ArduinoMania

Instagram is, once again, trying to explain how its recommendations work in an attempt to dispel “misconceptions” about how the app’s algorithm works and whether or not the company engages in “shadowbanning” of certain creators. In a new blog post from Instagram’s top exec Adam Mosseri, he offers one of the most detailed explanations to date on how the app ranks content in various parts of the app.

“Instagram doesn’t have a singular algorithm that oversees what people do and don’t see on the app,” Mosseri explains. Instead, he says, there are multiple algorithms and ranking systems underpinning different aspects of the app, like Explore, Reels, Stories and search. Each of these uses a variety of signals to determine how content is ranked for each user.

For example, the order of posts in your main feed is determined by your past activity, as well as previous interactions with the person who made each post. Likewise, Stories posts take into account viewing history as well as “closeness,” or “how likely you are to be connected as friends or family.” On the other hand, recommendations in Explore are largely based “posts you’ve liked, saved, shared and commented on in the past,” but are more likely to come from accounts you’ve never interacted with.

One of the more interesting sections of Mosseri's blog post is titled “addressing shadowbanning.” Mosseri notes that there isn’t a universal definition for the word, but acknowledges that many creators “use the term to imply that a user’s account or content is limited or hidden without a clear explanation or justification.” And he says that the company is working to increase transparency around when creators’ content or accounts are blocked from the app’s recommendations.

Specifically, he calls out the app’s “account status” feature, which can alert users if one of their posts or their account is considered “ineligible” for recommendations. The feature also offers an appeals process. While it’s not the first time Instagram has addressed the issue, which has been the subject of much speculation and conspiracy theories over the years, there has been a notable shift in the way the company is talking about “shadowbanning.”

In a similar post from two years ago, Mosseri wrote that “the truth is most of your followers won’t see what you share, because most look at less than half of their Feed.” Now, he says Instagram is working on increasing transparency in cases when a creator’s content isn’t widely distributed due to a policy violation. “If anything makes your content less visible, you should know about it and be able to appeal,” he wrote.

He added that Instagram is testing “new notifications to help creators understand when the reach of their reel may be limited due to a watermark” (the company has tried to discourage users from posting recycled TikToks to Reels for years).

While some creators may still find these explanations unsatisfying — and there are more than a few who fall into that camp, judging by the comments on Mosseri’s own Instagram post — the new details underscore just how central algorithmic recommendations are becoming to Instagram. While the app re-introduced an optional chronological feed, Mark Zuckerberg has said his goal is to transform Instagram and Facebook into a “discovery engine” more focused on recommendations than posts from friends.

Reddit’s recently-announced plan to charge for API access could price out the developer of one of the most popular third-party Reddit apps. The developer of Reddit client Apollo is raising the alarm on the new API pricing, saying the changes would require him to spend millions of dollars to keep his app going in its current form.

Reddit announced sweeping changes to its API rules last month, citing the rise of AI companies using their platform to train large language models. “The Reddit corpus of data is really valuable,” Reddit CEO Steve Huffman told The New York Times. “But we don’t need to give all of that value to some of the largest companies in the world for free.”

But it now seems that independent app makers will also be subject to the pricier new plans, which are set to take effect June 19th. While Reddit hasn’t officially disclosed its API pricing, Christian Selig, Apollo’s sole developer, says he would have to pay $20 million to keep his app going “as-is” under the new policies.

“50 million requests costs $12,000, a figure far more than I ever could have imagined,” he wrote in a post on Reddit, citing multiple conversations he’s had with Reddit representatives about the upcoming API changes. “Apollo made 7 billion requests last month, which would put it at about 1.7 million dollars per month, or 20 million US dollars per year.”

That pricing leaves Selig and Apollo, which has been widely praised for its design details and for providing functionality beyond Reddit’s native app, in a tough position. While the app does offer subscriptions, its current revenue isn’t enough to cover the steep API cost. He says that the average user makes about 344 API calls a day, which would require him to raise subscription prices to at least $2.50 a month (currently, he says, most subscribers pay $0.99 a month). Furthermore, that wouldn’t account for Apollo’s power users, who use the app at much higher rates, or the app’s free users. “Even keeping the existing, subscription only users I would be SUBSTANTIALLY in the red each month,” Selig tells Engadget.

In a statement, a Reddit spokesperson said that Selig was provided “pricing per 1,000 API calls, not a monthly bill,” but declined to share details. “Our pricing is based on usage levels that we measure to be as equitable as possible,” the spokesperson said. “We’ve been, and will continue, to work with third-party apps to help them improve efficiency, which can significantly impact overall cost.”

If all of this sounds oddly familiar, there are striking similarities between Reddit’s new developer rules and the drastic changes Twitter has made to its API policies under Elon Musk. In Twitter’s case, the company decided to ban third-party client apps while simultaneously making its API extraordinarily expensive for the researchers and businesses that previously depended on higher levels of access to Twitter data.

Of note, Reddit hasn’t been as outwardly hostile to developers. Selig notes that he’s had multiple calls with Reddit and that reps he’s spoken to have been “communicative and civil” about the changes. And a Reddit spokesperson suggested the company wants to keep third-party apps around.

“We’re committed to fostering a developer ecosystem around Reddit – developers and third-party apps can make Reddit better,” the spokesperson said. “Our Data API has powered thousands of applications, such as tools to make moderation easier, and utilities that help users stay up to date on their favorite topics, and games. Developers are incredibly valuable to the Reddit ecosystem, so much so that we recently updated our Developer Platform.”

Still, Selig said he’s uncertain about how he will handle the changes. “I hope it goes without saying that I don't have that kind of money,” he shared on Reddit. “This is going to require some thinking.”

Two employee groups at Amazon have joined together to stage a corporate walk out today, uniting to protest the company's return-to-office policy and to raise concerns about Amazon's climate impact.

Standing in front of Amazon's Seattle Headquarters, the group streamed the event live on Twitter — featuring speakers for both groups advocating for their united cause. Some speakers vented their frustrations with the company's policy to have workers return to the office for at least three days a week, telling stories about how the remote work kicked off by the COVID pandemic bought them precious hours at home with their family and saved them from hours of daily commute time. Another speaker married this idea to the company's climate goals, highlighting how remote work allowed more families to become one-car households. This dovetails into some of the groups' complaints that Amazon is failing to meet its own goals in its climate pledge of reaching zero emissions by 2040.

According to the Amazon Employees for Climate Justice Twitter page, more than 1900 Amazon employees pledged to participate in the walk out. Amazon did not immediately respond to a request for comment. 

Amazon isn't just settling over Ring doorbell privacy concerns. The online retailer has agreed to pay a $25 million settlement over Federal Trade Commission (FTC) allegations it violated child privacy through Alexa. The company allegedly fell afoul of both the FTC Act and the Children's Online Privacy Protection Act (COPPA) through its handling of kids' voice data.

While the company said it limited access to voice data and promised to delete info on request, it kept children's recordings and transcripts indefinitely by default until September 2019, according to the FTC. The Commission also says Amazon failed to delete transcripts for a "significant period of time," and was still holding on to voice info and geolocation data.

In response, Amazon has issued a statement that it 'disagrees' with the FTC's claims and is settling to put the matter "behind us." It maintains that Amazon Kids was designed with COPPA in mind, and that parents have easy ways to delete recordings and transcripts on the web. The company also says it will delete child profiles that haven't been used for at least 18 months, addressing an FTC concern that data could linger for years.

This is a developing story. Please check back for updates.

GM announced today that it’s making five active safety features standard on all its 2023 and newer EVs. In addition, the automaker says it will surpass its commitment to add Automatic Emergency Braking (AEB) to 95 percent of its vehicles; the feature is included in 98 percent of GM’s 2023 models — including all its EVs. The disclosure follows the US National Highway Traffic Safety Administration (NHTSA) stating today that all new passenger cars and light trucks will be required to include AEB within three years.

In addition to AEB, GM says four other safety features will be standard in most of its 2023 model-year vehicles. These include Forward Collision Alert (warns you if a front-facing crash is likely), Front Pedestrian Braking (alerts you to oncoming pedestrians, automatically braking if needed), Lane Keep Assist with Lane Departure Warning (lets you know if you’re drifting out of your lane) and IntelliBeam (automatic high-beam headlight controls). “As we look ahead toward a future vision of zero crashes, zero emissions and zero congestion, these technologies are significant building blocks with proven benefits for reducing common crashes,” said John Capp, GM’s director of Vehicle Safety Technology, Strategy and Regulations.

The automaker touts its adherence to safety research data to inform the decision, citing a 2023 study at the University of Michigan Transportation Research Institute (UMTRI) that showed AEB combined with Forward Collision Alert reduced rear-end crashes by 42 percent. The same research found that Front Pedestrian Braking cut head-on pedestrian collisions by 23 percent, while Lane Keep Assist with Lane Departure Warning lowered roadway departure accidents (crashes stemming from a car veering off-road or into other lanes) by 15 percent. Finally, the automaker sourced a separate 2022 UMTRI study demonstrating that IntelliBeam cut nighttime crashes involving pedestrians, bicyclists and animals by 22 percent.

GM is also upgrading the front sensors in the Cadillac Lyriq, 2023 Chevy Colorado and GMC Canyon. The enhanced sensors, with a wider field of vision, extend AEB operation to speeds of up to 80 mph while adding Bicyclist Automatic Emergency Braking to Front Pedestrian Braking — a welcome feature in cities with cyclists zipping in and out of traffic. In addition, the sensors enable smoother Lane Keep Assist engagement and Blind Zone Steering Assist (a short / sharp turn to avoid lane-change crashes). The automaker says the upgraded sensors and corresponding features will expand to other models in coming years.

The company emphasized that its safety features aren’t only for the wealthy and will also cover some of its cheapest gas-powered vehicles. For example, the 2024 Chevrolet Trax ($21,495 and up) will include all five of the safety features now standard in its more expensive EVs. “With this commitment, customers across all price points and trim levels will benefit from this set of proven safety technologies,” the company said.

Amazon will pay $5.8 million to settle a recently announced privacy lawsuit raised by the Federal Trade Commission. According to the federal complaint, Ring had allowed employees and third-party contractors full access to customer video content. This unfettered access was abused by some employees and contractors and was used to violate the privacy of both Ring customers and other workers.

The lawsuit highlights several cases where hacked cameras were used to sexually harass and spy on women, stalk female co-workers who used Ring products, utter racial slurs toward children and more. The FTC alleges that Amazon failed to address these issues after buying the company in 2018, noting that security issues persisted from 2016 through 2020.

Amazon has previously fired employees for inappropriate use of customer data, telling US Senators in 2020 that at least four employees had been terminated for accessing or attempting to access data that "exceeded what was necessary for their job functions."

This isn't the first time Ring Doorbells have contended with privacy complaints, either. The company faced scrutiny last year for giving police footage from Ring products without user consent. In 2020, hackers used illicit access to Ring cameras to use in livestream swattings.

Amazon did not immediately respond to a request for comment.

• Read more about trigger pullup switch with another controller output ?
• Log in or register to post comments

Streaming services often upload a pilot on YouTube to lure in new subscribers, but never an entire season. Paramount+ did just that, however, throwing up the entire first season of the critically-acclaimed Star Trek: Strange New Worlds on YouTube for all to see. That’s ten hours of free Star Trek goodness, for those keeping score.

It’s no coincidence that this move comes hot on the heels of season two of the sci-fi property, which premieres June 15th and promises more space-based tomfoolery and an anticipated crossover event with sibling show Star Trek: Lower Decks. The upcoming season also promises more Kirk and more Klingons, if classic franchise iconography is your bag.

Star Trek: Strange New Worlds is a prequel to the original series that follows the adventures of Enterprise captain Christopher Pike, his science officer Spock, a fresh-faced comms officer named Nyota Uhura and the rest of the crew. The show’s been praised for its “back to basics” formula, eschewing extensive serialization in favor of standalone episodes.

This promotion isn’t just for Trek fans, as Paramount+ also dropped full seasons of iCarly, Joe Pickett and Queen of the Universe. It goes without saying that this is a limited time promo offer, so open up those YouTube links and get binging.

Sony has revealed the PlayStation Plus monthly games lineup for June. These are the three titles that folks on the Essential, Extra and Premium tiers can all claim and retain access to as long as they continue to subscribe. The headliner for June is NBA 2K23. The most recent edition of the long-running series features the return of the Jordan Challenge, in which you can play through 15 key moments from Michael Jordan’s legendary career.

Also in the lineup is Jurassic World Evolution 2, a construction and management sim in which you'll build your own Jurassic World theme park. Perhaps you'll have better luck at keeping the dinosaurs in their enclosures than the characters from the movies.

The third game PS Plus subscribers will able to snag next month is Trek to Yomi. It's an Akira Kurosawa-inspired action-adventure set in the Edo period of Japan. You play as a young samurai who seeks revenge against those who attacked his village. I enjoyed Trek to Yomi quite a bit. It's more style than substance, but goodness is that style gorgeous — I've used a screenshot as my desktop background for the last year.

You can add these games to your library at no extra cost starting on June 6th. PS4 and PS5 versions of all three will be available. In the meantime, you still have a chance to claim May's games: Grid Legends, Chivalry 2 and Descenders.

The National Eating Disorder Association (NEDA) was forced to take down its Tessa chatbot after it “may have given information that was harmful and unrelated to the program”, according to an official social media post. Simply put, the AI chatbot was intended to help people dealing with emotional distress, but instead just made things worse by offering dieting advice and urging users to weigh and measure themselves.

Multiple users and experts in the field of eating disorders have experienced the issues first hand, claiming that the bot didn’t respond to simple prompts like “I hate my body” and that it constantly emphasized the importance of dieting and increased physical activity, as reported by Gizmodo. Again, this is a helpline for those with an eating disorder, not a weight loss support group.

The organization says this is a temporary shutdown until it fixes whatever “bugs” and “triggers” led to the chatbot dispensing dangerous information like an appointment with Dr. Oz. You’d think with such an extreme outcome, they’d be thinking about trashing the project entirely, but there’s more to the story.

The whole reason NEDA was relying on the chatbot in the first place is because it allegedly fired human staffers after they tried to unionize, as originally reported by Vice. The long-running phone helpline was manned by both paid staffers and volunteers, and former staff claim the mass firing was a direct consequence of the pro-union effort.

“NEDA claims this was a long-anticipated change and that AI can better serve those with eating disorders. But do not be fooled—this isn’t really about a chatbot. This is about union busting, plain and simple,” former helpline associate Abbie Harper wrote in a blog post on Labor Notes.

Even with this recent snafu, the helpline’s still set to vanish tomorrow. Before this issue came to the public’s attention, NEDA had been transitioning unpaid volunteers away from one-on-one conversations with sufferers and toward training the chatbot. We’ll see if that changes. In the meantime, union-busters gonna bust, am I right?