Afraid so: the machines are now able to beat us at this game too. CC-licensed photo by Chris on Flickr.
Ahead of No. 1,000, send in your three favourite links – leave a comment, email or DM me. Popular so far: Why drowning doesn’t look like drowning (May 2018); why I hope we don’t find extraterrestrial life (Aug 2016); the heroes of the cave dive rescue (Jan 2019). What do you remember best?
You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.
A selection of 9 links for you. Borderless. I’m @charlesarthur on Twitter. Observations and links welcome.
We’re now taking steps that wouldn’t have been possible even just a few years ago – for example, this year we plan to spend more on safety and security than our whole revenue at the time of our IPO, and the artificial intelligence required to help manage content at scale didn’t exist until recently. But as people use these networks to shape society, it’s critical we continue making progress on these questions.
At the same time, there is another force at play as well. As networks of people replace traditional hierarchies and reshape many institutions in our society – from government to business to media to communities and more – there is a tendency of some people to lament this change, to overly emphasize the negative, and in some cases to go so far as saying the shift to empowering people in the ways the internet and these networks do is mostly harmful to society and democracy.
To the contrary, while any rapid social change creates uncertainty, I believe what we’re seeing is people having more power, and a long term trend reshaping society to be more open and accountable over time. We’re still in the early stages of this transformation and in many ways it is just getting started. But if the last 15 years were about people building these new networks and starting to see their impact, then the next 15 years will be about people using their power to remake society in ways that have the potential to be profoundly positive for decades to come.
I have a question: are those people going to be unaccountable Russians working in low-rise buildings and pretending to be African-Americans based in Chicago protesting against police brutality in order to stir up division? Just so we’re clear on meaning, you understand.
link to this extract
VPNs are supposed to help you protect your data. But the Facebook flap shows that there’s one party that has full access to everything you’re doing: the VPN provider itself. And it’s a concern with several Chinese-owned VPNs, which reportedly send data back to China.
Recently, Top10VPN – a review site for VPN services – looked into the world’s 30 most downloaded free VPN apps. Among them, VPN Master, Turbo VPN, and Snap VPN claim the right to gather private information like IP addresses, time zones, and IMEIs (the unique number that identifies your phone). They also state that they may route personal data to China.
Another Chinese-owned app, VPN 360, notes that they may log and share an individual’s usage data with government authorities and law enforcement when required by law.
Unlike Facebook’s semi-secret “market research” app, these VPN services are readily available for everyone to download from Google Play and the iOS App Store.
And it means that while Facebook has said it will shut down its controversial market research app, other questionable VPN services are still being downloaded every day, with little transparency on where the data they collect will go.
Why, it’s as if VPNs aren’t a panacea to put you on the golden path to privacy at all, but instead might just mine your data. I guess you could ask the malicious hackers who have been busted via their VPN activity what they think for a second opinion.
link to this extract
Bitcoin’s ‘Five-Year Bet’ now has an official winner between Ben Horowitz and Felix Salmon • Bitcoinexchangeguide
Horowitz and Salmon both agreed to meet after five years and assess the state of the market during their 2014 appearance on NPR’s podcast. They also agreed on an official bet, the conditions for which are as follows:
“If 10% of Americans or more said they’d bought something with Bitcoin in the past month, Ben would win. If the number was lower, Felix would win.”
Before declaring the winner of the bet on episode #891, Planet Money published a poll on their website in an effort to gauge whether or not BTC (or any other major altcoin) had really been able to break into the financial mainstream.
Not only that, even Ipsos recently conducted a poll that took into consideration the opinions of 900 Americans who were asked the simple question:
“Have you purchased anything using Bitcoin as your payment within the past month?” In response, only a meager 3% of the respondents replied in the affirmative.
The transcript of the podcast makes for fun reading. That 3% is definitely a ceiling; in quite a few of the cases, people saying yes were using it to but other cryptocoins, and in some cases they claimed to have used it in places which don’t exchange bitcoin.
Salmon’s prediction – that bitcoin’s deflationary tendencies (forcing its price up) would kill its use as currency – turned out to be correct.
link to this extract
Recently there have been efforts to develop game-like CAPTCHAs, tests that require users to rotate objects to certain angles or move puzzle pieces into position, with instructions given not in text but in symbols or implied by the context of the game board. The hope is that humans would understand the puzzle’s logic but computers, lacking clear instructions, would be stumped. Other researchers have tried to exploit the fact that humans have bodies, using device cameras or augmented reality for interactive proof of humanity.
The problem with many of these tests isn’t necessarily that bots are too clever — it’s that humans suck at them. And it’s not that humans are dumb; it’s that humans are wildly diverse in language, culture, and experience. Once you get rid of all that stuff to make a test that any human can pass, without prior training or much thought, you’re left with brute tasks like image processing, exactly the thing a tailor-made AI is going to be good at.
“The tests are limited by human capabilities,” Polakis says. “It’s not only our physical capabilities, you need something that [can] cross cultural, cross language. You need some type of challenge that works with someone from Greece, someone from Chicago, someone from South Africa, Iran, and Australia at the same time. And it has to be independent from cultural intricacies and differences. You need something that’s easy for an average human, it shouldn’t be bound to a specific subgroup of people, and it should be hard for computers at the same time. That’s very limiting in what you can actually do. And it has to be something that a human can do fast, and isn’t too annoying.”
Figuring out how to fix those blurry image quizzes quickly takes you into philosophical territory: what is the universal human quality that can be demonstrated to a machine, but that no machine can mimic? What is it to be human?
Really it comes down to our tendency to dither when we don’t know. Or else be too certain when we don’t know. Unfortunately, machines can copy that too.
link to this extract
The shapes of the plots can tells us more about the nature of attention to the topic. The duration of news events is dependent on the speed at which an event develops, and whether or not its outcome was expected. The North Korea summit, for example, was in the news in the lead up to the event, and continued to be reported on afterwards, producing a symmetrical interest plot. An event like July’s blood moon, by contrast, was rarely mentioned after the fact, resulting in a leftward skew in the plot. An unexpected event, on the other hand, like the death of Anthony Bourdain, can yield a rightward skew in the plot as the public continues to process unanticipated information. Lastly, some events can even produce multiple peaks, like the government shutdown of January 2018 that was followed by the threat of a second shutdown in early February, resulting in a bimodal search interest plot.
As a journalist, the question one always wanted to be able to answer was: how important will this story appear to readers? Is this just a short hit or does it tug at something deeper? After a while you’d get a feel for that, but could still be surprised by things. This tries to offer a clearer view. (Via @Sophiewarnes’s Fair Warning newsletter.)
link to this extract
Like all inventors, Khan was paranoid about knockoffs. Even so, he was caught by surprise when Huawei, a potential customer, began to behave suspiciously after receiving the meticulously packed sample [of a screen coated on one side with artificial diamond]. Khan was more surprised when the US Federal Bureau of Investigation drafted him and Akhan’s chief operations officer, Carl Shurboff, as participants in its investigation of Huawei. The FBI asked them to travel to Las Vegas and conduct a meeting with Huawei representatives at last month’s Consumer Electronics Show. Shurboff was outfitted with surveillance devices and recorded the conversation while a Bloomberg Businessweek reporter watched from safe distance.
This investigation, which hasn’t previously been made public, is separate from the recently announced grand jury indictments against Huawei. On Jan. 28, federal prosecutors in Brooklyn charged the company and its chief financial officer, Meng Wanzhou, with multiple counts of fraud and conspiracy. In a separate case, prosecutors in Seattle charged Huawei with theft of trade secrets, conspiracy, and obstruction of justice, claiming that one of its employees stole a part from a robot, known as Tappy, at a T-Mobile US Inc. facility in Bellevue, Wash. “These charges lay bare Huawei’s alleged blatant disregard for the laws of our country and standard global business practices,” Christopher Wray, the FBI director, said in a press release accompanying the Jan. 28 indictments. “Today should serve as a warning that we will not tolerate businesses that violate our laws, obstruct justice, or jeopardize national and economic well-being.” Huawei has denied the charges…
The first sign of trouble came two months later, in May, when Huawei missed the deadline to return the sample. Shurboff says his emails to Han requesting its immediate return were ignored. The following month, Han wrote that Huawei had been performing “standard” tests on the sample and included a photo showing a big scratch on its surface. Finally, a package from Huawei showed up at Gurnee on Aug. 2.
Shurboff remembers opening it. It looked just like the package Akhan had sent months earlier. Inside the cardboard box was the usual protective packaging—air bags, plastic case, gel insert, and wax paper. But he could tell something was wrong when he picked up the case. It rattled. The unscratchable Miraj sample wasn’t just scratched; it was broken in two, and three shards of diamond glass were missing.
Shurboff says he knew there was no way the sample could have been damaged in shipping—all the pieces would still be there in the case.
I don’t really want a flying car, but I do want to be able to shed pennies (and fractions of pennies) as I browse news or read fiction online. I want to easily support artists and writers without having to set up an account, create a password, fork over my credit card details, and commit to an ongoing relationship that involves receiving a new piece of spammish email at least once a week.
What would such a system look like? It would be as seamless as browsing itself. It could have an automatic mode (a news subscription consortium, for instance, could silently disperse payments to individual publications as I read articles from members) or a one-click mode. (Stumble across a nice poem on some unfamiliar site? A small green button on your browser lights up, and you can make a one-time contribution.) And, much as Apple Pay already does now, vendors wouldn’t necessarily get your account information, just a cryptographic payment token that’s good for exchange or verification.
Of course, we already make payments online all the time, but under current conditions, frankly, it sucks to do so. If you buy things directly from small vendors, you’re stuck entering your credit card information, your email, and your billing address on site after site—sinking ever deeper into the surveillance economy as each digital form puts your personal details into someone else’s database, while also giving hackers ever more opportunities to filch your data.
Zeynep is usually reliably wonderful, but this is one area where she’s got a blindness to the subtle combination of economics and internet behaviour that would result. I used to have a running bet with Jakob Nielsen: he said we would soon have micropayments, I said we wouldn’t. We gave up after I’d been correct four, or possibly five, years in a row; that was about 2003.
I wrote about why this won’t ever happen a week short of ten years ago. Not a single thing about the dynamic has changed since, despite the invention of bitcoin. Micropayments have too many perverse incentives to ever happen.
link to this extract
Robots have already mastered games like chess and Go. Now they’re coming for Jenga • The Washington Post
AI long ago mastered chess, the Chinese board game Go and even the Rubik’s cube, which it managed to solve in just 0.38 seconds.
Now machines have a new game that will allow them to humiliate humans: Jenga, the popular game —— and source of melodramatic 1980s commercials —— in which players strategically remove pieces from an increasingly unstable tower of 54 blocks, placing each one on top until the entire structure collapses.
A newly released video from MIT shows a robot developed by the school’s engineers playing the game with surprising precision. The machine is quipped with a soft-pronged gripper, a force-sensing wrist cuff and an external camera, allowing the robot to perceive the tower’s vulnerabilities the way a human might, according to Alberto Rodriguez, the Walter Henry Gale career development assistant professor in the Department of Mechanical Engineering at MIT.
“Unlike in more purely cognitive tasks or games such as chess or Go, playing the game of Jenga also requires mastery of physical skills such as probing, pushing, pulling, placing, and aligning pieces,” Rodriguez said in a statement released by the school. “It requires interactive perception and manipulation, where you have to go and touch the tower to learn how and when to move blocks.”
“This is very difficult to simulate, so the robot has to learn in the real world, by interacting with the real Jenga tower,” he added.
These things are really ruining party games.
link to this extract
UK emissions have declined from around 600m tonnes of CO2 (MtCO2) in 1990 to 367MtCO2 in 2017. If underlying factors driving emissions had not changed, Carbon Brief’s analysis shows that a growing population and a constant electricity generation mix would have led to emissions increasing by around 25% compared to 1990 levels.
Instead, emissions actually fell by 38% to 367MtCO2, as shown in the black area in the figure below. Each coloured wedge in the figure shows one factor contributing to this decline.
As the chart shows, no single factor was responsible for more than around a third of the total reduction in the UK’s CO2. Overall, emissions in 2017 were 51% lower than they would have been without these changes.
The surprising data point: UK CO2 emissions peaked in 1973, because we were burning so much coal.
link to this extract
Errata, corrigenda and ai no corrida: none notified