AMC Entertainment, the largest movie theater chain in the world, will offer open captioning at 240 locations in the United States, a move that the company's chief executive described as "a real advance for those with hearing difficulties or where English is a second language." From a report: Movie theaters provide closed captioning through devices that some customers describe as inconvenient and prone to malfunctioning. Open captions, however, are displayed on the screen in a way similar to subtitles; everyone in the theater sees the same captions, on the same screen. Advocates for the deaf and hard of hearing have long sought more and higher-quality captioning, but theater owners worry that people who aren't deaf simply don't like seeing captions at the movies.

"In some cases, putting open captions on the screen diminishes ticket sales for the movie," said John Fithian, the president and chief executive of the National Association of Theatre Owners, although he noted that the evidence was mostly anecdotal. He said the industry, whose business has been battered by the pandemic, was studying the relationship between open captions and ticket sales. Christian Vogler, a professor at Gallaudet University, a school in Washington that serves the deaf, said in an email, "Detractors of open captions often have argued that the wider hearing audience would revolt over them, or that these would be a losing business proposition for theaters." He praised AMC's move, which was announced last week, saying, "The fact that a large national chain has had a change of heart is significant, and may even open the floodgates for others to follow suit."

"Facebook is trying to mend its reputation in the wake of whisleblower Frances Haugen's testimony," reports Engadget, "and that includes promises of features lessening the potential harm for teens." CNN and Reuters report that Facebook Global Affairs VP Nick Clegg promised Instagram would introduce a "take a break" feature that encouraged teens to simply stop using the social network for a while.

Clegg didn't say when it would be ready, but this was clearly meant to reduce addiction and other unhealthy behavior.

The social media exec also said Facebook would "nudge" teens away from material in its apps that "may not be conducive to their well-being." He didn't provide specifics for this new approach. He did, however, suggest that Facebook's algorithms should be "held to account," including by regulation if needed, to be sure real-world results matched intentions... Breaks and nudges may reduce exposure to harmful content, but they won't remove the content in question. Clegg's statements also reflect a familiar strategy at Facebook. It likes to invite regulation, but only the regulation it's comfortable with. While the proposed changes could help, politicians may demand more — in part to prevent Facebook from dictating its own regulation.
According to Reuters, Clegg also "said he could not answer the question whether its algorithms amplified the voices of people who had attacked the U.S. Capitol on January 6th."

Gizmodo takes a hard look at a "growing sense of excitement" about collaboration between bitcoin-mining operations and nuclear power plants (which are now plagued by high operating costs compared to renewables as well as natural gas): Of the three partnerships between bitcoin companies and nuclear energy that the Wall Street Journal mentioned, two involve bitcoin miners partnering with existing nuclear sources to power their operations... These are not companies investing in the future, but rather companies searching for anything that will help keep the profits flowing using existing power plants. It's pretty safe to say that some cash-strapped owners of nuclear plants will be using mining partnerships not to make any technological strides, but rather to simply keep the old plants operating.

"The plants themselves are pretty well-run, and they know what they're doing," said Alex Gilbert, a project manager at the think tank Nuclear Innovation Alliance. "It really is a matter of the economics. There's a certain point where you're definitely unprofitable, and you're going to be likely to close because you're not getting enough money in power markets. But if a bitcoin operation takes 10 to 15 to 30 percent of your power at a reasonable price, that tips you into profitability." This profitability means the plants can stay open, giving miners a little carbon-free energy as a treat while keeping the U.S.'s biggest source of zero-emissions power operational. This is especially a good idea while we wait for more renewables — and policies that favor them — to come online, in what could be the first real-world proof bitcoin is doing some societal good instead of being a waste of energy and resources....

A few small-to-medium reactors should be ready for licensing in a few years and some over the next decade, he said, helped along by private and federal funding. To actually get to a point where the kinds of smaller reactors could be developed that would be competitive with the (rapidly falling) price of renewables, Gilbert said, would take a significantly larger bump from private capital — as well as more customers. "Providing early demand for advance reactors, especially microreactors, that's how bitcoin can most help the nuclear sector," he said.... I'm not a technofuturist who dreams of a libertarian paradise, but I have to admit that there's kind of a cool idea here. If the bitcoin community really believes cryptocurrencies are the money of the future, let them be the first to invest in a budding technology that could be the energy of the future.

In the interim, however, they shouldn't be allowed to rest on their greenwashing laurels while continuing to churn out emissions as they wait for fast reactor technology to become feasible in 10 years. Government regulations are, of course, anathema to crypto true believers. But a mandate that any new mining facilities source power from nearby nuclear plants could go a long way toward cleaning up bitcoin's act and ensuring the carbon-free energy we desperately need stays on the grid while fancy fast reactors come online.

An anonymous reader quotes a report from Motherboard: The FCC has announced (PDF) it's investigating deals the broadband industry strikes with landlords that block broadband competition in apartment complexes, condos, and developments. While the FCC passed rules in 2008 attempting to prevent such deals, Internet Service Providers (ISPs) have exploited massive loopholes in the restrictions for more than a decade. "With more than one-third of the U.S. population living in condos and apartment buildings, it's time to take a fresh look at how exclusive agreements between carriers and building owners could lock out broadband competition and consumer choice," interim FCC boss Jessica Rosenworcel said of the announcement. "I look forward to reviewing the record."

The inquiry comes after President Biden signed an executive order in July urging regulators to take a closer look at competition and monopoly issues in several sectors. The order also mandated the creation of a competition council, which urged the FCC to take a closer look at the anticompetitive nature of these arrangements. The FCC's existing rules technically bar landlords and ISPs from colluding to restrict broadband competition. But in a 2016 piece in Wired, Harvard Law Professor Susan Crawford outlined the various ways big telecom wiggles around the restrictions -- often by simply calling what they're doing -- something else. "Sure, a landlord can't enter into an exclusive agreement granting just one ISP the right to provide Internet access service...but a landlord can refuse to sign agreements with anyone other than Big Company X, in exchange for payments labeled in any one of a zillion ways," Crawford wrote. "Exclusivity by any other name still feels just as abusive."

For example, to get around FCC rules expanding access to an ISP's in-building wiring, companies like Comcast or Charter will often deed ownership of these wires to a landlord, then turn around and pay that landlord to ensure that nobody else can have access. Because the landlord now technically owns the wires, the FCC rules no longer apply. ISPs also pay landlords to sign agreements that ban any other competing ISPs from advertising in the building. If you're a landlord that violates such arrangements, you can then expect a nastygram from a company like Comcast for violating your deal. In addition, many landlords will charge "door fees" to any company that needs access to a building to install new wiring, creating an additional layer of difficulty and expense for smaller broadband competitors trying to compete with dominant ISPs. Collectively such restrictions serve the same function as blocking broadband competition outright. Much as it does on the national level, this lack of block by block competition directly contributes to higher prices, slower speeds, and comically-terrible customer service.

"ShotSpotter" is an AI-powered tool that claims it can detect the sound of gunshots. To install it can cost up to $95,000 per square mile — every year — reports the Associated Press.

There's just one problem. "The algorithm that analyzes sounds to distinguish gunshots from other noises has never been peer reviewed by outside academics or experts." "The concern about ShotSpotter being used as direct evidence is that there are simply no studies out there to establish the validity or the reliability of the technology. Nothing," said Tania Brief, a staff attorney at The Innocence Project, a nonprofit that seeks to reverse wrongful convictions.

A 2011 study commissioned by the company found that dumpsters, trucks, motorcycles, helicopters, fireworks, construction, trash pickup and church bells have all triggered false positive alerts, mistaking these sounds for gunshots. ShotSpotter CEO Ralph Clark said the company is constantly improving its audio classifications, but the system still logs a small percentage of false positives. In the past, these false alerts — and lack of alerts — have prompted cities from Charlotte, North Carolina, to San Antonio, Texas, to end their ShotSpotter contracts, the AP found.
And the potential for problems isn't just hypothetical. Just ask 65-year-old Michael Williams: Williams was jailed last August, accused of killing a young man from the neighborhood who asked him for a ride during a night of unrest over police brutality in May... "I kept trying to figure out, how can they get away with using the technology like that against me?" said Williams, speaking publicly for the first time about his ordeal. "That's not fair." Williams sat behind bars for nearly a year before a judge dismissed the case against him last month at the request of prosecutors, who said they had insufficient evidence.

Williams' experience highlights the real-world impacts of society's growing reliance on algorithms to help make consequential decisions about many aspects of public life... ShotSpotter evidence has increasingly been admitted in court cases around the country, now totaling some 200. ShotSpotter's website says it's "a leader in precision policing technology solutions" that helps stop gun violence by using "sensors, algorithms and artificial intelligence" to classify 14 million sounds in its proprietary database as gunshots or something else. But an Associated Press investigation, based on a review of thousands of internal documents, emails, presentations and confidential contracts, along with interviews with dozens of public defenders in communities where ShotSpotter has been deployed, has identified a number of serious flaws in using ShotSpotter as evidentiary support for prosecutors. AP's investigation found the system can miss live gunfire right under its microphones, or misclassify the sounds of fireworks or cars backfiring as gunshots.

Forensic reports prepared by ShotSpotter's employees have been used in court to improperly claim that a defendant shot at police, or provide questionable counts of the number of shots allegedly fired by defendants. Judges in a number of cases have thrown out the evidence... The company's methods for identifying gunshots aren't always guided solely by the technology. ShotSpotter employees can, and often do, change the source of sounds picked up by its sensors after listening to audio recordings, introducing the possibility of human bias into the gunshot detection algorithm. Employees can and do modify the location or number of shots fired at the request of police, according to court records. And in the past, city dispatchers or police themselves could also make some of these changes.
Three more eye-popping details from the AP's 4,000-word exposé

• "One study published in April in the peer-reviewed Journal of Urban Health examined ShotSpotter in 68 large, metropolitan counties from 1999 to 2016, the largest review to date. It found that the technology didn't reduce gun violence or increase community safety..."

• "Forensic tools such as DNA and ballistics evidence used by prosecutors have had their methodologies examined in painstaking detail for decades, but ShotSpotter claims its software is proprietary, and won't release its algorithm..."

• "In 2018, it acquired a predictive policing company called HunchLab, which integrates its AI models with ShotSpotter's gunshot detection data to purportedly predict crime before it happens."

Chet Haase, who worked at several Silicon Valley tech companies and in 2010 joined the Android engineering team at Google and watched Android rise from the bottom of the smartphone field to where it is today, writes in a new book: The final part of the pitch (and the most important part, for the VCs they were pitching to) was how Android was going to make money. The open source platform described in the slides is essentially what the Android team eventually built and shipped. But if that was all there was, the company would not have been worth funding for VCs. Developing and giving away an open source platform sounds great from a save-the-world standpoint, but where's the payoff? Where's the upside for investors? That is, how did Android plan to make money off of a product that they planned to simply give away? Venture capitalists fund companies that they hope will make more (far more) than their investment back.

The path to revenue was clear for the other platform companies in the game. Microsoft made money by licensing its platform to Windows Phone partners; every phone sold contributed a per-device cost back to Microsoft. RIM made money both on the handsets they sold as well as the lucrative service contracts that their loyal enterprise customers signed up for. Nokia and the other Symbian adopters made money by selling the phones that they manufactured with variations of that operating system. Similarly, all of the other handset manufacturers funded their own software development through the revenue generated by the phones they sold.

So what was Android's play that would fund the development of this awesome platform that they had yet to build and which they would give away free to other manufacturers to build their own devices? Carrier services.

Carriers would provide applications, contacts, and other cloud-based data services to their customers for Android-based handsets. The carriers would pay Android for providing these services. Swetland explained: "Rather than running and hosting the services [like Danger did for its Hiptop phones], we would build the services and sell them to the carriers." (In fact, the system that the team eventually built and shipped stayed true to the vision laid out in the pitch deck, except for this part about revenue from carrier services, which went away entirely.)

A previously undocumented Android-based remote access trojan (RAT) has been found to use screen recording features to steal sensitive information on the device, including banking credentials, and open the door for on-device fraud. The Hacker News reports: Dubbed "Vultur" due to its use of Virtual Network Computing (VNC)'s remote screen-sharing technology to gain full visibility on targeted users, the mobile malware was distributed via the official Google Play Store and masqueraded as an app named "Protection Guard," attracting over 5,000 installations. Banking and crypto-wallet apps from entities located in Italy, Australia, and Spain were the primary targets. "For the first time we are seeing an Android banking trojan that has screen recording and keylogging as the main strategy to harvest login credentials in an automated and scalable way," researchers from ThreatFabric said in a write-up shared with The Hacker News. "The actors chose to steer away from the common HTML overlay development we usually see in other Android banking Trojans: this approach usually requires a larger time and effort investment from the actors to create multiple overlays capable of tricking the user. Instead, they chose to simply record what is shown on the screen, effectively obtaining the same end result."

Vultur [...] takes advantage of accessibility permissions to capture keystrokes and leverages VNC's screen recording feature to stealthily log all activities on the phone, thus obviating the need to register a new device and making it difficult for banks to detect fraud. What's more, the malware employs ngrok, a cross-platform utility used to expose local servers behind NATs and firewalls to the public internet over secure tunnels, to provide remote access to the VNC server running locally on the phone. Additionally, it also establishes connections with a command-and-control (C2) server to receive commands over Firebase Cloud Messaging (FCM), the results of which, including extracted data and screen captures, are then transmitted back to the server.

ThreatFabric's investigation also connected Vultur with another well-known piece of malicious software named Brunhilda, a dropper that utilizes the Play Store to distribute different kinds of malware in what's called a "dropper-as-a-service" (DaaS) operation, citing overlaps in the source code and C2 infrastructure used to facilitate attacks. These ties, the Amsterdam-based cybersecurity services company said, indicate Brunhilda to be a privately operating threat actor that has its own dropper and proprietary RAT Vultur.

Forget the home office -- 45% of American teleworkers regularly work from a couch, 38% regularly work from bed and 20% often work outside, according to a study by the home improvement marketing firm CraftJack. Axios reports: People have spent an average of $268 trying to improve their remote work setups, but a whopping 50% still say the pain and discomfort of working from home is enough to send them back to the office. It's not enough for companies to provide stipends for teleworkers to buy ergonomic chairs or desks, Axios' Kia Kokalitcheva notes. Many people simply do not have the space allocated inside their homes for an office setup, and it can be too expensive to move to a bigger place.

Today, Amaon said it will be upgrading almost every plug-in Echo smart speaker to support Matter, a cross-platform open-source standard coming later this year. This includes most Echo and Echo Dot speakers and every Echo Studio, Echo Show, Echo Plus, and Echo Flex. "In fact, the only Echo smart speakers that won't get upgraded to Matter are the first-gen Echo, first-gen Echo Dot and Echo Tap," reports The Verge. From the report: While the company doesn't provide a timeline for those upgrades, the general idea is that Matter will launch by late 2021, so it shouldn't be long until Amazon's newest and / or more popular devices receive the capability. A bigger question is whether any of them will work as Matter hubs. Google announced in May that in addition to upgrading its Nest devices to Matter, it would allow its devices that support the Thread protocol (like the Nest Wi-Fi, Nest Hub Max, and second-gen Nest Hub) to double as connection hubs for Matter, too, not simply as a voice assistant to control Matter gadgets. But while Amazon's Eero routers were early to adopt Thread, Amazon's Echo smart speakers were not.

Reactions are starting to come in for GitHub's new Copilot coding tool, which one site calls "a product of the partnership between Microsoft and AI research and deployment company OpenAI — which Microsoft invested $1 billion into two years ago." According to the tech preview page: GitHub Copilot is currently only available as a Visual Studio Code extension. It works wherever Visual Studio Code works — on your machine or in the cloud on GitHub Codespaces. And it's fast enough to use as you type. "Copilot looks like a potentially fantastic learning tool — for developers of all abilities," said James Governor, an analyst at RedMonk. "It can remove barriers to entry. It can help with learning new languages, and for folks working on polyglot codebases. It arguably continues GitHub's rich heritage as a world-class learning tool. It's early days but AI-assisted programming is going to be a thing, and where better to start experiencing it than GitHub...?"

The issue of scale is a concern for GitHub, according to the tech preview FAQ: "If the technical preview is successful, our plan is to build a commercial version of GitHub Copilot in the future. We want to use the preview to learn how people use GitHub Copilot and what it takes to operate it at scale." GitHub spent the last year working closely with OpenAI to build Copilot. GitHub developers, along with some users inside Microsoft, have been using it every day internally for months.

[Guillermo Rauch, CEO of developer software provider Vercel, who also is founder of Vercel and creator of Next.js], cited in a tweet a statement from the Copilot tech preview FAQ page, "GitHub Copilot is a code synthesizer, not a search engine: the vast majority of the code that it suggests is uniquely generated and has never been seen before."

To that, Rauch simply typed: "The future."

Rauch's post is relevant in that one of the knocks against Copilot is that some folks seem to be concerned that it will generate code that is identical to code that has been generated under open source licenses that don't allow derivative works, but which will then be used by a developer unknowingly...
GitHub CEO Nat Friedman has responded to those concerns, according to another article, arguing that training an AI system constitutes fair use: Friedman is not alone — a couple of actual lawyers and experts in intellectual property law took up the issue and, at least in their preliminary analysis, tended to agree with Friedman... [U.K. solicitor] Neil Brown examines the idea from an English law perspective and, while he's not so sure about the idea of "fair use" if the idea is taken outside of the U.S., he points simply to GitHub's terms of service as evidence enough that the company can likely do what it's doing. Brown points to passage D4, which grants GitHub "the right to store, archive, parse, and display Your Content, and make incidental copies, as necessary to provide the Service, including improving the Service over time." "The license is broadly worded, and I'm confident that there is scope for argument, but if it turns out that Github does not require a license for its activities then, in respect of the code hosted on Github, I suspect it could make a reasonable case that the mandatory license grant in its terms covers this as against the uploader," writes Brown. Overall, though, Brown says that he has "more questions than answers."
Armin Ronacher, the creator of the Flask web framework for Python, shared an interesting example on Twitter (which apparently came from the game Quake III Arena) in which Copilot apparently reproduces a chunk of code including not only its original comment ("what the fuck?") but also its original copyright notice.

In a surprise announcement today, IBM announced that Jim Whitehurst, who came over in the Red deal, would be stepping down as company president just 14 months after taking over in that role. From a report: IBM didn't give a lot of details as to why he was stepping away, but acknowledged his key role in helping bring the 2018 $34 billion Red Hat deal to fruition and helping bring the two companies together after the deal closed. "Jim has been instrumental in articulating IBM's strategy, but also, in ensuring that IBM and Red Hat work well together and that our technology platforms and innovations provide more value to our clients," the company stated.

He will stay on as a senior advisor to Krishna, but it begs the question why he is leaving after such a short time in the role, and what he plans to do next. Oftentimes after a deal of this magnitude closes, there is an agreement as to how long key executives will stay. It could be simply that the period has expired and Whitehurst wants to move on, but some saw him as the heir apparent to Krishna and the move comes as a surprise when looked at in that context.

Patrick Klepek writes via Motherboard: Few role-playing experiences are as beloved as the original Final Fantasy games, which is why Square Enix announcing a new brand it's calling Pixel Remasters for the first six games was greeted with equal parts shock and horror. For every brilliant reinvention, like last year's Final Fantasy 7 Remake, you have these nightmarish updates to classics like Final Fantasy 6 that are so abjectly awful to look at that fans created mods to try and replace the visuals. It's not really clear what Square Enix wants to accomplish with these Pixel Remasters, but what's abundantly clear is that Square Enix intends to revisit the visuals across each 2D game. The new sprites aren't massive departures from the originals, but they're different, and it's led to speculation about whether the company is going to address a longstanding issue with older games being released on fancy new televisions and computer monitors.

I've always loved the way video games looked -- fuzzy and crunchy -- on those humorously heavy and bulky older cathode-ray tube (CRT) TVs that used to populate family rooms. What I didn't know until earlier this year, however, was the science behind it all. It's not just that high-definition displays provide a crisper look at art made in earlier eras of video games, but that art was specifically drawn knowing it would ultimately pipe through a CRT, and when that art is viewed on a modern, non-CRT display, you're actually losing some intended detail. [...] The problem is many people will never experience it in real-life, and so filters and similar technologies are essentially forms of emulation for television tech. More than 705 million CRT TVs have been sold in the United States since 1980, and the vast majority of these environmentally unfriendly devices are in the process of being broken down and recycled. That process will take years. But more practically, nobody is making CRT TVs anymore, and as the existing supply naturally breaks down, it falls to hobbyists to keep them ticking. No great shock to learn that Starkweather isn't a huge fan of Square Enix's approach for the Pixel Remasters, partially because it risks erasing the work of the original artists. One solution that Starkweather proposes is Square Enix spending time on a refined CRT filter.

"Filters are simply filters and they change visuals without having any artistic intention behind," said renowned pixel artist Thomas Feichtmeir. "I have not yet seen any CRT filter implemented in a game which truly simulated a realistic CRT experience." While naive folks like myself learned about CRT through a Twitter account, Feichtmeir had a similar realization years ago. At home, Feichtmeir had a CRT monitor next to an LCD laptop, and as he transferred his dawn pixels from one to the other, it dawned upon him that they looked different. He noticed a similar issue playing games re-released on modern displays. "If you make a piece of pixel art on a LCD and you put it on a CRT," he said, "it's the equivalent of taking one of your articles, putting it through Google Translate and to expect that the other language it comes out [with] will have perfect meaning and grammar. A whole field of 'localization' exists for writing and in the game industry to address those issues." Though Feichtmeir has no specific insight into what Square Enix is or isn't planning for its Pixel Remasters series, watching what's been released gave him pause on the CRT theory. "Considering the couple of screenshots and snippets we saw in the presentation, I would not say any of it really accounts for the gap between CRT and LCDs," he said. "We still can see a lot of techniques which theoretically should stay on a CRT -- like overly dithered textures or just color optimized battle backgrounds. The biggest change are the characters, where they basically removed the volumetric shading in exchange for a dark outlined flat style. In my eyes this just changes the style to something which does not feel close to the original. And I think what a remaster should deliver on is to recreate the feeling how the original game felt."

"The official Python software package repository, PyPI, is getting flooded with spam packages..." Bleeping Computer reported Thursday.

"Each of these packages is posted by a unique pseudonymous maintainer account, making it challenging for PyPI to remove the packages and spam accounts all at once..." PyPI is being flooded with spam packages named after popular movies in a style commonly associated with torrent or "warez" sites that provide pirated downloads: watch-(movie-name)-2021-full-online-movie-free-hd-... Although some of these packages are a few weeks old, BleepingComputer observed that spammers are continuing to add newer packages to PyPI... The web page for these bogus packages contain spam keywords and links to movie streaming sites, albeit of questionable legitimacy and legality...

February of this year, PyPI had been flooded with bogus "Discord", "Google", and "Roblox" keygens in a massive spam attack, as reported by ZDNet. At the time, Ewa Jodlowska, Executive Director of the Python Software Foundation had told ZDNet that the PyPI admins were working on addressing the spam attack, however, by the nature of pypi.org, anyone could publish to the repository, and such occurrences were common.

Other than containing spam keywords and links to quasi-video streaming sites, these packages contain files with functional code and author information lifted from legitimate PyPI packages... As previously reported by BleepingComputer, malicious actors have combined code from legitimate packages with otherwise bogus or malicious packages to mask their footsteps, and make the detection of these packages a tad more challenging...

In recent months, the attacks on open-source ecosystems like npm, RubyGems, and PyPI have escalated. Threat actors have been caught flooding software repositories with malware, malicious dependency confusion copycats, or simply vigilante packages to spread their message. As such, securing these repositories has turned into a whack-a-mole race between threat actors and repository maintainers.

An anonymous reader quotes a report from MIT Technology Review: In 1998 a couple of Stanford graduate students published a paper describing a new kind of search engine: "In this paper, we present Google, a prototype of a large-scale search engine which makes heavy use of the structure present in hypertext. Google is designed to crawl and index the Web efficiently and produce much more satisfying search results than existing systems." The key innovation was an algorithm called PageRank, which ranked search results by calculating how relevant they were to a user's query on the basis of their links to other pages on the web. On the back of PageRank, Google became the gateway to the internet, and Sergey Brin and Larry Page built one of the biggest companies in the world. Now a team of Google researchers has published a proposal for a radical redesign that throws out the ranking approach and replaces it with a single large AI language model, such as BERT or GPT-3 -- or a future version of them. The idea is that instead of searching for information in a vast list of web pages, users would ask questions and have a language model trained on those pages answer them directly. The approach could change not only how search engines work, but what they do -- and how we interact with them.

[Donald Metzler and his colleagues at Google Research] are interested in a search engine that behaves like a human expert. It should produce answers in natural language, synthesized from more than one document, and back up its answers with references to supporting evidence, as Wikipedia articles aim to do. Large language models get us part of the way there. Trained on most of the web and hundreds of books, GPT-3 draws information from multiple sources to answer questions in natural language. The problem is that it does not keep track of those sources and cannot provide evidence for its answers. There's no way to tell if GPT-3 is parroting trustworthy information or disinformation -- or simply spewing nonsense of its own making.

Metzler and his colleagues call language models dilettantes -- "They are perceived to know a lot but their knowledge is skin deep." The solution, they claim, is to build and train future BERTs and GPT-3s to retain records of where their words come from. No such models are yet able to do this, but it is possible in principle, and there is early work in that direction. There have been decades of progress on different areas of search, from answering queries to summarizing documents to structuring information, says Ziqi Zhang at the University of Sheffield, UK, who studies information retrieval on the web. But none of these technologies overhauled search because they each address specific problems and are not generalizable. The exciting premise of this paper is that large language models are able to do all these things at the same time, he says.

"Salon reports amateur-hour mistakes in the attempted rollout of FRANK, a social media site envisioned by Mike Lindell of MyPillow," writes Slashdot reader Tom239. "A Drupal expert described the code as 'not even student work.'" From the report: Speaking to Salon on Thursday afternoon about Lindell's site, one "Acquia Certified Drupal Grand Master," who oversees a technology firm that employs numerous other "grandmasters," said that Lindell's site was set up for failure from its inception, noting that its developers -- whom Lindell compared to Navy SEALs -- had failed to carry out basic "Drupal 101" tasks. One coder who spoke to Salon in great detail explained the potential shortcomings of the pillow maven's program code and the patchy work done by his developer team. "Drupal can power high powerful websites, sites with lots of traffic," the expert said, adding that it isn't the right software to build a social media site with, since it's not designed to handle a large amount of user-generated content. "Lindell's website was basically trying to make soup for scratch for everybody," said the expert, who claimed more than 25 years of experience in the IT field.

"In my professional opinion, it will be extremely unlikely, if not impossible, for Lindell to accomplish his vision with Drupal and his own servers," the expert told Salon. "Despite how much I love it, Drupal simply isn't the right tool for the number of users with the features that he wants to provide. It would take a massive effort of 12 to 18 months to build out the needed hosting setup and application architecture, and this would come with an enormous degree of risk. The idea that he could do this in just a couple of months is patently absurd, and I think the results speak for themselves."

"When I was looking at the code, in the browser, they basically launched the site while it was still in development mode," one expert told Salon, citing the fact that developers had failed to check a box to aggregate files on the platform as the first red flag he ran across. "Their files were not aggregated, and by the way, that's a check box in Drupal -- you literally check a box and click save, My jaw dropped when I saw that. I was like, 'They did not try to launch this thing without aggregation turned on!'" The second major red flag another Drupal expert found was that Lindell's site was spitting out coded error messages to users, which leaves the platform vulnerable to attacks. "This is a shit show," the expert said, calling this an "obvious" issue that coders learn how to prevent in "Drupal 101."
Elsewhere it was reported that Lindell's supposed free-speech haven will not allow swearing, pornography, or the use of 'god's name in vain'.

The New York Times explores whether Substack is just a company that makes it easy to charge for newsletters — or a new direct-to-consumer media that's part of a larger cultural shift? This new ability of individuals to make a living directly from their audiences isn't just transforming journalism. It's also been the case for adult performers on OnlyFans, musicians on Patreon, B-list celebrities on Cameo. In Hollywood, too, power has migrated toward talent, whether it's marquee showrunners or actors. This power shift is a major headache for big institutions, from The New York Times to record labels. And Silicon Valley investors, eager to disrupt and angry at their portrayal in big media, have been gleefully backing it. Substack embodies this cultural shift, but it's riding the wave, not creating it...

A New York Times opinion writer, Charlie Warzel, is departing to start a publication on Substack called Galaxy Brain... The Times wouldn't comment on his move, but is among the media companies trying to develop its own answer to Substack and recently brought the columnist Paul Krugman's free Substack newsletter to the Times platform... [T]he biggest threat to Substack is unlikely to be the Twitter-centric political battles among some of its writers. The real threat is competing platforms with a different model. The most technically powerful of those is probably Ghost, which allows writers to send and charge for newsletters, with monthly fees starting at $9. While Substack is backed by the venture capital firm Andreessen Horowitz, Ghost has Wikipedia vibes: It is open-source software developed by a nonprofit...

And it's easy to leave. Unlike on Facebook or Twitter, Substack writers can simply take their email lists and direct connections to their readers with them. Substack's model of taking 10 percent of its writers' subscriptions is "too greedy of a slice to take of anyone's business with very little in return," said Ghost's founder and chief executive, John O'Nolan, a tattooed, nomadic Irishman who is bivouacked in Hollywood, Fla. He said he believed subscription newsletter publishing was "destined to be commoditized."

But Ghost represents an even purer departure from legacy media. More than half of the sites on the platform simply run the software off their own servers. "The technology is designed to be decentralized, and there's no one institution or one corporation that can decide what is OK," he said.
The article also notes that Twitter recently bought the newsletter platform Revue, while Facebook "is developing ambitious plans for a rival that will provide a platform for local journalists, among other writers."

And in a section on indie spirit, it adds as an aside that Bustle Digital Group "confirmed to me that it's reviving the legendary blog Gawker under a former Gawker writer, Leah Finnegan."

Astronauts Mike Hopkins and Victor Glover floated back outside the International Space Station Monday and completed solar array battery replacement work that began four years ago. They also installed three new video cameras and made preparations for upcoming work to install new roll-out solar blankets to upgrade the lab's power system. From a report: The work took less time than expected and the astronauts were able to carry out a variety of lower-priority "get-ahead" tasks before returning to Quest airlock and calling it a day, closing out a five-hour 20-minute spacewalk. Taking one last look around before entering the airlock, Hopkins said simply, "Beautiful view." The excursion began at 7:56 a.m. EST when the two men switched their spacesuits to battery power, officially kicking off the 234th spacewalk, or EVA, devoted to station assembly and maintenance since construction began 23 years ago in 1998.

"OK, it's a beautiful day. Let's go for a walk outside," astronaut Bob Hines radioed from mission control in Houston. The first priority was electrically connecting a replacement battery in the station's solar power system. The batteries are critical to station operation. They are recharged by the arrays when the lab is in sunlight and then provide that stored power during periods of orbital darkness. Starting in 2017, spacewalking astronauts began work to replace all 48 of the station's original nickel-hydrogen batteries with 24 smaller, more powerful lithium ion units. During battery replacement work in March 2019, one of the new lithium ion power packs used by the station's left-side inboard set of arrays failed and was replaced with one of the older nickel hydrogen batteries that had been removed. All of the other older batteries were replaced as planned during multiple spacewalks. Over the weekend, flight controllers operating the station's robot arm by remote control disconnected the one remaining nickel hydrogen battery and robotically installed a fresh lithium ion unit. Hopkins and Glover electrically connected the battery, finally completing the power system upgrade four years after the work began.

A new research study at the University of Missouri is using CRISPR gene-editing technology to produce mosquitoes that are unable to replicate Zika virus and therefore cannot infect a human through biting. Slashdot reader wooloohoo shared an announcement from Cornell's Alliance for Science: Alexander Franz, an associate professor in the MU College of Veterinary Medicine, collaborated with researchers at Colorado State University... Their work was recently published in the journal Viruses. Franz added that the genetic modification is inheritable, so future generations of the altered mosquitoes would be resistant to Zika virus as well... "[W]e are simply trying to expand the toolbox and provide a solution by genetically modifying the mosquitoes to become Zika-resistant while keeping them alive at the same time."

Franz' research is designed to help prevent another outbreak of Zika virus disease from occurring while also addressing concerns that have some have raised about reducing populations of mosquitoes, which are a food source for some animals...

The study was funded by the National Institutes of Health.

After Twitter banned President Trump's account last week, the site and its executives, including Twitter CEO Jack Dorsey, were largely silent in justifying their decision. That changed last night when Dorsey, in a series of tweets, explained that he felt banning Trump's account was the right move for the social network. The Verge reports: "Offline harm as a result of online speech is demonstrably real, and what drives our policy and enforcement above all," he wrote. Dorsey blamed Twitter's failure "to promote healthy conversation," acknowledged that Twitter needs to "look critically at inconsistencies of our policy and enforcement," and said that social platforms needed more transparency around moderation. Then, Dorsey plugged an effort to build a decentralized standard for social media he began in 2019 when he sought to hire five engineers to work on it. That's how, eight tweets into a thread on why his company banned Donald Trump, the CEO of Twitter managed to change the subject to his passion for bitcoin.

Banning the RealDonaldTrump Twitter account had "real and significant ramifications," Dorsey wrote. Dorsey said that the widespread suspension of the president by many platforms challenged the notion that if people didn't like Twitter's rules, they could simply go somewhere else. And though the president can issue a press release or call a press conference whenever he wishes -- or simply go on television -- Dorsey expressed concern that the enforcement actions might "erode a free and open global internet."

And that was when bitcoin came up. Dorsey is also CEO of Square, an internet payment company, that bought $50 million of bitcoin as part of a bet on cryptocurrency. Square has accepted bitcoin since 2014. According to Dorsey, bitcoin provides a model for a decentralized model for social media. Dorsey did not elaborate on how such a network might address Twitter's failures in moderation, creating healthy conversations, or provide for more consistent policy enforcment. "It's important that we acknowledge this is a time of great uncertainty and struggle for so many around the world," Dorsey wrote on Twitter. "Our goal in this moment is to disarm as much as we can, and ensure we are all building towards a greater common understanding, and a more peaceful existence on earth. I believe the internet and global public conversation is our best and most relevant method of achieving this."

The California Bar released data last week confirming that during its use of ExamSoft for the October Bar exam, over one-third of the nearly nine-thousand online examinees were flagged by the software. The Electronic Frontier Foundation is concerned that the exam proctoring software is incorrectly flagging students for cheating "due either to the software's technical failures or to its requirements that students have relatively new computers and access to near-broadband speeds." From the report: This is outrageous. It goes without saying that of the 3,190 applicants flagged by the software, the vast majority were not cheating. Far more likely is that, as EFF and others have said before, remote proctoring software is surveillance snake oil -- you simply can't replicate a classroom environment online, and attempting to do so via algorithms and video monitoring only causes harm. In this case, the harm is not only to the students who are rightfully upset about the implications and the lack of proper channels for redress, but to the institution of the Bar itself. While examinees have been searching for help from other examinees as well as hiring legal counsel in their attempt to defend themselves from potentially baseless claims of cheating, the California Committee of Bar Examiners has said "everything is going well" and called these results "a good thing to see" (13:30 into the video of the Committee meeting).

That is not how we see it. These flags have triggered concern for hundreds, if not thousands, of test takers, most of whom had no idea that they were flagged until recently. Many only learned about the flag after receiving an official "Chapter 6 Notice" from the Bar, which is sent when an applicant is observed (supposedly) violating exam conduct rules or seen or heard with prohibited items, like a cell phone, during the exam. In a depressingly ironic introduction to the legal system, the Bar has requested that students respond to the notices within 10 days, but it would appear that none of them have been given enough information to do so, as Chapter 6 Notices contain only a short summary of the violation. These summaries are decidedly vague: "Facial view of your eyes was not within view of the camera for a prolonged period of time"; "No audible sound was detected"; "Leaving the view of the webcam outside of scheduled breaks during a remote-proctored exam." Examinees do not currently have access to the flagged videos themselves, and are not expected to receive access to them, or any other evidence against them, before they are required to submit a response. The report goes on to say that some of these flags are technical issues with ExamSoft. For example, Lenovo laptops appear to have been flagged en masse for an issue with the software's inability to access the internal microphone.

Other flags are likely due to the inability of the software to correctly recognize the variability of examinees' demeanors and expressions. "We implore the California Bar to rethink its plans for remotely-proctored future exams, and to work carefully to offer clearer paths for examinees who have been flagged by these inadequate surveillance tools," the EFF says in closing. "Until then, the Bar must provide examinees who have been flagged with a fair appeals process, including sharing the videos and any other information necessary for them to defend themselves before requiring a written response."