Start Up: numberplate readers online, PC sales fall again, celebrity death hoaxing, and more

Afraid so – machine learning is coming for your poker game. Photo by Jeremy Brooks on Flickr.

A selection of 14 links for you. It’s a lot, isn’t it? I’m @charlesarthur on Twitter. Observations and links welcome.

The Internet IS a Thing: use it to find license plate readers • networked inference

Kenneth Lipp:


A few years ago I was searching for info about Automatic License Plate Readers, ALPR, Googling by manufacturer “Genetec.” I found a “Read Me” file from an FTP server, that it turned out was a live ALPR server for the City of Boston, with classified watchlists including thousands on a “Gang/Terror” watch.

At the time I was just searching Google, but once I found something interesting I used a variety of other tools to remotely scan the server and its directories to see what I could learn.

I found out recently that you could search Shodan, the Search Engine for the Internet of Things, for text in the “html” field. Shodan indexes, among other data about the physical architecture of the Web, headers which provide response codes (like 100, 200, “404 Error,” etc), and often descriptive information about the function of the computer (or machine, or site, for the purposes of Shodan’s heuristic, host).

In Boston I located a Genetec AutoVu server. I searched Shodan for html:”autovu”

This brings back four results, once filtered to just the US (probably two ALPR systems, 3 IP addresses, one is repeated in the results)…

…What we have in this case appears to be a field test for an occupancy-detection/plate-reader hybrid solution, located in Boydton, in Virginia, all the way down on the North Carolina border — where would you look at that there’s a Microsoft Data Center.

I know off the top of my head that Massachusetts state was working with Xerox on such a system on state roads, which in addition to scanning and archiving plate numbers of passing cars used a computer vision algorithm to determine how many people are in the car (or perhaps whether the car has only one or more than occupant — this type of system is frequently deployed for monitoring occupancy-restricted lanes — “carpool lanes.”)


And this stuff is on the internet, at least in the US. Lipp has written much more on this in a series of blogposts.
link to this extract

Facebook risks breaking its perfect business model • Bloomberg Gadfly

Shira Ovide on Facebook’s decision to start revenue-sharing on pre-roll and mid-roll (yuk) ads in videos:


Facebook has become – surprisingly – the perfect business for the smartphone age, and a big reason is it has spent essentially nothing to keep users enthralled. For the most part, companies that publish political articles or cooking videos on Facebook don’t make money directly from that material, although they use those items to assemble a big fan base and then point those people to websites and apps where the companies make money selling ads or subscriptions.

Those articles and cooking videos keep users hanging out on Facebook, and the company keeps all the money it makes from selling advertisements that fill in gaps between those posts and videos they paid nothing to publish. It may not be fair, but it has made for a wildly successful and profitable business. 

If Facebook is now willing to give 55% of ad dollars from those video ads, that means cracks are emerging in Facebook’s free ride with its army of content suppliers. (Facebook also has experimented with splitting ad dollars with semiprofessional video stars who have attracted television-sized audiences on YouTube.)

Sharing money is more equitable but could damage Facebook’s finances. Consider Alphabet Inc.’s YouTube. The video website makes roughly one-third of the money Facebook generates from each user. It’s not clear exactly why. Facebook may be doing a better job stuffing ads into every spot it can. Surely part of the gap is explained by Facebook paying almost nothing to stock the social network with posts, photos and video, while YouTube hands off 55 cents of every dollar it generates to the creators of popular videos.


link to this extract

How surveillance changes people’s behavior • Harvard Magazine

Jonathan Shaw, in a very long article on this topic:


“Google was ground zero,” [Wilson professor of business administration Shoshana] Zuboff begins. At first, information was used to benefit end users, to improve searches, just as Apple and Amazon use their customers’ data largely to customize those individuals’ online experiences. Google’s founders once said they weren’t interested in advertising. But Google “didn’t have a product to sell,” she explains, and as the 2001 bubble fell into crisis, the company was under pressure to transform investment into earnings. “They didn’t start by saying, ‘Well, we can make a lot of money assaulting privacy,’” she continues. Instead, “trial and error and experimentation and adapting their capabilities in new directions” led them to sell ads based on personal information about users. Like the tinkerers at Ford, Google engineers discovered “a way of using their capabilities in the context of search to do something utterly different from anything they had imagined when they started out.” Instead of using the personal data to benefit the sources of that information, they commodified it, using what they knew about people to match them with paying advertisers. As the advertising money flowed into Google, it became a “powerful feedback loop of almost instantaneous success in these new markets.”

“Those feedback loops become drivers themselves,” Zuboff explains. “This is how the logic of accumulation develops…and ultimately flourishes and becomes institutionalized. That it has costs, and that the costs fall on society, on individuals, on the values and principles of the liberal order for which human beings have struggled and sacrificed much over millennia—that,” she says pointedly, “is off the balance sheet.”


link to this extract

How celebrity death hoaxes power fake news • Digiday

Lucia Moses:


Web traffic-goosing tricks come and go. But if there’s one that has enduring appeal, it’s the celebrity death hoax.

In the past few weeks alone, internet pranksters have “killed off” Queen Elizabeth, Tony Hawk, Miley Cyrus and Hugh Hefner, to name a handful that have been debunked by the website Gossip Cop. Some trace its peak to the site Global Associated News, a fake-news headline generator that web entrepreneur Rich Hoover said he started as a joke.

Since then, others have discovered the celebrity death hoax as a tried-and-true scheme to drive traffic to their sites, which they’re monetizing with ads. The fake stories follow a loose pattern: Often coming from sites with legit-sounding names like and, the stories tend to focus on young, popular celebrities with many fans who would be shocked by their premature death, causing a burst of traffic to the site, which is paid for with programmatically served ads.

Michael Lewittes, founder of Gossip Cop, said he once was seeing as many as two or three death hoaxes a week. “Eddie Murphy and Adam Sandler have died, probably collectively, 15 to 20 times on the internet,” he said.

The rise of platforms and ad tech have enabled the spread of celebrity death, just as it has other types of fake news. Some pranksters have used Twitter to fool people about celebrity deaths, using accounts that sound like actual news outlets.


Celebrity *anything* tends to drive clicks. Hell, you can even get people to *vote* for you with the power of celebrity, even if you’re talentless.
link to this extract

US appeals court revives antitrust lawsuit against Apple • Reuters

Stephen Nellis and Dan Levine:


iPhone app purchasers may sue Apple Inc over allegations that the company monopolized the market for iPhone apps by not allowing users to purchase them outside the App Store, leading to higher prices, a U.S. appeals court ruled on Thursday.

The 9th U.S. Circuit Court of Appeals ruling revives a long-simmering legal challenge originally filed in 2012 taking aim at Apple’s practice of only allowing iPhones to run apps purchased from its own App Store. A group of iPhone users sued saying the Cupertino, California, company’s practice was anticompetitive.

Apple had argued that users did not have standing to sue it because they purchased apps from developers, with Apple simply renting out space to those developers. Developers pay a cut of their revenues to Apple in exchange for the right to sell in the App Store.

A lower court sided with Apple, but Judge William A. Fletcher ruled that iPhone users purchase apps directly from Apple, which gives iPhone users the right to bring a legal challenge against Apple.


That could get interesting.
link to this extract

The eye of the storm: a look at EyePyramid, the malware supposedly used in high-profile hacks in Italy • TrendLabs Security Intelligence Blog

Federico Maggi:


Two Italian citizens were arrested last Tuesday by Italian authorities (in cooperation with the FBI) for exfiltrating sensitive data from high-profile Italian targets. Private and public Italian citizens, including those holding key positions in the state, were the subject of a spear-phishing campaign that reportedly served a malware, codenamed EyePyramid, as a malicious attachment. This malware was used to successfully exfiltrate over 87 gigabytes worth of data including usernames, passwords, browsing data, and filesystem content.


Caught how? One of them bought a licence key for MailBee under his own name, and used the key in the attack code. Operational security is tough.
link to this extract

Asking the wrong questions • Benedict Evans

Evans looks at forecasts of the future from the 1960s:


Some of this has happened more or less as predicted – we did get air traffic control, automated subway trains and computerised taxation (except in the USA). There are some great comedy predictions here too – that ‘centralised wire tapping’ would take until 2030, or never, or that people in both 1964 and 2016 thought we’d have automated driving ‘by 2020’. 

However, to me the interesting thing is how often the order is wrong. What we now know to be the hard problems were going to be solved decades before what we now know were the easy ones. So it might take until 2020 to ‘fax’ a newspaper to your home, and automatic wiretapping might be impossible, but automatic doctors, radar implants for the blind, household robots and machine translation would be all done by 1990 and a machine would be passing human IQ tests at genius level by 2000. Meanwhile, there are a few quite important things missing – there is no general-purpose computing, no internet and no mobile phones. There’s no prediction for when everyone on earth would have a pocket computer connected to all the world’s knowledge (2020-2025). This aren’t random gaps – it’s just not that they thought X would work and didn’t know we’d invent Y. Rather, what’s lacking is an understanding of the structural impetus of computing and software as universal platforms that would shape how all of these things would be created. We didn’t make a home newspaper facsimile machine – we made computers.


link to this extract

Damn, Apple is losing a lot of people • Gizmodo

Christina Warren:


Here is a list of some of the high level employees who have left Apple since January 2016 and where they have gone, if that information is available:


It’s a long list (24 names), from multiple departments – though five are PR or media (is that a crucial department?). Notable that many in that list are going to Tesla. Either Tesla is the attractive place, or it’s good at the PR job of announcing recruitment wins.
link to this extract

Xiaomi stops disclosing annual sales figures as CEO admits the company grew too fast • TechCrunch

Jon Russell:


Xiaomi has forgone its tradition of revealing how many smartphones it sold the previous year. The strategy yielded many headlines for the highly-regarded Chinese outfit, but today its CEO admitted that Xiaomi has been in transition after growing “too fast”.

The writing was on the cards, even as early as January 2016 when Xiaomi revealed it had sold “over 70 million” devices in 2015. An impressive number, for sure, given the backdrop of slowing smartphone sales worldwide, but it was short of the company’s public target of 80m, which was reduced from an initial 100m.

It’s been fairly evident from analyst reports that 2016 wasn’t a year for blockbuster Xiaomi growth. While it featured near the top of the sales pile in China, and held steady in India, its top emerging market, there was no great acceleration as in past years. For example, sales jumped from 7.2m in 2012, to 18.7m in 2013 and 61m in 2014.

A Xiaomi rep confirmed to us that the company will not be disclosing its 2016 sales numbers.


That $45bn valuation (at its last funding round in December 2014) doesn’t look so solid any more.
link to this extract

Poker is the latest game to fold against artificial intelligence • MIT Technology Review

Will Knight:


In a landmark achievement for artificial intelligence, a poker bot developed by researchers in Canada and the Czech Republic has defeated several professional players in one-on-one games of no-limit Texas hold’em poker.

Perhaps most interestingly, the academics behind the work say their program overcame its human opponents by using an approximation approach that they compare to “gut feeling.”

“If correct, this is indeed a significant advance in game-playing AI,” says Michael Wellman, a professor at the University of Michigan who specializes in game theory and AI. “First, it achieves a major milestone (beating poker professionals) in a game of prominent interest. Second, it brings together several novel ideas, which together support an exciting approach for imperfect-information games.”

Later this week, a tournament at a Pittsburgh casino will see several world-class poker players play the same version of poker against a program developed at Carnegie Mellon University.


Computers playing poker get an advantage over humans: they don’t have emotions, and they don’t have “tells” – the unconscious signals we give off to indicate tension or otherwise.

However, poker is amazingly hard for computers:


tellingly, it contains levels of uncertainty, such as when an opponent may be bluffing, that are found in many real-world situations that AI has not yet mastered. Poker players cannot see their opponents’ hands, meaning that, in contrast to checkers, chess, or Go, not all of the information contained within the game is available to them.


About 10^160 paths for each hand in heads-up no-limit Texas Hold’em.
link to this extract

2016 marked fifth consecutive year of worldwide PC shipment decline • Gartner


For the year 2016, PC shipments totaled 269.7m units, a 6.2% decline from 2015. PC shipments have declined annually since 2012.

“Stagnation in the PC market continued into the fourth quarter of 2016 as holiday sales were generally weak due to the fundamental change in PC buying behavior,” said Mikako Kitagawa, principal analyst at Gartner. “The broad PC market has been static as technology improvements have not been sufficient to drive real market growth. There have been innovative form factors like 2-in-1s and thin and light notebooks, as well as technology improvements, such as longer battery life. This end of the market has grown fast, led by engaged PC users who put high priority on PCs. However, the market driven by PC enthusiasts is not big enough to drive overall market growth.”

“There is the other side of the PC market, where PCs are infrequently used. Consumers in this segment have high dependency on smartphones, so they stretch PC life cycles longer. This side of the market is much bigger than the PC enthusiast segment; thus, steep declines in the infrequent PC user market offset the fast growth of the PC enthusiast market.”


Hasn’t found the bottom yet, then. The other notable point is that the “Others” category (all those who aren’t Lenovo, HP, Dell, Asus, Apple or Acer) shrank by 19% in the fourth quarter, against 17% for the year – suggesting that the squeeze on the smaller players is getting worse. In total, “Others” went from 77.6m in 2015 to 64.5m. Those lost 13m sales are going to hurt balance sheets.
link to this extract

Saving you bandwidth on Google+ through machine learning • Google product blog

John Nack, Google+ product manager:


Traditionally, viewing images at high resolution has also meant using lots of bandwidth, leading to slower loading speeds and higher data costs. For many folks, especially those where data is pricey or the internet is spotty, this is a significant concern.

To help everyone be able to see the beautiful photos that photographers share to Google+ in their full glory, we’ve turned to machine learning and a new technology called RAISR. RAISR, which was introduced in November, uses machine learning to produce great quality versions of low-resolution images, allowing you to see beautiful photos as the photographers intended them to be seen. By using RAISR to display some of the large images on Google+, we’ve been able to use up to 75% less bandwidth per image we’ve applied it to.


“Google+ product manager” must be one of those depressing jobs. This is a great application, though on my slow connection the picture alone took absolutely ages – over 30 seconds – to load.
link to this extract

Techdirt’s First Amendment fight for its life • Techdirt

Mike Masnick, Techdirt’s founder:


As you may have heard, last week we were sued for $15m by Shiva Ayyadurai, who claims to have invented email. We have written, at great length, about his claims and our opinion — backed up by detailed and thorough evidence — that email existed long before Ayyadurai created any software. We believe the legal claims in the lawsuit are meritless, and we intend to fight them and to win.

There is a larger point here. Defamation claims like this can force independent media companies to capitulate and shut down due to mounting legal costs. Ayyadurai’s attorney, Charles Harder, has already shown that this model can lead to exactly that result. His efforts helped put a much larger and much more well-resourced company than Techdirt completely out of business.

So, in our view, this is not a fight about who invented email. This is a fight about whether or not our legal system will silence independent publications for publishing opinions that public figures do not like.
And here’s the thing: this fight could very well be the end of Techdirt, even if we are completely on the right side of the law.


I don’t agree with Masnick on everything (specifically: the value of copyright) but he always argues his case well; and he’s not afraid to call a spade a spade. It’s likely Techdirt will set up a legal defence fund. That could be a worthy cause for donations.

It’s also a sign of the US’s media getting an inkling of what it’s like to be in the news business in the UK: you always need to be thinking about libel risk.
link to this extract

The slow, sad, and ultimately predictable decline of 3-D printing •

John Brandon:


I often think of that Saab part [a cupholder which had broken, and which he wanted to replace] as a good model of what went wrong. (By the way, I promise not to use any metaphors for 3-D printing from now on.) For starters, I really wanted to print the cup holder. I even asked a well-known Thingiverse designer for help, and was going to pay him, but he said the part was too complex. Wait, what? Too complex for a well-known designer? He even used the word “hassle” in his email back to me. The part is not something you’d use on a NASA spaceship. It does have a spring attached to two pieces of plastic that fold together.

What about a water bottle cage for my bike? Shouldn’t be a big problem. There are plenty of designs. But when I actually printed one of them, it broke on my first ride. Also, a much more important piece of data: A water bottle cage costs about $4 at but even a relatively short spool of filament costs $65. The math doesn’t compute. And, it doesn’t make sense to spend the time.

From that experience, I knew something was wrong. As the Newsweek article notes, you can print only so many Yoda heads before you wonder why you bought the device. A 3-D printer won’t magically terraform anything right before your eyes, and it even has problems with slightly complex car parts. I even remember my nephew, who is working as an intern for me this year, saying the industry needs to figure out this problem. It’s fun for a while, but eventually you realize you need to do something practical after paying almost $1,000 for the product.


It’s that: the costs just never add up.
link to this extract

Errata, corrigenda and ai no corrida: “Alibaba” was mispelt as “Alababa” in the article about Yahoo yesterday. What’s an extra i between friends, though?

You can now sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.