Start Up: Amazon’s music challenge, spotting photo fakes, Apple patents Siri dock, and more

Mozilla wants to create an open-source speech recognition database. But why, exactly? Photo by royalconstatinesociety on Flickr.

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.

A selection of 10 links for you. Peu de trop. I’m @charlesarthur on Twitter. Observations and links welcome.

Ethereum co-founder says crypto coin market is a timebomb • Bloomberg

Camila Russo:


Initial coin offerings, a means of crowdfunding for blockchain-technology companies, have caught so much attention that even the co-founder of the ethereum network, where many of these digital coins are built, says it’s time for things to cool down in a big way.

“People say ICOs are great for ethereum because, look at the price, but it’s a ticking time-bomb,” Charles Hoskinson, who helped develop ethereum, said in an interview. “There’s an over-tokenization of things as companies are issuing tokens when the same tasks can be achieved with existing blockchains. People are blinded by fast and easy money.”

Firms have raised $1.3 billion this year in digital coin sales, surpassing venture capital funding of blockchain companies and up more than six-fold from the total raised last year, according to Autonomous Research. Ether, the digital currency linked to the ethereum blockchain, surged from around $8 after its ICO at the start of the year to just under $400 last month. It’s since dropped by about 50 percent.

Hoskinson, who runs technology research firm IOHK, is part of a growing chorus of blockchain watchers voicing concern about the rapid surge in cryptocurrency prices and digital coin crowdsales that have collected millions of dollars in minutes. Regulation is the biggest risk to the sector, as it’s likely that the U.S. Securities and Exchange Commission, which has remained on the sidelines, will step in to say that digital coins are securities, he said.


link to this extract

Can you tell whether this photo has been manipulated? • Science | AAAS

Giorgia Guglielmi:


If you were fooled by the recent photo of Donald Trump and Vladimir Putin having an intense discussion at the G20 summit, don’t feel bad. In a recent study people were only able to spot faked images 60% of the time. And almost half of the time they were not able to tell where an image had been altered, researchers report today in Cognitive Research: Principles and Implications.

To conduct the research, scientists sourced 10 photos from Google and altered six of them with image editing software. Then they asked more than 700 volunteers whether the images had been manipulated. Here’s one: 

Is this photo manipulated? (If you go to the article, you can click on it to find out.)

In a second experiment, the scientists developed an online test to judge people’s ability to locate manipulations. They asked participants to tell where an image had been manipulated, regardless of whether people said the image had been altered in the first place.


I’d say 60% is quite high for spotting fakes – usually these things are off and around the internet before anyone has queried them. (I’m terrible at spotting them.) The paper, linked, is educational.
link to this extract

Ransomware attack puts KQED in low-tech mode • San Francisco Chronicle

Marissa Lang:


The journalists at San Francisco’s public TV and radio station, KQED, have been stuck in a time warp.

All Internet-connected devices, tools and machinery have been cut off in an attempt to isolate and contain a ransomware attack that infected the station’s computers June 15. More than a month later, many remain offline.

Though the stations’ broadcasts have been largely uninterrupted — minus a half-day loss of the online stream on the first day of the attack — KQED journalists said every day has brought new challenges and revealed the immeasurable ways the station, like many businesses today, has become dependent on Internet-connected devices.

“It’s like we’ve been bombed back to 20 years ago, technology-wise,” said Queena Kim, a senior editor at KQED. “You rely on technology for so many things, so when it doesn’t work, everything takes three to five times longer just to do the same job.”


Notable that the only computer being used in the story is a Mac. Externalities of Windows are multifarious.
link to this extract

Apple granted patent on smart dock with Siri and wireless charging • The Verge

Chaim Gartenberg:


Apple was granted a new patent this week, one that’s particularly interesting given Apple’s upcoming HomePod and rumors of a wirelessly charging iPhone 8: an iPhone dock that could have Siri and a wireless charger built in.

The patent, officially for a “Smart dock for activating a voice recognition mode of a portable electronic device,” is pretty broad. It covers a dock that could recognize that an iPhone had been placed into it and activate a microphone that could listen for voice commands to allow users to control a phone from across a room. In other words, it’s a Siri dock. The patent also covers multiple ways of charging said iPhone, including wireless charging, and describes docks that range just simple connectors with a microphone and speaker to full-fledged miniature computers with buttons and displays.

Now, before I go off into rampant speculation, it’s worth remembering that this is a patent, not an actual product announcement. But the interesting part is how this could tie in to Apple’s HomePod strategy.


So it’s the Apple iPod Hi-Fi living again?
link to this extract

Mozilla is crowdsourcing a massive speech-recognition system • Fast Company

Sean Captain:


From Amazon’s Alexa to Apple’s Siri, speech recognition and response are becoming mainstays of how we interact with computers, apps, and internet services. But the technology is owned by giant corporations. Now the Mozilla Foundation, maker of the free Firefox browser, is recruiting volunteers to train an open-source speech recognition system.

Project Common Voice recruits web surfers to spend a couple minutes reading sentences aloud and/or listening to other people’s recordings to check their accuracy. It’s a very minimal contribution for volunteers: Just read three sentences to help the system understand what everyday speech sounds like. No need to go to a soundproof room or get a high-quality microphone. “We want the audio quality to reflect the audio quality a speech-to-text engine will see in the wild,” reads the projects FAQ. “This teaches the speech-to-text engine to handle various situations—background talking, car noise, fan noise—without errors.

Mozilla is out to collect at least 10,000 hours to train a database that anyone can use for free.


Not quite sure who benefits from this. An open-source speech recognition system is only going to be as good as its training data – and it needs millions of peoples’ voices.
link to this extract

Amazon is now the 3rd biggest music subscription service • Music Industry Blog

Mark Mulligan:


At MIDiA we have long argued that Amazon is the dark horse of streaming music. That horse is not looking so dark anymore. We’ve been tracking weekly usage of streaming music apps on a quarterly basis since 2016 and we’ve seen Amazon growing strongly quarter upon quarter. To the extent that Amazon Music is now the 2nd most widely used streaming music app, 2nd only to Spotify which benefits from a large installed base of free users to boost its numbers. So, in terms of pure subscription services, Amazon has the largest installed base of weekly active users.

But it’s not just in terms of active users that Amazon is making such headway. It is racking up subscribers too. Based on conversations with rights holders and other industry executives we can confirm that Amazon is now the 3rd largest subscription service. Amazon has around 16 million music subscribers (ie users of Amazon Prime Music and also Amazon Music Unlimited subscribers). This puts it significantly ahead of 4th and 5th placed players QQ Music and Deezer and gives it a global market share of 12%.


Makes sense, if you assume people are playing music via their Echo device. And so of course all the use is concentrated in the Prime markets – US, UK, Germany, Japan – where it has 40m potential users.

Mulligan estimates there are 13m Echos in use. All of which could be good for Apple’s plan to get people to play music through a home smart speaker. Though it’s 13m behind. (There are about 125m households in the US, 27.1m in the UK, 37.5m in Germany, 49m in Japan. So a fair bit of room to expand into for everyone.)
link to this extract

Google to launch automated feed based on users’ interests • FT

Richard Waters:


Google took a step closer to competing with Facebook’s core news feed service on Wednesday as it announced a new automated “feed” of its own to deepen its connection with users on mobile devices.

Unlike the mobile feeds of services such Facebook and Twitter, however, Google’s version will be based entirely on what the company knows about its users’ interests rather than their social connections, drawing on the personalised data and technology platform already built to support its core search engine.

The Google feed will initially be available only on the company’s main mobile search app, though it will eventually also appear in browsers and on the page, said Ben Gomes, the company’s vice-president of engineering.

The prospect of preloading information on to the normally pristine search page echoes a shortlived attempt more than a decade ago to turn it into a home for personalised information.

The feed is designed to contain news and information tied to users’ interests, based on things they have searched for before. It will also draw on other things Google knows about its users, for instance serving up a range of information in anticipation of an upcoming trip.


The “shortlived attempt” would be iGoogle, launched in May 2005 and killed in November 2013 “because the company believed the need for it had eroded over time”. Nope.
link to this extract

Google Glass 2.0 is a startling second act • WIRED

Steven Levy:


The difference between the original Glass and the Enterprise edition could be summarized neatly by two images. The first is the iconic photo of Brin alongside designer Diane von Furstenberg at a fashion show, both wearing the tell-tale wraparound headband with display stub. The second image is what I saw at the factory where Erickson works, just above the Iowa state line and 90 miles from Sioux Falls, South Dakota. Workers at each station on the tractor assembly line—sporting eyewear that doesn’t look much different from the safety frames required by OSHA—begin their tasks by saying, “OK, Glass, Proceed.” When they go home, they leave their glasses behind.

These Jackson, Minnesota, workers may be onto something. A recent Forrester Research report predicts that by 2025, nearly 14.4 million US workers will wear smart glasses. It wasn’t referring to fashion runways. It turns out that with Glass, Google originally developed something with promising technology—and in its first effort at presenting it, failed to understand who could use it best and what it should be doing. Now the company has found a focus. Factories and warehouses will be Glass’s path to redemption.


It’s been selling by the hundred, apparently. A niche; even if that Forrester report pans out, don’t expect that Google will have the market all to itself. Though I’ve always said that Google Glass’s best chance was in the business, not consumer, market.
link to this extract

How I discovered the first big mobile privacy scandal • Motherboard

Alasdair Allan:


But as Pete [Warden] put it at the time, “The main reason we went public with this was exactly because it already seemed to be an open secret among people who make their living doing forensic phone analysis, but not among the general public.”

Apple’s immediate response to the story was also perhaps somewhat disingenuous. “The iPhone is not logging your location,” it said. “Rather, it’s maintaining a database of Wi-Fi hotspots and cell towers around your current location, some of which may be located more than one hundred miles away from your iPhone, to help your iPhone rapidly and accurately calculate its location when requested.” This ignored that fact that if the phone is storing a list of access points and cell towers around your location then the center of those separate points will be a good approximation of your location. After all, that was the whole point of storing them in the first place. Contrary to Apple’s claims in its initial response, the phones continued to store location data even when location services were disabled…

…By 2013, Apple was still collecting location data. But this time they were exposing it in the user interface and allowing users to manage it. These days, locationgate wouldn’t even be a story.

Since then, people have become a lot more comfortable with the idea of sharing location data, while at the same time becoming a lot more nuanced about how that data is shared. Recent privacy scandals, such as when Uber updated its app asking users to share their location all the time, even when the app wasn’t running—is illustrative. People are OK with their phone tracking their location, but want control over how it’s shared.


I had the “newspaper” exclusive on this: Alasdair and Pete had an O’Reilly blogpost, and I had a Guardian article, and they went live at the same time. It was a huge, huge story at the time.
link to this extract

Bixby feels more like a return of the old Samsung than a path to the future – The Verge

Dan Seifert:


then there’s Bixby Voice, which is the heart of the Bixby experience. Samsung touts Bixby Voice as a way to interact with your phone more than a virtual assistant, but in reality, the line between the two is far blurrier than the company describes. You can ask Bixby Voice for weather information, facts, upcoming appointments, and so on. You know, all of the stuff that you can also ask the Google Assistant to do.

Bixby’s big difference, as Samsung would like you to believe, is that you can also ask Bixby to open apps and perform actions on your device. You can ask Bixby to open the settings app and turn up your screen’s brightness, or ask it to send the picture you’re looking at in the gallery app to your significant other through a text message. Ideally, it will do these things automatically, without requiring you to manually switch apps or type in a contact’s info.

But in practice, using Bixby Voice to do things on your phone is not any faster or easier than just tapping the touchscreen the old-fashioned way. It takes so long for Bixby Voice to launch (whether I press and hold the Bixby button to activate listening mode or use the “Hi Bixby” wake command) and hear what I’m saying that I could have performed the task three times by the time it has processed what I said and performed my action.

Further, there are many times when I ask Bixby to do something and it either doesn’t understand me or just doesn’t do what I expect it to. Just this morning I asked Bixby to “take a screenshot and share it to Twitter.” It got the screenshot part right, but then it attempted to share the image in a private DM conversation instead of a public tweet. I had to start the process all over again in order to do what I could have done manually the first time.

Basically, after weeks of using Bixby Voice, I still can’t trust that it will do what I want it to. That means that I won’t use it and will continue to use my phone the same way I always have. Trust will be even more important when Samsung brings Bixby to other appliances.


And he doesn’t have any in it. This is years late.
link to this extract

Errata, corrigenda and ai no corrida: none notified

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 )

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.