YCLOAK

Television journalist Julia Kirienko was sheltering with Ukrainian soldiers and medics two miles (three kilometers) from the front when their cellphones began buzzing over the noise of the shelling. Everyone got the same text message at the same time.

“Ukrainian soldiers,” it warned, “they’ll find your bodies when the snow melts.”

Text messages like the one Kirienko received have been sent periodically to Ukrainian forces fighting pro-Russian separatists in the eastern part of the country. The threats and disinformation represent a new form of information warfare, the 21st-century equivalent of dropping leaflets on the battlefield.

“This is pinpoint propaganda,” said Nancy Snow, a professor of public diplomacy at the Kyoto University of Foreign Studies.

The Associated Press has found that the messages are almost certainly being sent through cell site simulators, surveillance tools long used by U.S. law enforcement to track suspects’ cellphones. Photos, video, leaked documents and other clues gathered by Ukrainian journalists suggest the equipment may have been supplied by the Kremlin.

The texts have been arriving since 2014, shortly after the fighting erupted. The AP documented nearly four dozen of them, including the one that Kirienko received on Jan. 31 in Avdiivka, a battle-scarred town outside the principal rebel-held city of Donetsk.

The messages typically say things such as “Leave and you will live” or “Nobody needs your kids to become orphans.” Many are disguised to look as if they are coming from fellow soldiers.

Fake towers

In 2015, Ukrainian soldiers defending the railroad town of Debaltseve were sent texts appearing to come from comrades claiming their unit’s commander had deserted. Another set of messages warned that Ukrainian forces were being decimated. “We should run away,” they said.

“They were mostly threatening and demoralizing, saying that our commanders had betrayed us and we were just cannon fodder,” said Roman Chashurin, who served as a tank gunner in Debaltseve.

Ukrainian military and intelligence services had no comment on the phenomenon, but government and telecommunications officials are well aware of what’s going on.

A 2014 investigation by a major Ukrainian cellphone company concluded that cell site simulators were to blame for the rogue messages, according to an information security specialist who worked on the inquiry. He spoke on the condition that neither he nor his former firm be identified, citing a nondisclosure agreement.

Col. Serhiy Demydiuk, the head of Ukraine’s national cyberpolice unit, said in an interview that the country’s intelligence services knew the devices were being used as well.

“Avdiivka showed that the Russian side was using fake towers,” he said. “They are using them constantly.”

Cell site simulators work by impersonating cellphone towers, allowing them to intercept or even fake data. Heath Hardman, a former U.S. Marines signals analyst who operated the devices in Iraq and Afghanistan, said they were routinely used by American military intelligence officers to hunt insurgents.

Sending mass text messages in wartime isn’t entirely new. The Islamic militant group Hamas sent threatening messages to random Israelis during the 2009 conflict over Gaza, for example, though it is not clear how that was done.

Effectiveness of texts

Cell site simulators significantly sharpen the ability of propagandists to tailor their messages to a specific place or situation, according to Snow, the academic.

“There’s just something about viewing a message on your phone that just makes people more susceptible or vulnerable to its impact,” she said.

The type of hardware involved remains a matter of speculation. But last year, the Ukrainian investigative website InformNapalm published a video and photographs appearing to show a LEER-3, a Russian truck-mounted electronic warfare system, in the Donetsk area. InformNapalm also disclosed what it described as leaked Russian military documents discussing the LEER-3’s deployment to the Luhansk area of eastern Ukraine.

A 2015 article in Russia’s Military Review magazine said the LEER-3 has a cell site simulator built into a drone that is capable of acting over a 6-kilometer-wide area and hijacking up to 2,000 cellphone connections at once. That makes it a “pretty plausible” source for the rogue texts in Ukraine, said Hardman, the former signals analyst.

Russia’s Defense Ministry did not return a request for comment. Moscow has long denied any direct role in the fighting in Ukraine, despite a wealth of evidence to the contrary.

The effectiveness of the propaganda texts is an open question. Soldiers say they typically shrug them off.

“I can’t say that it had any influence on us,” said Chashurin, the former tank gunner. “We were even joking that they must be so afraid of us the only thing they can do is to spam us with these texts.”

But Svetlana Andreychuk, a volunteer who has made frequent trips to the front line to distribute food and supplies, said the threats and mockery sometimes hit a nerve in a grinding conflict that has claimed more than 9,900 lives.

“Some people are psychologically influenced,” she said. “It’s coming regularly. People are so tired. You see people dying. And then you face this.”
…

In just a few years, well-mannered self-driving robotaxis will share the roads with reckless, law-breaking human drivers. The prospect is causing migraines for the people developing the robotaxis.

A self-driving car would be programmed to drive at the speed limit. Humans routinely exceed it by 10 to 15 mph (16 to 24 kph) — just try entering the New Jersey Turnpike at normal speed. Self-driving cars wouldn’t dare cross a double yellow line; humans do it all the time. And then there are those odd local traffic customs to which humans quickly adapt.

 

In Los Angeles and other places, for instance, there’s the “California Stop,” where drivers roll through stop signs if no traffic is crossing. In Southwestern Pennsylvania, courteous drivers practice the “Pittsburgh Left,” where it’s customary to let one oncoming car turn left in front of them when a traffic light turns green. The same thing happens in Boston. During rush hours near Ann Arbor, Michigan, drivers regularly cross a double-yellow line to queue up for a left-turn onto a freeway.

 

“There’s an endless list of these cases where we as humans know the context, we know when to bend the rules and when to break the rules,” said Raj Rajkumar, a computer engineering professor at Carnegie Mellon University who leads the school’s autonomous car research.

 

Although autonomous cars are likely to carry passengers or cargo in limited areas during the next three to five years, experts say it will take many years before robotaxis can coexist with human-piloted vehicles on most side streets, boulevards and freeways. That’s because programmers have to figure out human behavior and local traffic idiosyncrasies. And teaching a car to use that knowledge will require massive amounts of data and big computing power that is prohibitively expensive at the moment.

 

“Driverless cars are very rule-based, and they don’t understand social graces,” said Missy Cummings, director of Duke University’s Humans and Autonomy Lab.

 

Driving customs and road conditions are dramatically different across the globe, with narrow, congested lanes in European cities, and anarchy in Beijing’s giant traffic jams. In India’s capital, New Delhi, luxury cars share poorly marked and congested lanes with bicycles, scooters, trucks, and even an occasional cow or elephant.

 

Then there is the problem of aggressive humans who make dangerous moves such as cutting cars off on freeways or turning left in front of oncoming traffic. In India, for example, even when lanes are marked, drivers swing from lane to lane without hesitation.

 

Already there have been isolated cases of human drivers pulling into the path of cars such as Teslas, knowing they will stop because they’re equipped with automatic emergency braking.

 

“It’s hard to program in human stupidity or someone who really tries to game the technology,” says John Hanson, spokesman for Toyota’s autonomous car unit.

 

Kathy Winter, vice president of automated driving solutions for Intel, is optimistic that the cars will be able to see and think like humans before 2030.

 

Cars with sensors for driver-assist systems already are gathering data about road signs, lane lines and human driver behavior. Winter hopes auto and tech companies developing autonomous systems and cars will contribute this information to a giant database.

 

Artificial intelligence developed by Intel and other companies eventually could access the data and make quick decisions similar to humans, Winter says.

 

Programmers are optimistic that someday the cars will be able to handle even Beijing’s traffic. But the cost could be high, and it might be a decade or more before Chinese regulators deem self-driving cars reliable enough for widespread public use, said John Zeng of LMC Automotive Consulting.

 

Intel’s Winter expects fully autonomous cars to collect, process and analyze four terabytes of data in 1 { hours of driving, which is the average amount a person spends in a car each day. That’s equal to storing over 1.2 million photos or 2,000 hours of movies. Such computing power now costs over $100,000 per vehicle, Zeng said. But that cost could fall as more cars are built.

 

Someday autonomous cars will have common sense programmed in so they will cross a double-yellow line when warranted or to speed up and find a gap to enter a freeway. Carnegie Mellon has taught its cars to handle the “Pittsburgh Left” by waiting a full second or longer for an intersection to clear before proceeding at a green light. Sensors also track crossing traffic and can figure out if a driver is going to stop for a sign or red light. Eventually there will be vehicle-to-vehicle communication to avoid crashes.

 

Still, some skeptics say computerized cars will never be able to think exactly like humans.

 

“You’ll never be able to make up a person’s ability to perceive what’s the right move at the time, I don’t think,” said New Jersey State Police Sgt. Ed Long, who works in the traffic and public safety office.
…

China is testing the ability for future astronauts to stay on the moon for extended periods, as Beijing accelerates its space program and looks to put people on the surface of the moon within the next two decades.

The official Xinhua news agency said volunteers would live in a “simulated space cabin” for between 60-200 days over the next year helping scientists understand what will be needed for humans to “remain on the moon in the medium and long terms”.

Chinese President Xi Jinping has called for China to become a global power in space exploration, with plans to send a probe to the dark side of the moon by 2018, the first ever such trip, and to put astronauts on the moon by 2036.

“While it remains unclear exactly how long China’s first lunar explorers will spend on the surface, the country is already planning for longer stays,” Xinhua said.

Two groups of four volunteers will live in the simulated cabin “Yuegong-1” to test how a life-support system works in a moon-like environment. A similar 105-day trial was carried out successfully in 2014.

The system, called the Bioregenerative Life Support System (BLSS), allows water and food to be recycled and is key to any Chinese probes to the moon or beyond.

“The latest test is vital to the future of China’s moon and Mars missions and must be relied upon to guarantee the safety and health of our astronauts,” Liu Zhiheng of the Chinese Academy of Sciences told the news agency.

The Yuegong-1 cabin has a central living space the size of a “very small urban apartment” and two “greenhouses” for plants.

In March, China announced plans to launch a space probe to bring back samples from the moon this year, while the country’s first cargo space craft docked with an orbiting space lab in April, a major step as Beijing looks to establish a permanently manned space station by 2022.

Despite the advances in China’s space program for military, commercial and scientific purposes, China still lags behind the United States and Russia.
…

Electric carmaker Tesla has added another product to its lineup: Solar roof tiles.

As of Wednesday, customers worldwide could order a solar roof on Tesla’s website. Installations will begin next month in the U.S., starting with California. Installations outside the U.S. will begin next year, the company said.

The glass tiles were unveiled by Tesla last fall just before the company merged with solar panel maker SolarCity Corp. They’re designed to look like a traditional roof, with options that replicate slate or terracotta tiles. The solar tiles contain photovoltaic cells that are invisible from the street.

Guaranteed for life of home

Tesla CEO Elon Musk said one of the drawbacks to home solar installations has been the solar panels themselves: They’re often awkward, shiny and ugly. Buyers will want Tesla’s roof, he said, because it looks as good or better than a normal roof.

“When you have this installed on your house, you’ll have the best roof in the neighborhood. The aesthetics are that good,” Musk said in a conference call with media.

The roof is guaranteed for the life of the home, which is longer than the 20-year lifespan for a typical, nonsolar roof, Musk said. It has gone through the same hail, fire and wind testing that normal roofs endure.

Tesla’s website includes a calculator where potential buyers can estimate the cost of a solar roof based on the size of their home, the amount of sunlight their neighborhood receives and federal tax credits. They can also put down a refundable $1,000 deposit to reserve a place in line.

$42 per square foot

Tesla said the solar tiles cost $42 per square foot to install, making them far more costly than slate, which costs around $17 per square foot, or asphalt, which costs around $5. But homes would only need between 30 and 40 percent of their roof tiles to be solar; the rest would be Tesla’s cheaper nonsolar tiles, which would blend in with the solar ones.

It would cost $69,100 to install a solar roof with 40-percent solar tiles on a 2,600-square-foot roof in suburban Detroit, according to Tesla’s website. That includes a $7,000 Tesla Powerwall, a battery unit that stores the energy from the solar panels and powers the home. The roof would be eligible for a $15,500 federal tax credit and would generate an estimated $62,100 in electricity over 30 years. Over that time period, Tesla estimates, the homeowner would save $8,500.

Tesla said the typical homeowner can expect to pay $21.85 per square foot for a Tesla solar roof. The cost can be rolled into the homeowner’s mortgage payments and paid for over time, the company said.

Sales to be slow at first

Musk wouldn’t say how many orders the company expects to get this year. He expects the initial ramp-up to be slow.

“It will be very difficult and it will take a long time, and there will be some stumbles along the way. But it’s the only sensible vision of the future,” Musk said.

Palo Alto, California-based Tesla Inc. is making the solar tiles at its Fremont, California, factory initially. But eventually all production will move to a joint Tesla and Panasonic Corp. factory in Buffalo, New York. Panasonic makes the photovoltaic cells used in the solar tiles.

Tesla said it would be installing equipment in the Buffalo factory over the next few months.
…

Earlier this year, engineers at Wal-Mart Stores Inc. who track rivals’ prices online got a rude surprise: the technology they were using to check Amazon.com several million times a day suddenly stopped working.

Losing access to Amazon.com Inc.’s data was no small matter. Like most big retailers, Wal-Mart relies on computer programs that scan prices on competitors’ websites so it can adjust its listings accordingly. A difference of even 50 cents can mean losing a sale.

But a new tactic by Amazon to block these programs — known commonly as robots or bots — thwarted the Bentonville, Arkansas-based retailer.

Its technology unit, @WalmartLabs, was unable to work around the blockade for weeks, forcing it to retrieve Amazon’s data through a secondary source, according to a person familiar with the matter who was not authorized to speak publicly.

The previously unreported incident offers a case study in how Amazon’s technological prowess is helping it dominate the retail competition.

Now the largest online retailer in the world, Amazon is best known by consumers for its fast delivery, huge product catalog and ambitious moves into areas like original TV programming. But its mastery of the complex, behind-the-scenes technologies that power modern e-commerce is just as important to its success.

Dexterity with bots allows Amazon not only to see what its rivals are doing, but increasingly to keep them in the dark when it undercuts them on price or is quietly charging more.

“Benchmarking against Amazon is going to become hard,” said Guru Hariharan, a former Amazon manager who now sells pricing software to retailers as chief executive of Mountain View, California-based Boomerang Commerce.

A Wal-Mart spokesman declined to discuss the January episode but said the company improves its technology regularly and has multiple tools for tracking items. He said the company offers value not only through pricing but from discounts for in-store pickup and other benefits.

A spokeswoman for Amazon said the company is aware of competitors using bots to check its listings and denied any “campaign” to stop them. “Nothing has changed recently in how we manage bots on our site,” she said. Still, she said, “we prioritize humans over bots as needed.”

Bots can slow down a website, a big motivator for retailers to block them.

Reuters interviewed 21 people familiar with bots and how they are deployed, including current and former Wal-Mart employees, former Amazon employees and outside specialists. Many spoke only on condition of anonymity because they were not authorized to discuss the issues publicly.

Most pointed to Amazon’s leadership in the burgeoning bot wars.

The company’s technological edge has been good for its profit margin, and it’s proving a winning formula for investors.

Shares of the internet powerhouse have risen about 15-fold since the market’s bottom in March 2009, while the S&P 500 has more than tripled in value. Amazon hit $100 billion in annual sales in 2015 — faster than any company in history, it said.

Brave new world

Bot-driven pricing has represented a massive change for the retail industry since Amazon helped pioneer the practice more than a decade ago.

Traditionally, brick-and-mortar stores changed prices no more than weekly because of the time and expense needed to swap labels by hand.

In the world of e-commerce, though, retailers update prices with ease, sometimes multiple times a day, helped by algorithms that consider inventory levels, sales forecasts and rivals’ pricing data.

To stay in the game, companies such as online wholesaler Boxed, based in New York, depend on a variety of methods including bots to ensure they do not lag others’ price moves for even 20 minutes.

“That’s like a lifetime during Christmas,” said Chief Executive Chieh Huang, whose company sells bulk staples like toilet paper and pet food. “If we’re not decently priced, we’ll see it almost immediately” in sales declines.

  

Disguised as humans

Using bots to view massive amounts of data on public websites — a process known as crawling or scraping — has many purposes. Alphabet Inc.’s Google, for example, constantly crawls the Web to gather information for its search engine results and to sell ads.

In e-commerce, though, the use of bots has developed into a cat-and-mouse game. Companies try to thwart the practice on their own websites while aiming to penetrate their competitors’ defenses. Third-party services abound to help less-savvy retailers.

To protect data from rivals, some retail websites use what’s known as a “CAPTCHA” — typically a distorted string of letters and numbers that humans can read but most bots can’t. Amazon shies away from the practice because it annoys some customers.

For merchants seeking to evade such defenses, disguising their computer programs as real shoppers is key. Some pricing technology experts have programmed computer cursors to meander through a Web page in the way a person might, instead of going directly to the prized data. Another technique is to use multiple computer addresses so that retailers cannot track a barrage of clicks to a single source.

“It is an arms race,” said Keith Anderson, a senior vice president at e-commerce analytics firm Profitero, based in Ireland. “Every week or every month, there’s some new approach from both sides.”

Amazon’s maneuver that halted Wal-Mart in January took aim at a specialized Web browser called PhantomJS. Unlike, say, Internet Explorer, this browser is designed specifically for programmers — a telltale clue that its users are not typical shoppers. Amazon put up a digital curtain to hide its listings from PhantomJS users, according to three people familiar with the situation.

It was unclear how the move, which was not aimed at Wal-Mart in particular, affected other companies.

Tests conducted in recent weeks for Reuters show that among major U.S. retail chains, Amazon had by far the most sophisticated bot detection in place, both for its home page and for two popular items selected by Reuters because they change price frequently — a De’Longhi coffee maker and a Logitech webcam.

The tests were run by San Francisco-based Distil Networks, which sells anti-bot tools. In one of the tests, Distil programmed bots to hit each retailer’s website 3,000 times, but slowly enough to mimic a person clicking through listings. This tricked most retail behemoths, but not Amazon.

Blocked bots would not have seen, for instance, that Amazon’s price for the De’Longhi espresso machine changed four times in a single 24-hour period starting on the morning of April 25, according to price tracking website camelcamelcamel.com. During that time, the price swung by more than 10 percent, from a low of $80.06 to $88.16.

Swarming with bots

Despite Amazon’s capabilities, the sheer volume of crawling on its site is staggering. At times, as many as 80 percent of the clicks on Amazon product listings have been from bots, people familiar with the matter say, compared with just a third or more of the traffic on other large sites.

In addition to rivals seeking price data, that traffic includes bots from university researchers studying competition, search engines, advertising services and even fraudsters trying to break into Amazon accounts.

For Wal-Mart, a small group in Silicon Valley directs its automated pricing strategy while dozens of engineers in India and around the world handle the code, current and former Wal-Mart employees said.

Amazon had about 40 engineers who would covertly extract and organize rivals’ data with bots as of several years ago, one of the people interviewed said. Amazon did not discuss the size or structure of its teams working with bots.

According to one U.S. patent application, Amazon is working on encryption technology that would force bots, but not humans, to solve a complicated algorithm to gain access to its Web pages.

“Amazon has both the competency to detect bot traffic and the wherewithal to do something about it,” said Scott Jacobson, a former Amazon manager and now managing director of Madrona Venture Group. That “isn’t the case for most retailers.”

 
…

Microsoft Corp. on Wednesday turned up the heat on other technology giants by launching new image and video recognition products which could help it court businesses worried about running ads next to offensive content.

The Redmond, Washington-based company said its new Video Indexer can identify faces, voices and emotions in moving pictures. Separately, its Custom Vision Search lets companies build apps that recognize images with just a few lines of code.

For brands, knowing what’s in the videos that they sponsor has become a hot-button issue since major companies began canceling ad deals with Alphabet Inc’s Google this year over hate speech playing on its subsidiary YouTube.

Microsoft’s Video Indexer has similarities to a tool Google launched in March; Amazon.com Inc also said last month it could flag insulting images via a cloud-based service.

Microsoft’s latest moves underscore how its focus has evolved from its staple Windows software to the cloud, where it is competing with Amazon to sell data storage and computing power. Extra analytics such as image recognition may prove key to luring Web developers.

“It’s hard to understand what’s in the video” the longer it is, said Irving Kwong, a senior product director at Microsoft, in an interview ahead of the company’s developer conference Build. He said Video Indexer, which analyzes videos far faster than humans can, could help a user “harness and get more out of the video content that you have.”

The tools launched in preview by the Microsoft Cognitive Services unit on Wednesday, including a decision recommendation service, have one aim apart from winning business: data.

Microsoft views the tools as a way to put powerful computing into people’s hands and improve the tools at the same time, because processing more data is key to reaching artificial intelligence. Others including Amazon are pursuing this strategy, with the prize being a new revenue stream.

Research firm International Data Corporation has forecast the market for such tools will balloon to over $47 billion in sales in 2020 from $8 billion in 2016.

Microsoft pulled back the curtain on experiments that are further afield, too. It announced a new Cognitive Labs unit and the so-called Project Prague: technology to allow people to control computers simply with hand gestures.
…

Facebook is planning to intensify its crackdown on so-called clickbait websites, saying it will begin giving lower prominence to links that lead to pages full of deceptive or annoying advertisements.

The downgrade of the links was expected to take effect beginning on Wednesday on News Feed, the home page of Facebook where people go to see posts from friends and family.

Facebook said it wanted to downplay links that people post to websites that have a disproportionate volume of ads relative to content, or that have deceptive or sexually suggestive ads along the lines of “5 Tips to be Amazing in Bed” or “1 Crazy Tip to Lose Weight Overnight!”

Links to websites with pop-up ads or full-screen ads also would be downplayed, it said.

People scrolling through their News Feed are often disappointed when they click on such links and do not find valuable information, Andrew Bosworth, Facebook’s vice president of ads and business platform, said in an interview.

“People don’t want to see this stuff,” he said. “We’re just trying to figure out how to find it and rank it further down News Feed when possible.”

Facebook uses a computer algorithm to determine which posts people see first from friends and family, and it frequently refines the algorithm to keep up with spam or other concerns.

The company said in August it was adjusting the algorithm to downplay news stories with clickbait-style headlines, a style of headline that intentionally withholds information or misleads people to get them to click on them.

In December, facing criticism that hoaxes and fake news stories spread too easily on Facebook in the run-up to the U.S. presidential election on November 8, the company made it easier for people to report those kinds of posts.

Facebook, the world’s largest social media network with 1.9 billion monthly users, has enormous power with its algorithms to potentially drive traffic to media publishers or stymie it.

The company said it reviewed hundreds of thousands of websites linked to from Facebook to identify those with little substance but lots of disruptive or shocking ads.

Bosworth declined to name any websites Facebook wants to target. He said only publishers of spam needed to worry about seeing less traffic, and other publishers could see their traffic go up.

“This is a small number of the worst of the worst,” he said.
…

New York University researchers have developed a system combining virtual reality with a pressure sensing mat they say could help people with vestibular dysfunction, which affects parts of the inner ear and brain and results in problems with balance, or those suffering from vertigo or dizziness as a result of a brain injury. Faith Lapidus reports.
…