Model Failure

This may be hard to believe, but I’m not about to talk about Bayesian modeling nor CompSci. Nope, I got dragged into an argument over implicit bias with a science-loving “skeptic,” and a few people mobbed me over the “model minority.”

Asian-Americans, like Jews, are indeed a problem for the “social-justice” brigade. I mean, how on earth have both ethnic groups done so well in such a profoundly racist society? How have bigoted white people allowed these minorities to do so well — even to the point of earning more, on average, than whites? Asian-Americans, for example, have been subject to some of the most brutal oppression, racial hatred, and open discrimination over the years. In the late 19th century, as most worked in hard labor, they were subject to lynchings and violence across the American West and laws that prohibited their employment. They were banned from immigrating to the U.S. in 1924. Japanese-American citizens were forced into internment camps during the Second World War, and subjected to hideous, racist propaganda after Pearl Harbor. Yet, today, Asian-Americans are among the most prosperous, well-educated, and successful ethnic groups in America. What gives?

What gives is simple demographics. Take it away, Jeff Guo of the Washington Post: [Read more…]

Fake Hate, Frequentism, and False Balance

This article from Kiara Alfonseca of ProPublica got me thinking.

Fake hate crimes have a huge impact despite their rarity, said Ryan Lenz, senior investigative writer for the Southern Poverty Law Center Intelligence Project. “There aren’t many people claiming fake hate crimes, but when they do, they make massive headlines,” he said. It takes just one fake report, Lenz said, “to undermine the legitimacy of other hate crimes.”

My lizard brain could see the logic in this: learning one incident was a hoax opened up the possibility that others were hoaxes too, which was comforting if I thought that world was fundamentally moral. But with a half-second more thought, that view seemed ridiculous: if we go from a 0% hoax rate to 11% in our sample, we’ve still got good reason to think the hoax rate is low.

With a bit more thought, I realized I had enough knowledge of probability to determine who was right.

[Read more…]

This Rings A Bell

Heavyweight tech investor and FDA-critic Peter Thiel is among conservative funders and American researchers backing an offshore herpes vaccine trial that blatantly flouts US safety regulations, according to a Monday report by Kaiser Health News.

The vaccine—a live but weakened herpes virus—was first tested in a 17-person trial on the Caribbean Island of St. Kitts without federal oversight or the standard human safety requirement of an institutional review board (IRB) approval. Biomedical researchers and experts have sharply rebuked the lack of safety oversight and slammed the poor quality of the data collected, which has been rejected from scientific publication. However, investors and those running the trial say it is a direct challenge to what they see as innovation-stifling regulations by the Food and Drug Administration.

It was around that point in Beth Mole’s article for Ars Technica that I got a sense of deja-vu. A quick Google search confirmed it was more than a feeling:

Biomedical research, then, promises vast increases in life, health, and flourishing. Just imagine how much happier you would be if a prematurely deceased loved one were alive, or a debilitated one were vigorous — and multiply that good by several billion, in perpetuity. Given this potential bonanza, the primary moral goal for today’s bioethics can be summarized in a single sentence.

Get out of the way.

A truly ethical bioethics should not bog down research in red tape, moratoria, or threats of prosecution based on nebulous but sweeping principles such as “dignity,” “sacredness,” or “social justice.” Nor should it thwart research that has likely benefits now or in the near future by sowing panic about speculative harms in the distant future.

That was an ill-informed opinion of Steven Pinker from two years ago. I took it to task back then, but I wonder if Pinker has changed his mind in the intervening years. I checked his Twitter feed, and came away empty. Stick a pin in that one, it may become interesting.

Activist Self-Protection

The hosts of Feminist Killjoys outdid themselves with their latest episode, when they interviewed a member of “Redneck Revolt,” an AntiFa group. The conversation was pretty one-sided and animated, but you get a great summary of what they do.

00:08:13,760 –> 00:08:50,120
… we were asked by anarchist people of color to go and defend Justice Park. Our mission in Charlottesville was purely defensive. We never moved – and I want to make this really clear, and I hope this message gets out – we never moved beyond a very fixed perimeter. We were highly disciplined, we had a clear mission: keep people safe, keep the state and the Nazis out of the park. [We were] successful, partially because 1) we were asked to be there, so we knew who had our back and who wanted us there and 2) we knew what was to our front, the state and the Nazis.

00:08:50,120 –> 00:09:31,040
We never mixed into the larger protest, and there’s been some discussion, I think, out in the internet world that “yeah, we’re just wandering around with guns.” I mean, we’re not operators – this isn’t SEAL team 6 cosplay. We kept our muzzles down, and we wanted to project the force and power that not only our group possesses, but what we knew was streaming behind us and through us: as AntiFa columns, groups of Quakers marched- BLM folks moved- queer liberation activists… all these people move through our line to go and face down white supremacy.

00:09:31,040 –> 00:09:41,780
White supremacists came to face us, but we were in complete concert with the people that were deploying other tactics, and that again is an enormous power that really can’t be underestimated.

I can’t find flaw in the tactics; when white supremacists are willing to murder and terrorize to get their way, and the police aren’t keeping the peace, this is precisely what you need. The interviewee also dropped an interesting citation.

00:20:05,330 –> 00:20:38,250
People should go read “This Non-Violence Stuff Will Get You Killed.” Great, amazing book about how weapons provided a militant armed self-defense backbone to the civil rights movement. It sweeps away the whitewashed narrative of Martin Luther King, and describes an entire interior world of African American and allied folks willingness – and sometimes actual use – of firearms to preserve the sanctity and lives of the people dedicated to that struggle.

I’m not that surprised to find guns mixed with social justice movements. The police and FBI have not been kind to activists, and in some cases have been infiltrated by white supremacists. Some sort of self-defense against state violence is sensible in those circumstances.

But what did surprise me was how common guns were.

Visiting Martin Luther King Jr. during the Montgomery, Alabama, bus boycott, journalist William Worthy almost sat on a loaded pistol. “Just for self-defense,” King assured him. It was not the only weapon King kept for such a purpose; one of his advisors remembered the reverend’s Montgomery, Alabama, home as “an arsenal.”

MLK Jr? Armed to the teeth? I’ve gotta pick up that book.

[HJH 2017-09-29] Speaking of which, Politically Reactive also interviewed Mark Bray about AntiFa. The more I hear about it, the more I believe that Anonymous isn’t that original.

Tracing Trolls

We hear a lot about the Kremlin’s hacking exploits (speaking of which, get acquainted with Rinat Akhmetshin), but less about their social media game. Sensitive documents cannot damage public perception if they don’t wind up getting publicity; the DNCC emails didn’t make much of a splash when they were first posted on “DCLeaks,” but made a tremendous splash when they landed on WikiLeaks.

The Kremlin has “troll factories” to do that dirty work, but how do you spot their handiwork? Salty Current points to a great Twitter thread on just that topic.

A pattern you may have noticed: many bot and troll accounts have usernames that end in 8 random digits.

I searched through two recent datasets (propagators of #FireMcMaster and #UniteTheRight hashtags) and found 824 such accounts.

Searching their followers for similarly named accounts, and subsequently their followers’ followers yielded 63099 accounts.

Here’s the follower network formed by those 63099 accounts. Larger circle = more accounts with the 8-digit numbers among its followers.

All very troll-like, but not evidence of a Kremlin op, right?

Let’s look at the largest node in the network, DavidJo52951945. This account’s been around for a while – since early 2013, 136K tweets.

Here’s an interesting observation – David is posting 8 AM – 8 PM every day, Moscow time. Almost like it’s his job or something.

What’s he tweeting about? This figure illustrates the volume of DavidJo52951945’s tweets mentioning various topics over the years.

A history of what "DavidJo52951945" has been Twetting about over the last four years. UKIP, Brexit, and migrants dominate.The messaging is very interesting. Let’s go back to 2013, check out DavidJo52951945’s tweets about Ukraine: #TrumpRussia

Several Tweets from "DavidJo52951945" about Ukraine. They have a heavy pro-Kremlin bias.So four years ago, this Twitter account tweeted on Moscow time and was heavily biased towards the Kremlin; four years later, it’s pumping out pro-Trump propaganda that also benefits the Kremlin. If this is a false-flag op, it began two years before Trump announced he was running for US president. Highly unlikely.

Also, if you think this is just a problem for US Republicans, I’ve got another Twitter thread for you.

Everyone knows about Putin’s alt-right pro-Trump trolls. He’s using left-wing anti-Trump trolls too. Exhibit A: meet @MarcusC22973194 /2

Marcus grabbed a great handle, NotMyPresident. Looking like the perfect #Resistance liberal, he’s amassed 16,500 followers in 9 months /4

Marcus is getting to be a big deal. With a Klout score of 70, he’s more influential than @SallyQYates (58) @renato_mariotti (63)… /7

Marcus is as influential as @AngrierWHStaff (70) & @DavidPriess (70) & close to & gaining on @CarlBernstein (71) & @TheRickWilson (72). /9

So how can we tell Marcus is a Russian troll? We can tell by his profile, how much he tweets, where he tweets from, when he tweets,… /12

his problems with English, how he can’t keep his cover story straight, how he plagiarizes others, the threats he makes, the bizarre… /13

things he says, how he’s too good to be true & how he pushes Kremlin propaganda. /14

The timing of Marcus’ tweets is consistent with someone working paid shifts at a Russian troll factory. archive.fo/s40DH /20

Marcus claims to be American but constantly uses British spellings revealing he didn’t learn to speak in the US writing things like… /35

behaviour colour glamour honour humour labour neighbour rumour saviour archive.is/vVk6s, and also things like… /36

Instead of calling the USSR premier “Khrushchev” like Americans would, Marcus uses “Chrustjev” like a Russian. archive.is/m3xJD /48

Russian trolls have strict quotas – they need to post 100+ times/day. This hard when you’re English not great. /69

So trolls steal other ppl’s work & pass it off as their own. Marcus does this faster than a cop hands out tickets on the 31st. /70

And so on. It’s excellent detective work, and shows the Kremlin is trying to infiltrate US left-wing politics as well. Compare and contrast this with a typical 4chan op, and you see the handiwork is quite different: the command of English is better; the Twitter handles don’t have eight numeric digits appended; the heavy use of picture memes; and of course, planning the entire thing on a public message board that many people monitor.

It isn’t that hard, once you know what to do. So why not take a boo at what the Kremlin is currently peddling, and roll up your sleeves too?

The Total Package

Mano Singham and PZ Myers aren’t that interested in eclipses. I’m sort of the opposite, as a group of us drove 13 hours to reach totality, arriving only an hour and a half before the eclipse started… and departing for the return trip fifteen minutes after totality ended. Why the heck would anyone go to such extreme lengths for a few minutes of darkness?

The Corona

The solar corona is the hottest part of the sun we can see, far hotter than the surface, and we don’t know why that is. Despite being so hot, the corona is also very diffuse and thus the cooler chromosphere blasts out far more light than it does. This means you can’t see it if any part of the sun is visible, and the physics of choronographs means they block off significantly more of the corona than the Moon does during an eclipse.

While that’s all very nice and intellectual, there’s also something satisfying about looking up in the sky and seeing what looks like Albert Einstein being consumed by a black hole.

An image of the corona from "the Camera for Photographing Eclipse Photographic Collection."Mercury

The planet Mercury is likely the last visible-eye planet discovered. Because it clings so tightly to it, you need to blot out the Sun by exploiting sunsets and sunrises, and even then you need a view close to the horizon and the planet in a certain orientation. A solar eclipse accomplishes the same, only during the middle of the day. I’m not convinced I actually saw Mercury yesterday, as it was faint and appeared in the wrong spot too close to the sun, but oh well.

Sunsets on Every Horizon

I can verify this actually happens. The physics is pretty simple: the Moon’s shadow occupies a finite area. If you’re perfectly centred under it, every horizon is in the direction of a patch of earth which has at least some sunlight falling on it. That sunlight bounces back up into or scatters through the atmosphere, producing something that looks like a sunset. It is wicked cool!

Watching The Shadow

The shadow of the Moon is quite fuzzy, so if you’re expecting to see a sharp line you’ll be disappointed. The best view is definitely from space, though an airplane will do in a pinch; on Earth, I could spot the Eastern horizon getting gradually lighter even as we were in totality.

Shadow Bands

As NASA puts it, “Shadow bands are thin wavy lines of alternating light and dark that can be seen moving and undulating in parallel on plain-coloured surfaces immediately before and after a total solar eclipse.” Scientists aren’t entirely sure what they are, but the best guess is atmospheric cells warping light in a similar way to stellar flicker. They aren’t guaranteed to show up, but I insisted on laying out a white blanket to make them more visible. We missed seeing them as totality was approaching, but as the Sun started peeking back I strongly suggested everyone stare at the blanket. And we saw them!

A Chill In The Air

The Sun radiates a lot of heat our way, which is absorbed and scattered by the ground and atmosphere. Take away that source, and you’re just left with the radiation from said ground and atmosphere as it cools down. This is at its strongest during totality, and collectively we could feel the atmosphere was notably chillier just after the eclipse than it was in the lead up. I’m estimating the difference was about 5-10C.

People Losing Their Shit

My photos of the eclipse were pretty lousy, as I didn’t have any money to invest in the proper gear. Derek Muller of Vertasium was much luckier, but his video is more notable for the audio; he, and everyone around him, were losing their minds as they reached totality. You don’t get that from a partial solar eclipse.

Don’t listen to Singham or Myers. Total solar eclipses are the coolest, and if one happens to fall near you I recommend you take full advantage.

Words Said and Unsaid

In the process of researching my last post, I stumbled on some excellent sources which didn’t fit but are worth sharing anyway. First up, a short but powerful read from Zenobia Jeffries on media whitewashing.

On Saturday, NBC said, “Charlottesville rally turned deadly.”  CNN said, “1 dead, 19 injured after crash near Unite the Right rally.” What took place was not a rally. Who wears paramilitary gear and carries automatic weapons to a rally? Who takes shields and helmets and pepper spray and bats and sticks to a rally? The car didn’t “crash”— it was driven at full speed into a crowd of counter-protesters.

What happened in Charlottesville was White nationalist extremists inciting a riot. We cannot unite, come together, overcome, Kumbaya, or whatever else, until we get some truth-telling. Media professionals need to get it right this time.

Next, Walter D. Greason set up a long Twitter thread on the history of white supremacist violence in the US, from the view of someone who taught the subject.

Fifteen years ago, I taught a course on collective racial violence in the US. It is the only course I decided to never teach again. #Thread

The students were traumatized by the weekly meetings, and I decided to break the material into multiple courses so it was easier to handle.

That said, I’ll provide some of the available materials here in hopes that we can better understand the danger #Charlottesville represents.

 Other scholars, educators, & writers: please post other resources that advance the public conversation on white supremacist violence.

And finally, the Huffington Post has a long read about the current white supremacist movement, both who’s in it and their current tactics.

Yiannopoulos had exposed a rift between the Spencer and Anglin wings of the alt-right. Both are dedicated white nationalists, but they differ on how to achieve their goals. Anglin is a purist. Spencer is willing to work with people outside the movement’s core. For instance, there is the so-called “alt-lite”—more casually bigoted mischief-makers, who might bandy about the N-word but are more likely to be upset about PC culture than, say, the Jews. A broader circle still—you could call it the “alt-white”—encompasses a large number of Trump voters. Cas Mudde, a political scientist at the University of Georgia who studies populist movements, described these people as “not necessarily racist or consciously racist. They just think they have a right to things they used to have and they don’t realize that was in a racialized and fairly racist structure.” […]

Spencer wants to meld both the alt-lite and alt-white into a viable political force. “What we should do is basically ride [Yiannopoulos’] coattails,” Spencer said. “If I wanted to create a movement that was 1488 white nationalist, I would have done that. But I didn’t because I recognized that is a total nonstarter. No one outside a hardcore coterie would identify with it. The whole point about alt-right is it’s open. Different people can identify with it. I thought that was strategically wise.”

Unfortunately, their tactics seem to be working.

The alt-right’s efforts to contaminate the zeitgeist have, by many measures, succeeded. “Everywhere now on normie sites I see our ideas and memes being pushed,” Anglin said. Since 2012, American white nationalist groups have seen their Twitter followers grow by more than 600 percent, according to a September report by the Program on Extremism at George Washington University. An ADL task force found that between August 2015 and July 2016, 2.6 million anti-Semitic tweets, many of them from Trump supporters, generated an estimated 10 billion impressions. This torrent of hate, the ADL suggested, could “contribute to reinforcing and normalizing anti-Semitic language on a massive scale.”

Attacks like those on Ioffe and Schrode have become commonplace, particularly against members of the media. According to the ADL, at least 800 journalists, most of them Jewish, were targeted by anti-Semitic attacks in the 11-month period the task force examined. Twitter, in particular, has proved ill-equipped to prevent trolls from running amok on its platform. A banned troll can set up another anonymous account within minutes and keep on trolling. And the savviest ones know exactly how far they can go: No specific threats against specific people. Nothing that can be construed as “inciting or producing imminent lawless action,” the legal standard created by the U.S. Supreme Court. The FBI initiated a threat assessment of the incidents involving Ioffe and Schrode, but did not find sufficient evidence to open an investigation, according to FBI officials in San Francisco.

That last paragraph is good evidence against civility pledges and free-speech absolutism. Bigots will happily exploit both to advance their message, and the raw numbers show it works.

Building a Science Detector

Oh, let us count the ways…

The Defense Advanced Research Projects Agency (DARPA) Defense Sciences Office (DSO) is requesting information on new ideas and approaches for creating (semi)automated capabilities to assign “Confidence Levels” to specific studies, claims, hypotheses, conclusions, models, and/or theories found in social and behavioral science research. These social and behavioral science Confidence Levels should rapidly enable a non-expert to understand and quantify the confidence they can have in a specific research result or claim’s reliability, reproducibility, and robustness.

First off, “confidence levels?” We’ve already got “confidence intervals,” and there’s been a decades-long push to use them in place of hypothesis testing.[1][2] This technique is fully compatible with frequentism (though over there it doesn’t mean what you think it does), and it even predates null-hypothesis significance testing! Alas, scientists find “we calculate a Cohen’s d of 0.3 +- 0.1” less satisfying to type than “we have refuted the null hypothesis.” The former shows a pretty weak effect, while the latter comes across as bold and confident. If those won’t do, what about meta-analyses? [3]

Second, these “confidence levels” would only apply to published research. Most research never gets published, yet those results are vital to understanding how strong any one finding is.[4] We can try to estimate the rate of unpublished works, and indeed over the decades many people have tried, but there is no current consensus on how best to compensate for the problem.[5][6][7]

Thirdly, “social and behavioral science?” The replication crisis extends much farther, into biomedicine, chemistry, and so on. Physics doesn’t get mentioned much, and there’s a reason for that (beyond their love of confidence intervals). Emphasis mine:

Even if you adjust the acceptable P value, a test of statistical significance, from 0.05 to 0.005—the lower it is, the more significant your data—that won’t deal with, let’s say, bias resulting from corporate funding. (Particle physicists demand a P value below 0.0000003! And you gotta get below 0.00000005 for a genome-wide association study.)

Just think on that. “p < 0.0000003″ means “if the null hypothesis is true, we would find a more extreme result in less than 1 in 3,333,333 trials on data like what we have observed.” If you wanted to see one of those exceptions, you’d have to do one experiment a day for 6,326 years just to have a better than 50/50 chance of spotting it. For comparison, the odds of a particular US citizen being struck by lightening over a year are 1 in 700,000; worldwide, the yearly odds of death by snake bite are about 1 in 335,000; and over the lifetime of a US citizen, the odds of them dying by dog attack are 1 in 112,400. p < 0.0000003 is a ridiculously high bar to leap, which means either a) false positives are easy to generate in physics, either via the law of large numbers or shoddy statistical techniques, or b) the field has been bitten so many times by results that can’t be replicated, even when they were real, that they’ve cranked the bar ridiculously high, or c) both.

Fourth, confidence isn’t everything. The Princeton Engineering Anomalies Research lab did studies where people tried to psychically bias random number generators. Over millions of trials, they got extremely significant results… but the odds of success were still around 50.1% vs. the expected 50%. Were they now confident that psychic abilities exist, or merely that luck and reporting bias could introduce a subtle skew into the data? Compacting those complexities into a number or label that a lay-person can understand is extremely difficult, perhaps impossible.

Basically, what DARPA is asking for has been hashed out in the literature for decades, and the best recommendations have been ignored.[8] They may have deep pockets and influence, but what DARPA wants requires a complete overhaul in how science is conducted across the globe, spanning everything from journals to how universities are organized.[9] When even quite minor tweaks to the scientific process are met with stiff opposition, pessimism seems optimistic.

[1] Gardner, Martin J., and Douglas G. Altman. “Confidence intervals rather than P values: estimation rather than hypothesis testing.” Br Med J (Clin Res Ed)292.6522 (1986): 746-750.

[2] Rozeboom, William W. “The fallacy of the null-hypothesis significance test.” Psychological bulletin 57.5 (1960): 416.

[3] Egger, Matthias, et al. “Bias in meta-analysis detected by a simple, graphical test.” Bmj 315.7109 (1997): 629-634.

[4] Rosenthal, Robert. “The file drawer problem and tolerance for null results.” Psychological bulletin 86.3 (1979): 638.

[5] Franco, Annie, Neil Malhotra, and Gabor Simonovits. “Publication bias in the social sciences: Unlocking the file drawer.” Science 345.6203 (2014): 1502-1505.

[6] Rosenberg, Michael S. “The file-drawer problem revisited: a general weighted method for calculating fail-safe numbers in meta-analysis.” Evolution 59.2 (2005): 464-468.

[7] Simonsohn, Uri, Leif D. Nelson, and Joseph P. Simmons. “P-curve: a key to the file-drawer.” Journal of Experimental Psychology: General 143.2 (2014): 534.

[8] Sedlmeier, Peter, and Gerd Gigerenzer. “Do studies of statistical power have an effect on the power of studies?.” Psychological bulletin 105.2 (1989): 309.

[9] Rawat, Seema, and Sanjay Meena. “Publish or Perish: Where Are We Heading?” Journal of Research in Medical Sciences : The Official Journal of Isfahan University of Medical Sciences 19.2 (2014): 87–89. Print.

All the President’s Bots

Trump appears cranky. It’s raining New Jersey, so he can’t golf work, which leaves him with no choice but to hate-watch CNN. Vets are angry with him, his policies are hurting his base, the polls have him at his lowest point since taking office, foreign diplomats view him as a clown, and he has nothing to show for his first six months.

He still has friends, though.

"@1lion: brilliant 3 word response to Hilary's 'I'm With You' slogan. @realDonaldTrump twitter.com/seanhannity/"Aww, at least one person likes Trump!

ilion on Twitter: STILL hasn't made a single Tweet.

… or maybe not? As an old Cracked article pointed out, Trump had a habit of quoting Tweets that didn’t exist, from people who just joined Twitter or were obvious bots. It was an easy way to make himself look more popular than he was, and stroke his ego. He put this to rest after winning the presidency, but that appears to be changing.

In a tweet on Saturday, President Donald Trump expressed thanks to Twitter user @Protrump45, an account that posted exclusively positive memes about the president. But the woman whose name was linked to the account told Heavy that her identity was stolen and that she planned to file a police report. The victim asserted that her identity was used to sell pro-Trump merchandise.

Although “Nicole Mincey” was the name displayed on the Twitter page, it was not the name used to create the account. The real name of the victim has been withheld to protect her privacy.

The @Protrump45 account also linked to the website Protrump45.com which specialized in Trump propaganda. All of the articles on the website were posted by other Twitter users, which also turned out to be fakes. Mashable noted that the accounts were suspected of being so-called “bots” used to spread propaganda about Trump. Russia has been accused of using similar tactics with bots during the 2016 campaign.

The “Nicole Mincey” scam was remarkably advanced, backed up by everything from paid articles pretending to be journalism to real-life announcers-for-hire singing her praises.

So the latest thing in the Trump resistance is bot-hunting. It’s pretty easy to do, once you’ve seen someone else do it, and the takedown procedure is also a breeze. It also silences a lot of Trump’s best friends.

If only we could do the same to Trump.