Camel Pilot writes: The new GOP tax plan -- which just passed the House -- will tax tuition waivers as income. Graduate students working as research assistants on meager stipends would have to declare tuition waivers as income on the order of $80,000 income. This will force many graduate students of modest means to quit their career paths and walk away from their research. These are the next generation of scientists, engineers, inventors, educators, medical miracle workers and market makers. As Prof Claus Wilke points out: "This would be a disaster for U.S. STEM Ph.D. education." Slashdot reader Camel Pilot references a report via The New York Times, where Erin Rousseau explains how the House of Representatives' recently passed tax bill affects graduate research in the United States. Rousseau is a graduate student at M.I.T. who studies the neurological basis of mental health disorders. "My peers and I work between 40 and 80 hours a week as classroom teachers and laboratory researchers, and in return, our universities provide us with a tuition waiver for school. For M.I.T. students, this waiver keeps us from having to pay a tuition bill of about $50,000 every year -- a staggering amount, but one that is similar to the fees at many other colleges and universities," he writes. "No money from the tuition waivers actually ends up in our pockets, so under Section 117(d)(5), it isn't counted as taxable income." Rousseau continues by saying his tuition waivers will be taxed under the House's tax bill. "This means that M.I.T. graduate students would be responsible for paying taxes on an $80,000 annual salary, when we actually earn $33,000 a year. That's an increase of our tax burden by at least $10,000 annually."

An anonymous reader writes: Robots are not taking over the world," the diplomat leading the first official talks on autonomous weapons assured on Friday, seeking to head off criticism over slow progress towards restricting the use of so-called "killer robots." The United Nations was wrapping up an initial five days of discussions on weapons systems that can identify and destroy targets without human control, which experts say will soon be battle ready. "Ladies and gentlemen, I have news for you: the robots are not taking over the world. Humans are still in charge," said India's disarmament ambassador, Amandeep Gill, who chaired the CCW meeting. "I think we have to be careful in not emotionalizing or dramatizing this issue," he told reporters in response to criticism about the speed of the conference's work. Twenty-two countries, mostly those with smaller military budgets and lesser technical knowhow, have called for an outright ban, arguing that automated weapons are by definition illegal as every individual decision to launch a strike must be made by a human. Gill underscored that banning killer robots, or even agreement on rules, remained a distant prospect.

An anonymous reader quotes a report from The Register: Three misconfigured AWS S3 buckets have been discovered wide open on the public internet containing "dozens of terabytes" of social media posts and similar pages -- all scraped from around the world by the U.S. military to identify and profile persons of interest. The archives were found by veteran security breach hunter UpGuard's Chris Vickery during a routine scan of open Amazon-hosted data silos, and these ones weren't exactly hidden. The buckets were named centcom-backup, centcom-archive, and pacom-archive. CENTCOM is the common abbreviation for the U.S. Central Command, which controls army operations in the Middle East, North Africa and Central Asia. PACOM is the name for U.S. Pacific Command, covering the rest of southern Asia, China and Australasia.

"For the research I downloaded 400GB of samples but there were many terabytes of data up there," he said. "It's mainly compressed text files that can expand out by a factor of ten so there's dozens and dozens of terabytes out there and that's a conservative estimate." Just one of the buckets contained 1.8 billion social media posts automatically fetched over the past eight years up to today. It mainly contains postings made in central Asia, however Vickery noted that some of the material is taken from comments made by American citizens. The databases also reveal some interesting clues as to what this information is being used for. Documents make reference to the fact that the archive was collected as part of the U.S. government's Outpost program, which is a social media monitoring and influencing campaign designed to target overseas youths and steer them away from terrorism.

An anonymous reader writes: One of the more surprising stories of the past year was Microsoft's announcement that it was going to use the Git version control system for Windows development. Microsoft had to modify Git to handle the demands of Windows development but said that it wanted to get these modifications accepted upstream and integrated into the standard Git client. That plan appears to be going well. Yesterday, the company announced that GitHub was adopting its modifications and that the two would be working together to bring suitable clients to macOS and Linux. Microsoft says that, so far, about half of its modifications have been accepted upstream, with upstream Git developers broadly approving of the approach the company has taken to improve the software's scaling. Redmond also says that it has been willing to make changes to its approach to satisfy the demands of upstream Git. The biggest complexity is that Git has a very conservative approach to compatibility, requiring that repositories remain compatible across versions.

Microsoft and GitHub are also working to bring similar capabilities to other platforms, with macOS coming first, and later Linux. The obvious way to do this on both systems is to use FUSE, an infrastructure for building file systems that run in user mode rather than kernel mode (desirable because user-mode development is easier and safer than kernel mode). However, the companies have discovered that FUSE isn't fast enough for this -- a lesson Dropbox also learned when developing a similar capability, Project Infinite. Currently, the companies believe that tapping into a macOS extensibility mechanism called Kauth (or KAuth) will be the best way forward.

From a wide-ranging interview of Richard Stallman, president of the Free Software Foundation, programming legend and recipient of at least 15 honorary doctorates and professorships: "The reason that we are subject now to more surveillance than there was in the Soviet Union is that digital technology made it possible," he says. "And the first disaster of digital technology was proprietary software that people would install and run on their own computers, and they wouldn't know what it was doing. They can't tell what it's doing. And that is the first injustice that I began fighting in 1983: proprietary software, software that is not free, that the users don't control." Here, Stallman is keen to stress, he doesn't mean free in the sense of not costing money -- plenty of free software is paid for -- but free in the sense of freedom to control. Software, after all, instructs your computer to perform actions, and when another company has written and locked down that software, you can't know exactly what it is doing. "You might think your computer is obeying you, when really its obeying the real master first, and it only obeys you when the real master says it's ok. With every program there are two possibilities: either the user controls the program or the program controls the users," he says. "It's free software if users control it. And that's why it respects their freedom. Otherwise it's a non-free, proprietary, user subjugating program."

Earlier this week, several publications published a holiday-themed data study about how families that voted for opposite parties spent less time together on Thanksgiving, especially in areas that saw heavy political advertising. The data came from a company called SafeGraph that supplied publications with 17 trillion location markets for 10 million smartphones. A report looks at the bigger picture: The data wasn't just staggering in sheer quantity. It also appears to be extremely granular. Researchers "used this data to identify individuals' home locations, which they defined as the places people were most often located between the hours of 1 and 4 a.m.," wrote The Washington Post. The researchers also looked at where people were between 1 p.m. and 5 p.m. on Thanksgiving Day in order to see if they spent that time at home or traveled, presumably to be with friends or family. "Even better, the cellphone data shows you exactly when those travelers arrived at a Thanksgiving location and when they left," the Post story says. To be clear: This means SafeGraph is looking at an individual device and tracking where its owner is going throughout their day. A common defense from companies that creepily collect massive amounts of data is that the data is only analyzed in aggregate; for example, Google's database BigQuery, which allows organizations to upload big data sets and then query them quickly, promises that all its public data sets are "fully anonymized" and "contain no personally-identifying information." In multiple press releases from SafeGraph's partners, the company's location data is referred to as "anonymized," but in this case they seem to be interpreting the concept of anonymity quite liberally given the specificity of the data.

A new paper from Stanford University reveals how artificial intelligence algorithms can be quickly trained to diagnose pneumonia better than a radiologist. "Using 100,000 x-ray images released by the National Institutes of Health on Sept. 27, the research published Nov. 14 (without peer review) on the website ArXiv claims its AI can detect pneumonia from x-rays with similar accuracy to four trained radiologists," reports Quartz. From the report: That's not all -- the AI was trained to analyze x-rays for 14 diseases NIH included in the dataset, including fibrosis, hernias, and cell masses. The AI's results for each of the 14 diseases had fewer false positives and false negatives than the benchmark research from the NIH team that was released with the data. The paper includes Google Brain founder Andrew Ng as a co-author, who also served as chief scientist at Baidu and recently founded Deeplearning.ai. He's often been publicly bullish on AI's use in healthcare. These algorithms will undoubtedly get better -- accuracy on the ImageNet challenge rose from 75% to 95% in just five years -- but this research shows the speed at which these systems are built is increasing as well.