Instagram Has a Massive Harassment Problem

Via Charles Arthur, The Atlantic on Instagram (owned by Facebook so don’t be surprised if they share the same values):

When Instagram introduces new features, the moderation-team members receive no warning, Andy [who works as a moderator; that’s not his real name] said. Consequently, they are left scrambling to understand how they work and what constitutes harassment on each format. “When the Questions feature rolled out, same way as every other new feature, we had no idea,” he said. “We didn’t know which part is the question, which is the answer, who says what? That makes such a big difference on whether you’re going to delete or ignore the post. The mods are just totally not kept up to date on how people use features.”

Alex, the current Instagram employee who asked to be referred to by a pseudonym, said the company prioritizes growth above all else, often at costs to user experience. “The focus is still on getting people to spend more time, getting more users, getting more revenue. That doesn’t change much internally,” Alex said. “There’s been a lot of effort to shape the narrative, but the reality is that it doesn’t drive business impact.”

At Instagram and Facebook, Alex said, “features can make whatever progress … but can’t hurt the other metrics. A feature might decrease harassment 10 percent, but if it decreases users by 1 percent, that’s not a trade-off that will fly. Internally right now, no one is willing to make that trade-off.”

Allie, a former employee at Instagram, agreed. “Instagram has terrible tools. I think people haven’t really focused on it much because so many harassment campaigns are just more visible on other platforms,” she said. Throughout her time there, she said, “many of the efforts to reduce harassment were oriented toward PR, but very few engineering and community resources were put toward actually decreasing harassment.”

Sometimes, it’s the data you’re missing that’s the key to understanding something

Via John Naughton, a salutary tale for data fiends:
How Not to Be Wrong opens with an extremely interesting tale from World War II. As air warfare gained prominence, the challenge for the military was figuring out where and in what amount to apply protective armor to fighter planes and bombers. Apply too much armor and the planes become slower, less maneuverable and use more fuel. Too little armor, or if it’s in the “wrong” places, and the planes run a higher risk of being brought down by enemy fire. To make these determinations, military leaders examined the amount and placement of bullet holes on damaged planes that returned to base following their missions. The data showed almost twice as much damage to the fuselage of the planes compared to other areas, most specifically the engine compartments, which generally had little damage. This data led the military leaders to conclude that more armor needed to be placed on the fuselage. But mathematician Abraham Wald examined the data and came to the opposite conclusion. The armor, Wald said, doesn’t go where the bullet holes are; instead, it should go where the bullet holes aren’t, specifically, on the engines. The key insight came when Wald looked at the damaged planes that returned to the base and asked where all the “missing” bullet holes to the engines were. The answer was the “missing” bullet holes were on the missing planes, i.e. the ones that didn’t make it back safely to base. Planes that got hit in the engines didn’t come back, but those that sustained damage to the fuselage generally could make it safely back. The military then put Wald’s recommendations into effect and they stayed in place for decades.

Eye Movements During Everyday Behavior Predict Personality Traits

Frontiers in Human Neuroscience:

Besides allowing us to perceive our surroundings, eye movements are also a window into our mind and a rich source of information on who we are, how we feel, and what we do. Here we show that eye movements during an everyday task predict aspects of our personality. We tracked eye movements of 42 participants while they ran an errand on a university campus and subsequently assessed their personality traits using well-established questionnaires. Using a state-of-the-art machine learning method and a rich set of features encoding different eye movement characteristics, we were able to reliably predict four of the Big Five personality traits (neuroticism, extraversion, agreeableness, conscientiousness) as well as perceptual curiosity only from eye movements. Further analysis revealed new relations between previously neglected eye movement characteristics and personality. Our findings demonstrate a considerable influence of personality on everyday eye movement control, thereby complementing earlier studies in laboratory settings. Improving automatic recognition and interpretation of human social signals is an important endeavor, enabling innovative design of human–computer systems capable of sensing spontaneous natural user behavior to facilitate efficient interaction and personalization.


Google says the words “African-American composers” and “African-American music” are “dangerous or derogatory.”

But the BIGGER lesson, as has been stated by so many people far smarter than I am, is that explicit prejudice isn’t necessary to create a discriminatory system. Just four steps:

1) Algorithm is created to flag offensive terms.

2) Algorithm is fed with data from survey respondents who may not have been selected in enough quantity and demographic diversity. Subconscious (or worse) prejudice in a segment of respondents is amplified dramatically; bias against a term leads to a full ban of that term.

3) On a separate team, support agents attempt to resolve problems one by one. Their goal? Click “resolved”, get a gold star, move on to the next case. There’s likely little incentive to escalate cases potentially symptomatic of larger and more malignant problems in the system.

4) Engineers remain blissfully unaware of the problem as it persists and even intensifies. After all, no one has reported any “offensive” ads, so it must be working. And a group’s voice has now been muted without requiring the slightest malicious intent from anyone involved.


Facebook Is Reviewing its Policy on White Nationalism After Motherboard Investigation, Civil Rights Backlash


Facebook told Motherboard it’s currently reviewing its policies on white supremacy, white nationalism, and white separatism after a series of meetings with civil rights leaders, reporting by Motherboard on these policies, and a forceful letter from a civil rights group formed under the direction of President John F. Kennedy.

Leaked internal documents show that Facebook’s content moderators are explicitly instructed to allow “white separatism” and “white nationalism” on the platform, but note that “white supremacy” is banned. Facebook makes this distinction because it argues in those documents that white nationalism “doesn’t seem to be always associated with racism (at least not explicitly.)”

Now, following that reporting, multiple leading civil rights groups and Black history scholars are calling for Facebook to change its stance, saying that separatism and nationalism are a thinly-veiled mask for white supremacy.

“The idea that they are making a distinction that is basically buying into what the white nationalists are trying to sell is deeply troubling,” Becky Monroe, the director of the Stop Hate Project at the Lawyers’ Committee for Civil Rights Under Law, told Motherboard in a phone call.

The organization was formed in 1963 at the request of John F. Kennedy at the height of the civil rights movement. Monroe told Motherboard the committee met with Facebook over the summer to discuss the issue, and, in a letter the committee wrote to the company earlier this month obtained by Motherboard, it says Facebook’s stance is at odds with the central tenet of Brown v. Board of Education, the foundational Supreme Court ruling which found the doctrine of racial segregation is inherently unequal.

“By attempting to distinguish white supremacy from white nationalism and white separatism, Facebook ignores centuries of history, legal precedent, and expert scholarship that all establish that white nationalism and white separatism are white supremacy,” the letter states (emphasis theirs.) “Indeed, when we met with your company this summer, both our staff as well as the staff at Facebook, were unable to identify an example of white nationalism or white separatism that was not white supremacist.


Alternative Influence: Broadcasting the Reactionary Right on YouTube

Rebecca Lewis’s new report for Data & Society:

presents data from approximately 65 political influencers across 81 channels to identify the “Alternative Influence Network (AIN)”; an alternative media system that adopts the techniques of brand influencers to build audiences and “sell” them political ideology.

Alternative Influence offers insights into the connection between influence, amplification, monetization, and radicalization at a time when platform companies struggle to handle policies and standards for extremist influencers. The network of scholars, media pundits, and internet celebrities that Lewis identifies leverages YouTube to promote a range of political positions, from mainstream versions of libertarianism and conservatism, all the way to overt white nationalism.

Notably, YouTube is a principal online news source for young people. Which is why it is concerning that YouTube, a subsidiary of Google, has become the single most important hub by which an extensive network of far-right influencers profit from broadcasting propaganda to young viewers.

“Social networking between influencers makes it easy for audience members to be incrementally exposed to, and come to trust, ever more extremist political positions,” writes Lewis, who outlines how YouTube incentivizes their behavior. Lewis illustrates common techniques that these far-right influencers use to make money as they cultivate alternative social identities and use production value to increase their appeal as countercultural social underdogs. The report offers a data visualization of this network to show how connected influencers act as a conduit for viewership.

How the Waffle House does it

Financial Times:

This morning we got a chance to talk briefly with Walt Ehmer, CEO of Waffle House. For our global readers unfamiliar with the American South, the Waffle House is a diner chain that functions like a medieval tavern: it sits at every crossroads and it is always open. The Waffle House is where travelers meet each other. It provides refuge for the tipsy, the joyful, the insomniacs. If you grew up below the Mason Dixon line, something happened to you once at the Waffle House.

The definition of a Waffle House is this: when you need one, it’s nearby and serving. Which is why the chain has to be so good at staying open after a hurricane. Here’s how Ehmer puts it:

I tell people all the time, I said, we’re really not that smart, we’re not that complicated. We just have a lot of want to. We want to be there for the community. We want to be there for our people. We want to be there for the first responders. 

The way the Waffle House stays open is, in fact, really smart and really complicated.  A 2010 case study for the International Journal of Production Economics walks through what “a lot of want to” looks like.

The Hidden Link Between Farm Antibiotics and Human Illness


For almost seven decades, we’ve routinely fed antibiotics to the animals we eat. That’s just a few years less than we’ve taken antibiotics ourselves. And for just about as long, it’s been clear that those antibiotic doses have been creating drug-resistant bacteria that pass from meat animals to make humans sick.

The first outbreaks of drug-resistant foodborne illness were spotted as early as the mid 1950s, when an epidemic of resistant salmonella swept through southeastern England. That was the first of waves of outbreaks that occurred over decades, some small and some very large and widespread. One of the largest foodborne outbreaks in US history, which made 634 people in 29 states and Puerto Rico sick in 2013-14, was tracked back to chickens that had been given antibiotics in their feed.

The connection isn’t universally accepted, of course. Most of the studies linking farm antibiotic use and human illness have been observational, not experimental — and that’s given ag and pharma room to insist that the case against farm antibiotic overuse isn’t solid. The argument has been that the bacterial traffic from animals to meat to humans isn’t proven — and until it can be established with 100 percent certainty, the practice of giving livestock preventative antibiotics should continue.

Now a new study, years in the making, goes further than any other to demonstrate that resistant bacteria can move from animals to humans via the meat they become. It also provides a model of how new surveillance systems might reduce that bacterial flow at its source on farms.

It’s just one study, but it possesses outsize significance, because it eliminates the uncertainty at the center of that bacterial flow. Outside of experimental conditions, it’s never been possible to prove that this antibiotic given to that animal gave rise to this bacterium that ended up in thathuman. But this new work dives so deeply into the genomics of bacterial adaptation in food animals and humans, it proves the link that ag would rather deny.

Good animal husbandry requires some use of antibiotics.  Factory farming relies on antibiotics and that causes widespread resistance to antibiotics in humans.  That’s a significant external cost that factory farmers should be paying for.  Antibiotic use should be taxed.