Day: 19 June 2019

Last year, I wrote a couple of articles for the Sweet Setup with my recommendations for great apps to shoot and edit photos on your iPhone. I wanted to be thorough, so I downloaded a couple of dozen apps to choose from for each review. And, just about every time, I was asked to confirm that it was okay for that app to use the camera, then to confirm that it was okay to use my location to tag the photos, and then to confirm that it was okay to access the photo library. If the app shot video, I’d maybe also be asked to grant permission to use the microphone.

In a vacuum, these prompts for permission make sense, and it’s not like they haven’t been borne of purpose. The reason you have to confirm the use of your contacts is directly because of the Path app. If we didn’t have to confirm the use of location data, you can bet every ad-supported app would exploit it. But, since we have to permit the use of our location, surveillance-friendly developers have accepted that they should respect a user’s decision not to opt in. Right?

Of course not. The economy of surveillance capitalism has an insatiable appetite for data and, if they can’t get it legitimately, mischievous developers will figure out another way.

It’s possible to use Bluetooth beacons to approximate the location of a device. So, in iOS 13, apps are now required to ask permission to use Bluetooth, and those of us using the beta seeds so far have been pretty surprised at the number and variety of apps that have been requesting Bluetooth permissions.

Marco Arment retweeted a screenshot from Benjamin Herrin today that shows nearly fifty apps on his device that have so far thrown a Bluetooth prompt, including Lyft, IKEA, and Smoothie King. But a majority of the apps in that list are largely for media and, likely, implement Google’s Chromecast “sender” SDK, which lists the CoreBluetooth framework as a dependency. As a result, if you want to output video from your phone to your Chromecast device you, inherently, give permission for the app to estimate your location using Bluetooth. If you explicitly give an app permission to use your location, that app might hand it over to a third-party marketing and analytics firm for their own use. That’s exactly what happened to users of Accuweather’s app.

Apple does its part in trying to combat these insidious privacy violations. They have rules that disallow tracking users without their knowledge; they display permissions dialogs. But nobody’s fooling themselves: bad developers will always find a way to excavate user data. Just look at the myriad shitty uses of JavaScript on the web. Besides, asking Apple to somehow police the shuffling of user data from one service to another is a ridiculous proposition. I’m not feeling sorry for big tech companies; I’m being realistic.

This is why it’s so critical for the privacy of user data to be enshrined in law. We are being universally surveilled and there’s little we can do to change that. Permissions dialogs are far more than decoration, and it is useful to inform users about what technologies the app wishes to use, but they are no longer capable of bearing the burden hoisted upon them. Users — we — need enforceable and realistic legal protections.

Casey Newton, the Verge:

In February, I wrote about the secret lives of Facebook contractors in America. Since 2016, when the company came under heavy criticism for failing to prevent various abuses of its platform, Facebook has expanded its workforce of people working on safety and security around the world to 30,000. About half of those are content moderators, and the vast majority are contractors hired through a handful of large professional services firms. In 2017, Facebook began opening content moderation sites in American cities including Phoenix, Austin, and Tampa. The goal was to improve the accuracy of moderation decisions by entrusting them to people more familiar with American culture and slang.

Cognizant received a two-year, $200 million contract from Facebook to do the work, according to a former employee familiar with the matter. But in return for policing the boundaries of free expression on one of the internet’s largest platforms, individual contractors in North America make as little as $28,800 a year. They receive two 15-minute breaks and a 30-minute lunch each day, along with nine minutes per day of “wellness” time that they can use when they feel overwhelmed by the emotional toll of the job. After regular exposure to graphic violence and child exploitation, many workers are subsequently diagnosed with post-traumatic stress disorder and related conditions.

Newton interviewed twelve Facebook moderators based at Cognizant’s Tampa office for this story, current and former, and even reading the descriptions of what they witnessed were enough to make me reel. I can’t imagine being required to sit through the minimum of fifteen seconds per video that they must view.

When I linked to Newton’s last piece about Cognizant, I wrote that this is a fundamental issue with the design of social networks. Automatic filtering — much of which, by the way, is an illusion — is inaccurate and will likely always be so, but subjecting low-wage contractors to a firehose of the worst of humanity is unconscionable. I’m not sure paying them more would help with the psychological toll, but I certainly wouldn’t argue against a pay bump. Cognizant or Facebook should be providing them with therapy, free of charge and without stigma. But something must change, too, with the very design of platforms like Facebook to make it harder for people to share and glorify disturbing, discriminatory, and abhorrent material.

One more thing from Newton’s article that requires a highlight:

Florida law does not require employers to offer sick leave, and so Cognizant workers who feel ill must instead use personal leave time.

It is a failure of ethics for executives and lawmakers if employees cannot expect time to recover from illness.

Update: Manton Reece:

Social networks like Facebook (and Twitter) are designed to reward the sensational video. The timeline algorithm, “like” counts, and quick re-sharing all contribute to surfacing both the best and worst content. Whatever drives engagement.

Until the tie is cut between engagement and revenue, I don’t think this will be solved. By that, I mean that I don’t think this will be solved — at least not by these incumbents.

See Also: Page 35 of Sarah Jeong’s “The Internet of Garbage” (PDF).

Keach Hagey, Rob Copeland, and Sam Schechner, Wall Street Journal:

Google’s competitors are pressing antitrust enforcers to look far and wide at the company’s practices. Perhaps the most common complaint against Google around the world in recent years is that it uses its search engine to privilege its own content at the expense of its competitors’.

For example, it created new design features like the “knowledge graph,” which populates the boxes that appear at the top of search, often answering a query without requiring the user to click through to another website. In March, 62% of Google searches on mobile were “no-click” searches, according to research firm Jumpshot. Google has argued that if consumers don’t find the rearranged content useful, they won’t click on it.

Via Michael Tsai, who also links to a tweet from Kyle Howells showing a Google search on his iPad that displayed only an information box, with an additional tap required to see further search results.

The proportion of people who, Jumpshot claims, do not click or tap on a search result on mobile is unsurprising to me. Google’s mobile search results page often displays a knowledge panel that occupies the entire height of the viewport. This may be enough to answer the query; however, it’s frequently inaccurate, and there’s little indication that different results are available which may better answer the searcher’s question.

See Also: Bad products selected by Amazon showing the inherent power of recommendations; inaccurate Featured Snippets; how Featured Snippets killed Celebrity Net Worth; and, recently, the wholesale copying of lyrics from Genius.