Google. That was a pretty "rad" name, too.

apathyon Award-Winning Nautilus Enters Rough Waters

The sole comment as of 11:26AM PDT, April 29th, 2017, is devastating. It suggests that science journalism is fucked. I'm going to poll a few friends that do freelance work for Nature & Science to see if this is the general consensus.

apathyon Mice treated with a protein from umbilical cord plasma showed improved memory

nb. Here is the link to the article:

http://www.nature.com/nature/journal/vaop/ncurrent/full/natu...

Here are the figures:

http://www.nature.com/nature/journal/vaop/ncurrent/fig_tab/n...

If you need a fulltext link that can be arranged, but it's not a bad idea to understand how DOIs work. Here is one:

doi:10.1038/nature22067

This is essentially a URI; the piece after the : is a URN, which is meant to uniquely identify a piece of published research (now expanded to include preprints and software).

To route it, you'll need a URL which your browser can resolve. Happily, if you visit

http://dx.doi.org/10.1038/nature22067

you will return to the paper. It is meant to provide a durable identifier, so that even Elsevier journals can't reliably hide the identity of a publication. What this also means is that "unlocking" services like Elbakyan's can use it as a primary key. Suppose you paste the URN part of that DOI onto http://sci-hub.cc/ or http://sci-hub.bz/ to complete the URL. What do you suppose happens?

(This is left as an exercise for the reader, and of course I cannot and do not condone violations of copyright law. So don't read whatever might pop up, ok?)

apathyon Mice treated with a protein from umbilical cord plasma showed improved memory

> I generally trust Nature

That's your first mistake. Don't take my word for it:

http://www.nature.com/nature/journal/v483/n7391/fig_tab/4835...

Glamor journals are glamorous because of their "impact" (people read and cite them), and that "impact" occurs because they publish sexy (not necessarily replicable or useful) findings. Note that Nature is not a bad actor per se, and in fact their editors have completely destroyed my lack of faith in peer review lately (as an author). But the incentives for CNS journals skew in favor of "sexy" rather than "scientific", and the general tendency of NIH to fund "sexy" means that almost all research in glamor journals must be treated as suspect until replicated, ideally in a clinical trial or by an un-incentivized third party lab.

> I can only access what appears to be more of an abstract

There's a site called http://sci-hub.bz/ that can help.

> these claims could be anomalies.

You are correct, and you should heed your instincts. Extraordinary claims require extraordinary evidence, and a single experiment or replication is one (possibly quite principled) sample from a distribution of potential outcomes. Don't invest yet ;-)

Side note: it seems like numeracy and statistical thinking is becoming the default in an educated population. You, too, are destroying my lack of faith in humanity. Thanks.

apathyon Programmers are most likely to work from home

My god this sounds exactly like academia: your job as a professor at a research university is to obtain grants, publications, and donations. Pretty much everything else is incidental. "Service" to a department, to students, etc. is "emphasized" in reviews, but if push comes to shove, the asshole with a bunch of big grants is going to steamroll the "nice" junior faculty who "mentors" the asshole's students. If you're teaching, preparing to teach, or holding office hours, you're not writing grants or manuscripts. And the odds are that everyone will say "thank you so much!!" as you get turned down for tenure, UNLESS you managed to satisfy your true primary responsibilities so efficiently as to leave time for the "fun" stuff (teaching, watching people discover, blah blah)

There's a balance -- you can do more with more hands, and they need to be well trained hands. But if they're not your direct reports, ultimately they're not your problem. (This also means that you will need to be responsible for the outcomes of those you mentor, and you'll need to learn how to let slow learners go. Oh noes, you're becoming management! It's almost as if there's a reason that good managers with domain expertise are hard to find.)

I guess it comes down to, are you going to be a "helpful" "mentor" or are you going to accept formal responsibility for the success or failure of mentees (vis-a-vis grants, contracts, successful projects, etc.)? Instead of being a martyr, it seems that explicitly accepting or denying responsibility for people who eat up your time (within reason) is the best long-term strategy. You can still succeed without doing this, and you can still fail while doing this, but the odds are in your favor if you demand accountability. (Also, there's nowhere to hide if you aren't as good as you think you are; I think that's a feature, rather than a bug, but opinions may vary.)

apathyon Programmers are most likely to work from home

> It doesn't sound like humbleMouse is getting paid to literally write code. It sounds like they're getting paid to write code and mentor junior developers.

As a contractor? I have to say I find this doubtful. Without seeing the contract, neither of us can say for sure, but I'd be surprised if it specified mentoring.

apathyon Programmers are most likely to work from home

perhaps you're working for the wrong boss. Every effective boss I've ever worked for wanted to see two things:

1) revenue 2) results

Those who got distracted by other crap tended not to be as successful by objective metrics. Just my experience though

apathyon Why We switched from Python To Go

His grammar is atrocious, yet this is one of the clearest and most persuasive things I've ever seen on medium.

I learned an important lesson about writing from this author.

apathyon Python vs. R: The battle for data scientist mind share

Both are nice and pleasant guys to work with, unlike a lot of the drooling idiots that act like this "rivalry" is some sort of football game. Funny how master craftsmen don't often blame their tools, they sharpen them instead.

Any time someone blames their tools for their own inadequacies, show them this video of Kelly Slater surfing better on an overturned table than most of us can surf on a 7 foot three fin board: https://m.youtube.com/watch?v=XQ4owd3yQ_4. Up your game instead!

Edit: hacker news doesn't use that part of markdown

apathyon Python vs. R: The battle for data scientist mind share

Statisticians need to make figures; most programmers, most of the time, don't. The tools reflect this :-)

apathyon Python vs. R: The battle for data scientist mind share

Let me guess you work in finance

apathyon Python vs. R: The battle for data scientist mind share

R is explicitly designed around S-expressions and as such lends itself to domain-specific languages like these. The choice was not an accident either by the original Rs (Robert Gentleman and Ross Ihaka) nor by Hadley.

Guido has explicitly stated that he does not want Python to be "more lispy" e.g. in regards to lambdas (asterisk). Thus I've seen many people even at, say, Stanford, Harvard, and Cambridge going back to R from Python. Sometimes there does not exist a language that best suits a workflow, and a DSL works better. That is where lispy languages hold an advantage.

Use the right tool for the the job, imho, but I fucking hate people that mix the two within a project intended for wide public release. Worst of both worlds, again imho.

(Asterisk) apparently functional data structures such as iterators and generators are OK though. Wtf guido

apathyon Denouement

Thank you for doing this -- all of it. It seems like you've built up a tremendous amount of goodwill and I hope it will pay off for you, in what seems like a much more focused new venture. If you can capture 10% of Craigslist's reach without the sketch, you should be OK :-)

apathyon A blog post leads to doubts about Cornell’s food laboratory

Ah, OK. I think we are on the same page. My mistake.

apathyon A blog post leads to doubts about Cornell’s food laboratory

Preregistration, as demanded for clinical trials, is essentially the only way to ensure that researcher degrees of freedom are avoided.

Buy the ticket, take the ride. Needless to say, it's a scary ride, so most groups won't do it.

apathyon A blog post leads to doubts about Cornell’s food laboratory

I love how you believe that most reviewers have time to conduct a thorough review for a journal, most of the time.

As opposed to something much closer to "LGTM!" With some "but be sure to cite my vaguely-relevant work in Foo et al, 1952".

apathyon A blog post leads to doubts about Cornell’s food laboratory

So, peer review doesn't scale, and some alternative is needed.

Sounds about right. How about we crowdsource the task? Since all the qualified specialists are busy writing grants against a 15% payline, we can farm it out to people without those pressures.

Oh... wait...

apathyon A startup is filling SF's empty luxury rentals

40 years later and the kids just never learn...

http://www.theverge.com/2013/11/4/5063258/ponte-tower-africa...

Oh but that would never happen here, Bay Area landlords and developers are all on the up and up, yada yada...

Full disclosure: I lived in a great little 3-bedroom in the mission from 1999-2001 where we each paid $650. And it was fuckinh awesome. The place "mysteriously" burned down a few years later, coincidentally releasing the owner from rent control. Who'd a thunk it?

apathyon A researcher's response to people pointing out 150 errors in four of his papers

What does any of this have to do with politics?

Wansink doesn't appear to be troubled by his prodigious contributions of unreliable results to the literature, nor does he appear inclined to correct them. All of science and discovery is a joke to him.

Left or right, the troubled extremes of politics start where respect for evidence and truth ends. Scientists are, nominally, people who search for truths about our universe with some degree of rigor, some respect for evidence, some appetite for truth (and usually an insanely competitive nature, but I digress). Absent that, what's the point?

This isn't about society, in my opinion. It's about having some respect for evidence and truth. Without these, science isn't science. And what wansink is doing isn't science. Cold fusion had more rigor. Yet he will probably be rewarded with grants of taxpayer money, taken from fellow citizens as taxes, under threat of force. Meanwhile pediatric bone marrow transplanters will shut down their labs for "lack of funding". Or, more accurately, lack of standards in science.

Yes, it affects people when dishonesty is accepted.

apathyon Intel’s first Optane SSD: 375GB that you can also use as RAM

wow, and here I thought that SSDs were disks made of RAM

apathyon Improving Bayesian Reasoning: What Works and Why?

Oh christ, another Frontiers in Briefing Trends series. These are among the most superficially reviewed journals on earth. Navarrete must have been up for tenure and needed a few publications (deans can't read, but some can count).

apathyon High quality graphics for Latex documents: The Matlab route

Why do you quote the word experts?

It doesn't appear that the objective merits of the tools under discussion yield a single best answer. For example, if you look through various implementations of assorted somewhat-esoteric algorithms & formulae, I have found that the FORTRAN implementations are usually cleanest. Not because the language is cleaner or anything like that, but because the most experienced people in the field didn't see the point of bowing to language fashion.

If fashion were the deciding point I suppose we'd all be compelled to use JavaScript, so, yeah. I'd rather not reimplement something that already works right, least of all when it's a fiddle problem and I've got deadlines to hit.

apathyon Ask HN: Which Berkeley Courses Should I Archive?

you and the above poster are heros.

    $ youtube-dl
    The program 'youtube-dl' is currently not installed.
    You can install it by typing:
      sudo apt install youtube-dl

    $ sudo apt install youtube-dl

Not many things I love more than discovering that someone wanted the program I'm about to write, and wrote it better than I'd have had time to. This is why I still love the Linux/BSD ecosystem after all these years