Rocking the Boat

YouTube needs to be tied; algorithmic bias alert; and much more.

  • This is civic tech: French PhD candidate Julien Carbonnell has pulled together a dataset of 500 civic tech professionals (with a strong tile toward French and American practitioners) and reports the following findings: Their median age is 32, half of them created their projects in the last three years (while only 25% have more than five years under their belt), nearly half are full-time employees of their organizations, 70% identify closely with the term “civic tech,” and the two main ways they engage with the marketplace is either by providing services to governments, candidates and community organizations, or by building tools for autonomous citizen empowerment.

  • Videos from many of the main talks and sessions from last month’s TICTeC (The Impacts of Civic Tech Conference) are now up on mySociety’s YouTube channel.

  • Speaking of civic tech conferences, Personal Democracy Forum Central-Eastern Europe is now gathering in Gdansk and you can follow along live here or via the #PDFCEE19 hashtag.

  • New Yorker Jeff Novich called 311 to report a honking taxi outside his apartment back in 2014, and after he learned that 311 didn’t have a way of tracking taxi-related complaints, he built his own solution: a mobile app called Reported. As Lisette Voytko reports for Gotham Gazette, a year and a half later 311 embraced his app, and since taxi reports now flow directly to the Taxi and Limousine Commission, users get results. In 2018, more than 4,500 complaints about taxis generated $380,000 in revenue back to the TLC as a result of fines imposed. AS Voytko writes, “The Reported app is an example of New Yorkers taking matters – including the city’s open data – into their own hands, creating ‘civic tech’ to serve their needs and better understand their surroundings.” But, she notes, while NYC has a strong open data law, improvements in collecting citizen feedback through services like 311 don’t always lead to improved enforcement.

  • Media matters: Older Americans are more civically active than their younger peers, and also less digitally literate, and as Craig Silverman reports for BuzzFeed News, that makes them more likely to be targeted by misinformation and hyper-partisan messages online, and more likely to share it.

  • Here’s a fascinating and well done report on how the town of Baraboo, Wisconsin, dealt with the photo of a bunch of high school boys making the Seig Heil salute, written by Joseph Bernstein for BuzzFeed News. It raises all kinds of questions about who town civic leaders decided needed protection after the picture blew up online (the boys far more than the town’s Jews, sadly), how social media can rip things from context (it seems clear that the salutes were at worst a stupid accident, not something anyone planned), and the tenuous civic fabric that barely manages to address the underlying issues at work (none of the town’s titular leaders stepped up much and most of the heavy-lifting aimed at healing the resulting situation was done by one high school boy and two older Jewish women who are town residents). One lesson of the story: we can’t wait til a crisis hits to work on strengthening the social fabric across identity lines.

  • Life in Googlestan: YouTube is one of seven Google-owned platforms with more than one billion users (the others are Gmail, Android, Chrome, Maps, Search and the Play Store), and like all the other Big Tech platforms that have prioritized growth over every other value, it is a leading vector of rising polarization, extremism and disinformation. Outside critics have been making this point for a while, but as Mark Bergen reports in a major investigative story for Bloomberg, “scores of employees inside YouTube and Google” have been raising concerns in recent years about “the mass of false, incendiary and toxic content that the world’s largest video site surfaced and spread.” He adds, “each time they got the same basic response: Don’t rock the boat.”

  • Bergen points out that the problems with YouTube accelerated around 2009, as Google took tighter control of the platform it had bought and set an internal goal of reaching one billion hours of viewing a day. So, just for those of you keeping track, at the same time the Obama administration welcomed a wave of Googlers in for a bout of government service (cued by the buddy-buddy relationship of Google chairman Eric Schmidt with President Obama), YouTube was shifting to the “engagement-uber-alles” mode that has helped power the populist wave now washing away the achievements of the Obama years.

  • Katherine Maher, the ED of the Wikimedia Foundation, comments, “I cannot believe this article ends and begins with pointing to @Wikipedia as *still* the best intervention @YouTube has come up with. Speechless at leadership paralysis on public responsibility.” YouTube points viewers of tendentious videos to Wikipedia pages as a supposed corrective. (Bonus comment from Anil Dash.)

  • Reacting to the Bloomberg YouTube story, New York Times opinion writer Charlie Warzel connects the dots to Facebook and Twitter, writing “Reports from inside YouTube echo insider tales from other large tech platforms — a familiar pattern of ruthless optimization for users and of attention gone wrong.”

  • Life in Facebookistan: A new academic paper that studied Facebook’s advertising algorithms has found that because they have been trained on historically biased data, they consistently deliver results that are biased by race and gender, Hal Hodson reports for The Economist. Otherwise identical ads with different pictures of black and white families were routed to different groups of people, researchers at Northeastern University, USC and Upturn (a DC advocacy group) found. For example, “An ad for cheap houses for sale, which depicted white families, was delivered to an audience that was 85% white. An identical ad that contained pictures of black families was served to an audience comprising around 73% white users.” As Hodson writes, “This suggests that fewer black people saw ads for cheap or affordable housing when those ads used pictures of white people.”

  • As Hodson notes, the study’s authors bring important new evidence to bear on whether Facebook should be held liable for the discrimination its own systems appear to be fostering, or whether Section 230 protections for platforms that act as neutral hosts for content still apply.

  • A new tip line launched in India by WhatsApp to respond to reports of misinformation as India’s general election begins is actually “not a helpline” and isn’t primarily designed to provide feedback, Ryan Mac and Pranav Dixit report for BuzzFeed News. It’s mainly a research tool, according to Proto, the company running the project.

  • Two troves of Facebook user data belonging to hundreds of millions of users have been sitting exposed on Amazon’s cloud servers, Issie Lapowsky reports for Wired, further proof of how badly the company stewarded user data that it allow third-party app developers to obtain.

  • Facebook is paying the Daily Telegraph to run a series of positive features about the company, Rob Price reports for Business Insider. Facebook spokesperson Vicky Gomes said that “this is a part of our larger marketing efforts in the UK with the goal of educating and driving awareness of our local investments, initiatives, and partnerships here in the UK that have a positive impact on people’s lives.” This is really great, because it kills at least two birds with one stone: Facebook says it wants to help prop up journalism, and it needs more positive coverage. 26 articles have run over the last month.

  • A group of top AI researchers are calling on Amazon to stop selling its Recognition AI facial recognition service to police departments. The tool, they say, has much higher error rates classifying the gender of darker skinned women.

  • Summing things up: Big Tech platforms, fueled by the culturally sanctioned greed and ambitions of their makers, are dramatically altering the larger cultural environment, rewarding emotionally engaging and polarizing content over facts, polluting and warping the public sphere both with their own marketing noise as well as in hidden ways with biased algorithms, and mostly sticking bandaids on the problems they are accelerating. Meanwhile, civic actors like Wikipedia and civic hackers are trying to foster a more authentic and responsible civic culture, but we’re swimming against a current that has yet to be slowed, not to mention reversed. Tech needs to do more to address the public/civic health crises it is fueling.

You are reading First Post, a twice-a-week digest of news and analysis of the world of civic tech, brought to you by Civic Hall, NYC’s community center for civic tech. If you are reading this because someone forwarded it to you, please consider joining Civic Hall as a network member and supporting our work or sign up to our newsletter to stay connected with the #CivicTech community.