ak_yo

I find that this guide unhelpfully conflates probability and inference in a few places. Probability theory on its own is interesting but not terribly useful without the infrastructure of estimation.

This is an interesting argument but I think it fails to coherently restate what it means for race to be socially constructed. Nobody is denying that genetic differences between people exist and are important for health. They’re saying that reusing racial categorization schemes in genetics is scientifically ungrounded because these racial categories have no inherent biological content— they’re more like receptacles for political and social debate.

It’s not clear what value it adds to use (e.g.) US Census racial categories in genetics research. The salience of racial identification changes over time and is deeply politicized (as the author notes). More sophisticated and granular categories that are actually based on genetics would be much more appropriate than trying to recuperate categories that weren’t developed for science.

Cool tutorial, but I'm not entirely sure what makes this ML -- aside from neural nets, this is more or less the material you'd encounter in a basic applied statistics or regression analysis course, minus material on estimating uncertainty, modeling survival or time-series data, and causal inference. I suspect you'd benefit more from a 50 minute tutorial on those than neural nets.

Experimental design and observational causal inference would be excellent skills to have. Especially if you’re working with people who are asking you “why” questions, ML is helpful but isn’t going to cut it alone.

Great guide! Another reason to be familiar with this stuff is for data science: splines in particular are really useful for fitting a linear predictor where you suspect some of the marginal relationships are nonlinear, but you don’t have a strong prior on what the nonlinearity looks like.

This is right -- plus lm() is faster! Although, from a statistical perspective, if you can't invert X'X, that should first make you think "I have data quality issues" (i.e. multicollinearity) rather than "I need a different algorithm to compute the inverse".

I'd encourage you to think of this from the perspective of the person who's fearing for their life. It's rational to assume that the likelihood of someone potentially harming you is much higher if they're vocally expressing hatred about your identity.

That's totally fair -- and I appreciate that you're willing to engage seriously about these things!

Your last point is well-taken. It may even be the case these days that the majority of people who hold views similar to yours think the way you do. But I think we need to take into account the uncertainty that people in these situations face. It's hard to know what the other person is thinking -- and if all you know is that person's opinion, it's hard to know what's going on when you're not looking. Bigotry is often dressed up in talk and behavior that seems polite, even respectful at first glance, but that is ultimately materially harmful.

For example, I think a lot of LGBTQ people from religious backgrounds have had an experience of being told that their family or community will "love the sinner, but hate the sin" -- and then subsequently being subjected to unfair and harmful treatment (bullying, psychological abuse, ostracism, being disowned).

I think it's completely fair to think that a lot of these thoughts are coming from a good place. But the people who are espousing these perspectives need to understand that there's a lot of really intense history behind ideas like this. When these ideas have surfaced in the past, it usually hasn't really been about "fixing the cons" -- and even when it has, it often has major unanticipated and unintended consequences that cause serious harms.

I guess what I mean is that even when folks are talking about these things in good-intentioned ways, they're not appreciating the weightiness of the ideas they're throwing around -- and depending on your priors, it's reasonable to worry about that.

I agree, we're making the same general point -- it does seem crazy to ask someone to "ensure they hold no offensive opinions". So why, for example, should the opinions of the person who wants to speak their mind about the immorality of gay marriage get precedence over the opinions of the person who thinks they should be allowed to get married?

I believe this evidence is consistent with my point that a robust (and perhaps growing!) network of campus conservatism exists.

Specifically, this evidence could be explained by an increasing rate of conservatism on campuses. The rate of disinvitation could be the same, it's just that there's a greater number of conservative speaker invitations going out from a greater number of conservative students. And right leaning speakers are disproportionately invited by right-leaning student organizations.

From my perspective, I don't want to work in an environment where people are voicing their opinion that (e.g.) gay marriage is illegitimate or wrong. How am I supposed to work with someone who thinks a huge part of my life is immoral? I would have an incredibly hard time believing that that person was taking me seriously, really wanted to work with me, wasn't going to undercut me, or trusted me.

It's not that you can't have these opinions or voice them -- but it's also not the case that the people who are most affected by those opinions are going to feel OK about it.

You can definitely stop science. There's a reason we don't secretly infect populations with various diseases just to see what happens anymore.

I completely agree. The exchange seems to go like this:

A: "I think black people are genetically inferior." B: "I think that's a bigoted thing to say that has historically caused a lot of harm to a lot of people." A: "Why can't we have a free exchange of ideas?"

Person B is engaging with that idea, it's just that they think it's completely indefensible. I have a hard time understanding how Person A thinks this conversation is supposed to go -- like are we supposed to entertain everything as if it were serious? If I go up to a physics professor and say "I think physics is completely inferior and wrong" what is she supposed to say?

In what world are these opinions not voiced? I don't think this argument stands up to empirical scrutiny -- there's an incredibly robust network of vocally conservative student organizations.

I think you're exactly right. His argument is completely self-defeating: it's a perfect example of shutting down criticism on the basis of tone. He's just dismissing critique out of hand (because it's "politically correct") without actually engaging with its substance.

To be clear, I'm agnostic of the guy's motive -- I'm just genuinely puzzled how, if you spent most of your waking hours supporting AI startups and expanding the tech industry, you could possibly see that as "inevitable" and not something that requires a ton of hard work. Surely if this was all just inexorably happening on its own then everyone at YC would be out of a job?

I don't quite get why this is written as if the author is a neutral observer of this phenomenon, when in reality he works extremely hard every day to make sure it happens.

This is an interesting perspective but it doesn't really mesh with my experience. If I had to describe the typical political sympathies of HN, it would be "somewhat right of center on everything, except far left on tech policy" (e.g. strongly pro net neutrality [1], lukewarm on race/gender disparities in hiring [2]). Can you think of a big thread that drew a solidly left-leaning reaction on a social issue?

[1] https://news.ycombinator.com/item?id=14754772

[2] https://hn.algolia.com/?query=diversity&sort=byPopularity&pr...

This problem has been a structural feature of the academic profession for a long time -- Max Weber famously wrote about it almost 100 years ago: http://anthropos-lab.net/wp/wp-content/uploads/2011/12/Weber...

To tag on to this: I'd argue that work and school are more engaging than video games. When was the last time you just stopped going to work because it wasn't fun? Having to do these things in order to survive is extremely exciting, it just that the excitement doesn't feel good.

I'd be interested in reading evidence that supports a causal link between AAVE usage and segregation. To the best of my knowledge, no research has shown this to be the case, but I might be missing something. Did you have a particular study in mind?

The figure in this article* is taken (without citation) from a 2012 article by sociologists Michael Rosenfeld and Reuben Thomas.

(open-access preprint: https://web.stanford.edu/~mrosenfe/Rosenfeld_How_Couples_Mee...)

* Edit: The paper on ArXiv cites R&T properly, it's the MIT Tech Review piece that doesn't.

If you like this, you might also like Max Weber's essay on bureaucracy: https://www.law.upenn.edu/institutes/cerl/conferences/ethics...

> P=0.05 is best interpreted as "There is a 95% chance that this data isn't pure noise".

This is incorrect. A p-value is the probability of observing a value of the test statistic that is at least as extreme as the reported value, assuming the null hypothesis is true.

> Anyone looking at the presented data should be able to reason the result without needing to conduct a statistical test.

Part of the problem, though, is that scientific results are often counterintuitive, or (particularly in the social sciences) are easy to view as obvious after we know what the answer is.

One famous example: the sociologist Paul Lazarsfeld once gave a talk where he asked whether urban or rural men had adjusted more easily to the day-to-day routines of military service in World War II. There are equally obvious reasons why rural men would adjust more easily (e.g. they have more experience with physical labor, have experience with guns) or urban men would (e.g. they're used to crowded conditions, have more experience with a hierarchical division of labor).

I agree that IRBs are not great -- the book cited above does a great job of unpacking how IRBs fail to live up to their promises, and this covers some good additional examples. I also agree that the author isn't necessarily advocating against ethical review as such.

I'm more concerned with the author's insistence that he knew the right way to conduct his study ethically. Peer review is a core scientific principle-- we don't allow researchers to make autonomous judgments about the quality of their work.

IRB review procedures are flawed, I agree. But ethical review by a third party is enormously important for research with human subjects. I'm alarmed by the notion of researchers deciding unilaterally that a study has only "nonexistent risks" -- that's a recipe for disastrously unethical scientific conduct.

The author doesn't seem to take the history of medical research ethics (e.g. the Tuskegee syphilis experiments [1]) seriously. Having a sophisticated understanding of the ethical risks of research is a huge part of doing good science, and this post certainly doesn't communicate a high level of sophistication. If the author thinks that ethics review is "Blindly trusting authority to make our ethical decisions for us", then he clearly doesn't understand what ethics review is or how it works.

IRBs have well-documented flaws (for a more comprehensive look, read Laura Stark's book on IRBs [2]) but "IRBs are bad because they say I can't do whatever I want even though it's obviously the right way to do it" is not one of them.

[1]: https://en.wikipedia.org/wiki/Tuskegee_syphilis_experiment

[2]: http://press.uchicago.edu/ucp/books/book/chicago/B/bo1218257...

How does this compare to 2-gram Markov chains? The results look pretty similar (though perhaps a bit more grammatical...)