Category: Uncategorised

Enlarge / This morning, Trending promoted this as its top story related to the trending topic "Megyn Kelly." The story was up for several hours, and is completely false. (credit: Washington Post)

Earlier this year, Facebook denied criticisms that its Trending feature was surfacing news stories that were biased against conservatives. But in an abrupt reversal, the company fired all the human editors for Trending on Friday afternoon, replacing them with an algorithm that promotes stories based entirely on what Facebook users are talking about. Within 72 hours, according to the Washington Post, the top story on Trending was about how Fox News icon Megyn Kelly was a pro-Clinton "traitor" who had been fired (she wasn't).

The original accusations of bias came from a disgruntled ex-editor at Facebook, who leaked internal Trending training materials to Gizmodo. The training package offered tips on, among other things, how to curate news from an RSS feed of reputable sources when the stories provided by Facebook users were false or repetitive. Though the human editors were always expendable—they were mostly there to train the Trending algorithm—they were still engaging in quality control to weed out blatant falsehoods and non-news like #lunch. And after Trending latched on to the fake Kelly scoop, it appears that human intervention might still be required to make Facebook's algorithms a legitimate source of news after all.

In a post about the changes, Facebook said the early move to eliminate human editors was a direct response to "the feedback we got from the Facebook community earlier this year," an oblique reference to the raging controversy unleashed by the Gizmodo revelations. Facebook explained that the new, non-human Trending module is personalized "based on a number of factors, including Pages you’ve liked, your location (e.g., home state sports news), the previous trending topics with which you’ve interacted, and what is trending across Facebook overall." Instead of paying humans to "write topic descriptions and short story summaries," the company said "we’re relying on an algorithm to pull excerpts directly from news stories." Which is why millions of Facebook readers this morning saw the "news" that Megyn Kelly is a traitor who has been fired.

Enlarge / Jennifer Youngman, 65, used a .410 gauge shotgun like this to take out a drone. (credit: Big Swede Guy)

With a single shotgun blast, a 65-year-old woman in rural northern Virginia recently shot down a drone flying over her property.

The woman, Jennifer Youngman, has lived in The Plains, Virginia, since 1990. The Fauquier Times first reported the June 2016 incident late last week. It marks the third such shooting that Ars has reported on in the last 15 months—last year, similar drone shootings took place in Kentucky and California.

Youngman told Ars that she had just returned from church one Sunday morning and was cleaning her two shotguns—a .410 and a .20 gauge—on her porch. She had a clear view of the Blue Ridge Mountains and neighbor Robert Duvall’s property (yes, the same Robert Duvall from The Godfather). Youngman had seen two men set up a card table on what she described as a “turnaround place” on a country road adjacent to her house.

(credit: Getty Images | Yuri_Arcurs)

The Federal Communications Commission has decided not to appeal a court decision that allows states to impose laws restricting the growth of municipal broadband.

The FCC in February 2015 voted to block laws in North Carolina and Tennessee that prevent municipal broadband providers from expanding outside their territories, but the states convinced a federal appeals court to keep the laws in place. The FCC could have asked for another appeals court review or gone to the Supreme Court but will instead let the matter drop.

"The FCC will not seek further review of the [US Court of Appeals for the] Sixth Circuit's decision on municipal broadband after determining that doing so would not be the best use of Commission resources," an FCC spokesperson told Ars today. The decision was also reported yesterday in The New York Times.

The classic model of scientific progress is that the field advances when new findings contradict or supersede old ones. But a new study reveals that this process isn't working today—at least, not in scientific journals, where most data is shared with colleagues. Indeed, the researchers found that "rebuttals scarcely alter scientific perceptions about the original papers."

For the study, a group of researchers looked at the citation rates on seven marine biology papers about fisheries. Citation rates are often used as a proxy for the "importance" of a scientific paper, with the notion that the more a paper is cited, the more influential it is. Each paper had been the subject of a rebuttal, also published in a scientific journal. The researchers wanted to know whether these rebuttals affected citation levels on the original papers—and, perhaps more importantly, whether they convinced people to question the interpretation of data in the original papers.

It turns out that rebuttals don't seem to affect the scientific community's understanding of the original papers in any way. "The original articles were cited 17 times more frequently than the rebuttals, an order of magnitude difference that overwhelms other factors," write the study authors in Ecosphere. "Our test score results emphasize that rebuttals have little influence: even the rare few authors who happened upon the rebuttals were influenced only enough to move from whole-hearted support of the original article (a score of five) to neutrality (a score of three), despite the fact that all of the rebuttals argue that the interpretations of data in the originals were incorrect. Astonishingly, 8 percent of the papers that cited a rebuttal actually suggested that the rebuttal supported the claims of the original article, an observation which may give pause to those contemplating writing a rebuttal in the future."

Enlarge (credit: Apple)

Apple has just send out press invitations for its next product event, which is happening at 10am Pacific on September 7 in the Bill Graham Civic Auditorium in San Francisco. This has become Apple's go-to event space in recent years, replacing smaller venues like the Moscone Center, the Yerba Buena Center for the Arts, and Apple's own town hall event space (the latter of which was officially retired earlier this year when the iPhone SE was announced).

New iPhones are a sure bet for the event—this year's models aren't expected to deviate too far from the existing iPhone 6S and 6S Plus design, but they're said to include better cameras, faster chips, and no headphone jack. Along with that new hardware, we can also expect release date announcements for macOS Sierra, iOS 10, watchOS 3, and tvOS 10.

Other hardware announcements are definitely possible—practically all of Apple's products are a year or more old at this point—but rumors have been less-than-consistent. A new Apple Watch model and new MacBook Pros are on the more likely end of the spectrum, but the larger iPad Pro and most of the Mac lineup are at least a year old, and a few of the Macs are even older. The fourth-generation Apple TV box is around a year old too, but Apple has never updated its set-top box on a regular yearly cadence the way it has with many of its other products.

Enlarge / Lakes formed by melting permafrost, on peatland. In Hudson Bay, Canada. (credit: Steve Jurvetson)

During the last deglaciation, between roughly 21,000 and 10,000 years ago, there was a rise in atmospheric carbon. This surge brought CO₂ levels up to where they were in preindustrial times and contributed to the warming that ended the glacial period. But there's a significant item missing from this picture: we don't know where the carbon came from.

Researchers had suggested that changes in the distribution of ice, driven by alterations in Earth's orbit and tilt, altered the ocean’s capacity to absorb CO₂. But a new paper performed a model-driven analysis of past changes in carbon levels and come up with a somewhat different answer. The authors' simulations showed that, when a permafrost carbon component was included, it was possible to reproduce the atmospheric CO₂ levels seen in ice core measurements—suggesting that carbon released by melting permafrost contributed to the rise of CO₂.

Carbon accounting

Data from the ice cores can help narrow down the possibilities, because it records something called δ¹³C (delta-thirteen-C), which is essentially a measure of the ratio of carbon-13 to carbon-12 in the atmosphere. (It’s mathematically a bit more complicated, but that’s the basic idea). As this ratio is influenced by biological activity, it can give some clues about the carbon's source. Even with these clues, however, previous simulations have failed to narrow down the possibilities. The researchers suspected that was because these weren't taking into account an important mechanism: change in permafrost.