Year: 2018

14 May 2018

You can now try Smart Compose in the new Gmail

Smart Compose, the experimental autocomplete feature in the new Gmail on the web that Google announced at its I/O conference last week, is now available for testing.

Smart Compose is an AI tool that promises to automatically finish your sentences for you, using what it has learned about how people typically write. Based on my experience so far, it’s not quite as good as Google’s demo made us believe it was, but it’s still quite useful and will likely save you a few keystrokes as you go about your day.

You’ll have to enable “Experimental Access” in the new Gmail settings to be considered for this first test. I did so last week and the new feature is now live in my account.

 

I admit that I always feel a bit empty inside when I use Smart Reply, the somewhat more limited version of this feature in the mobile Gmail app that provides you with a few potential two or three-word replies. And I always wonder if the person on the other end knows I was too lazy to write a real answer. But it also makes me feel more productive because I end up answering more emails. It’s a trade-off that Smart Reply is currently winning. My guess is, the same will happen with Smart Compose.

For now, though, Smart Compose is still quite limited (and only works in English). When it works, it’s almost magical, and the suggestion is almost always spot on. But it only works for rather trite sentences so far. If you go off the script, you could write paragraph after paragraph without ever seeing the prompt.

It’ll happily autocomplete any cliché and write “Hi [name],” at the top of your email, which I guess is something, but that doesn’t feel especially intelligent. We’re still looking at an experimental feature, though, and these tools tend to get better as they learn more about how users behave.

14 May 2018

NES Classic loaded with classic manga games raises hopes for other special editions

Japanese gamers and manga aficionados and every combination thereof will get a treat this summer with the release of a NES Classic Edition loaded with games from the pages of Weekly Jump. The beloved manga mag is celebrating its 50th anniversary and this solid gold Famicom is part of the festivities.

There’s basically no chance this Jump-themed NES will get a release in the US — first because hardly any Americans will have read any of these manga (with a couple exceptions) and second because even fewer will have played the Famicom games associated with them.

Familiar… and yet…

That said, this nurtures the hope inside me that we will at some point see other themed NES Classics; the original has, of course, a fantastic collection — but there are dozens more games I would have loved to see on there.

You can hack the thing pretty easily and put half the entire NES library on it, but Nintendo’s official versions will have been tested and perhaps even tweaked to make sure they run perfectly (though admittedly emulation problems aren’t common for NES games).

More importantly it’s possible these hypothetical themed consoles may come with new accessories that I desperately need, like a NES Advantage, Zapper (not sure how it would work), or NES Max. Perhaps even a Power Glove?

In the meantime, at least if you missed the chance to buy one the first time around, you can grab one come the end of June.

14 May 2018

Tinder’s upcoming location-based feature seems a bit creepy

Do you want random Tinder users to see where you’ve been? Uh, no? Well, great news: an upcoming Tinder feature called Places will allow for just that. According to screenshots detailing Tinder Places uncovered by The Verge, the dating app is developing a feature that tracks your location via its app, then shows potential matches where you’ve been. The idea is to allow people to come across their real-life missed connections, similar to how the dating app Happn works today.

There are some caveats about the new feature. For starters, this is something Tinder has in testing – the way it works at launch could be different. Also, the feature can be shut off, the documentation says – a toggle in the app’s settings let you turn it on or off at any time. And we’ve learned that, thankfully, this feature will be opt-in.

However, that’s a decision you should approach with caution.

Above: Places documentation, image credit: The Verge

The way Tinder has implemented the location feature is concerning. Instead of allowing users to explicitly “check in” to a given place – like their favorite coffee shop or a cool restaurant or bar – Tinder continuously tracks users’ location with its app, then makes a determination about which of your “places” it will show to your potential matches.

The company, at least, thought to remove things like doctors, dentists, banks, and the place where you live or work from this automated location-sharing option. It also won’t add a place to your list until after you’ve left – though it doesn’t say how long it waits to do so. (The documentation hedges on the timeframe by saying things like “we’ll wait a while” or “it’ll take some time.”)

While Tinder says your recent places will expire after 28 days – meaning, other Tinder users won’t be able to see where you’ve been past that point – the company does appear to be keeping a wider history of users’ location and travels for itself. The documentation explains that Tinder will use this Places information in order to improve the product – by learning which places lead to matches, which users are always deleting, and it will use the data to improve its ability to show users better matches.

Above: Tinder Places, image credit: The Verge

In other words, Tinder will be tracking you, as well as giving potential matches the ability to narrow down the parts of the city you frequent – right down to your daily habits. That means potential matches could figure our things like which bar you regularly hit up for after-work drinks, where you work out, what your favorite breakfast spot is, and so on.

The advantage to daters gaining access to this information about other Tinder users is fairly limited. After all, simply hitting up the same Starbucks in the morning isn’t any sort of signal about someone’s potential as a love match.

But it does put a lot more data into the hands of potential stalkers, while offering Tinder access to a massive treasure trove of location data – the selling of which, even anonymized and in aggregate, could be a big business. Even if Tinder doesn’t aim to sell the data directly, it clearly paves the way for the company to show more specific location-based ads in its product.

It also lets Tinder group users into cohorts regarding their interests – without explicitly asking for that data, like Facebook does. For example, Tinder would know if someone shows up at church every week, or regularly takes their dog to a dog park – things it could use to classify users and match them accordingly.

That’s useful to some extent, in a handful of cases – but just because you have a dog, doesn’t mean you need to date someone with a dog, too. In the end, it’s less useful to have “things” in common with people – it’s more useful to share the same values, experts say. And those values are more important than the initial attraction (which fades as the hormones wear off), and more important than a set of common interests – those can be negotiated in a relationship.

In the end, there’s far more for Tinder to gain here, than users to gain from the Places feature – especially with the downside regarding its potential for harassment or stalking.

One serious concern was whether Places would be opt out or opt in – the documentation The Verge found didn’t make this clear. However, we’re relieved to hear (from people familiar with product) that Places is an opt-in experience.

That this feature’s launch is nearly is not a surprise. Tinder already said it was working on rolling out a new location feature this year during its earnings calls, something it described as having the potential to bring in a new audience and “expand the definition of dating.” That could imply the company wants to make Places more of a social networking, or friend-finding feature, rather than just an option for finding dates.

14 May 2018

Say hello to Google One

Google is revamping its consumer storage plans today by adding a new $2.99/month tier for 200 GB of storage and dropping the price of its 2 TB plan from $19.99/month to $9.99/month (and dropping the $9.99/month 1 TB plan). It’s also rebranding its storage service as ‘Google One.’

Going forward, you’ll also be able to share your storage quota with up to five family members.

That by itself would be interesting, given how easy it is to max out 100 GB with 4K videos and high-res images these days, but there is one other feature here that explains the new brand name: free one-tap access to Google Experts for help with any Google consumer product and service.

That access to live experts — not some barely functional AI chatbot — comes with every Google One plan, including the $1.99/month 100 GB plan. In the U.S., these experts will be available 24/7 over chat, email and phone. In other countries, this lineup of support options may vary, but the company tells me that its objective is “to provide users with great one tap support and constantly improve it over time.”

Google already offered 24/7 support for paying business users with a G Suite account, but this is the first time it actively offers live support for consumers.

It’s worth stressing that the existing free quota of 15 GB will remain,

In addition to access to experts, the company also promises to provide subscribers with other benefits. Google One’s director Larissa Fontaine told me that those could include discounts on hotels you find in Google Search, preferred rates for other Google services, or credits on Google Play. “We hope to build those out over time,” she noted.

Brandon Badger, Google’s group product manager for Google One, told me that the team looked at how people use the storage plan. Users now have more devices, shoot more 4K video and share those files with more family members, who in turn also have more devices. “We are looking with this plan to accommodate that,” he said.

In addition, Fontaine noted that users with paid storage accounts also tend to be heavy Google users in general, so combining storage and support seemed logical.

Sadly, this isn’t an immediate change. Over the course of the next few months, Google will upgrade all existing storage plans to Google One accounts, starting in the U.S., with a global rollout following after that. Google also tells me that it will roll out a new Android app to help users manage their plan (not their files).

While the focus of today’s announcement is on storage, it’s hard not to look at this new offering in the context of the additional support and other bonus features that Google promises. Google One is clearly about more than simply a better storage plan offering. Instead, it feels like the beginning of a new, more ambitious offering that could be expanded to include other services over time. Maybe a single subscription to all Google consumer services including Drive, YouTube Red and Play Music (or whatever becomes of that)? Despite its name, Google One is currently only one of many subscription services the company offers, after all.

14 May 2018

A free web tool can predict your hair, skin, and eye color from DNA data

A new tool by researchers at the School of Science at IUPUI and Erasmus MC University Medical Center Rotterdam in the Netherlands can predict your hair, skin, and eye color from your DNA data. The system, which is essentially a web app that can accept DNA sequences, compares known color phenotypes to known data and tells you the probabilities of each color.

The app, called HIrisPlex-S, can tell colors from even small amounts of DNA like that left at a crime scene.

“We have previously provided law enforcement and anthropologists with DNA tools for eye color and for combined eye and hair color, but skin color has been more difficult,” said forensic geneticist Susan Walsh from IUPUI. “Importantly, we are directly predicting actual skin color divided into five subtypes — very pale, pale, intermediate, dark and dark to black – using DNA markers from the genes that determine an individual’s skin coloration. This is not the same as identifying genetic ancestry. You might say it’s more similar to specifying a paint color in a hardware store rather than denoting race or ethnicity. If anyone asks an eyewitness what they saw, the majority of time they mention hair color and skin color. What we are doing is using genetics to take an objective look at what they saw.”

You can actually try the web app here but be warned: it’s not exactly the most user friendly app on the web. It requires you to know specific alleles for your test subject or upload a set of alleles in a csv file. It is, however, free and looks like it could wildly useful in law enforcement and figuring out what your hair color was before you dyed it purple.

“With our new HIrisPlex-S system, for the first time, forensic geneticists and genetic anthropologists are able to simultaneously generate eye, hair and skin color information from a DNA sample, including DNA of the low quality and quantity often found in forensic casework and anthropological studies,” said Manfred Kayser of Erasmus MC.

14 May 2018

AWS introduces 1-click Lambda functions app for IoT

When Amazon introduced AWS Lambda in 2015, the notion of serverless computing was relatively unknown. It enables developers to deliver software without having to manage a server to do it. Instead, Amazon manages it all and  the underlying infrastructure only comes into play when an event triggers a requirement. Today, the company released an app in the iOS App Store called AWS IoT 1-Click to bring that notion a step further.

The 1-click part of the name may be a bit optimistic, but the app is designed to give developers even quicker access to Lambda event triggers. These are designed specifically for simple single-purpose devices like a badge reader or a button. When you press the button, you could be connected to customer service or maintenance or whatever makes sense for the given scenario.

One particularly good example from Amazon is the Dash Button. These are simple buttons that users push to reorder goods like laundry detergent or toilet paper. Pushing the button connects to the device to the internet via the home or business’s WiFi and sends a signal to the vendor to order the product in the pre-configured amount. AWS IoT 1-Click extends this capability to any developers, so long as it is on a supported device.

To use the new feature, you need to enter your existing account information. You configure your WiFi and you can choose from a pre-configured list of devices and Lambda functions for the given device. Supported devices in this early release include AWS IoT Enterprise Button, a commercialized version of the Dash button and the AT&T LTE-M Button.

Once you select a device, you define the project to trigger a Lambda function, or send an SMS or email, as you prefer. Choose Lambda for an event trigger, then touch Next to move to the configuration screen where you configure the trigger action. For instance, if pushing the button triggers a call to IT from the conference room, the trigger would send a page to IT that there was a call for help in the given conference room.

Finally, choose the appropriate Lambda function, which should work correctly based on your configuration information.

All of this obviously requires more than one click and probably involves some testing and reconfiguring to make sure you’ve entered everything correctly, but the idea of having an app to create simple Lambda functions could help people with non-programming background configure buttons with simple functions with some training on the configuration process.

It’s worth noting that the service is still in Preview, so you can download the app today, but you have to apply to participate at this time.

14 May 2018

Supreme Court allows states to legalize sports betting, opening floodgates for online gambling profits

Whoever had the over on DraftKings‘ boss Jason Robins and FanDuel chief executive Matt King being given a potential billion dollar windfall by the Supreme Court’s decision to allows sports betting should head to the cashier’s cage.

In a six-to-three decision (Justice Breyer was a partial dissent), Supreme Court Justices struck down a federal law that had banned gambling on sporting events in most states.

The implications of this for state tax revenues, and around arguments for making significant changes to the ways college athletes are compensated (or should be compensated), are huge, but clear winners from this ruling are the online betting companies… or any media company that has any sort of exposure to live streaming sporting events.

DraftKings and FanDuel seem like clear early winners, but really there’s a market for Facebook, Twitter, YouTube, and the major networks that hold sports broadcasting rights to open up new sources of revenue. For the two leaders in online sports betting, the decision is a new lease on life — although both companies had argued that they were games of skill and not chance, and should not be regulated as odds-based gambling companies.

At stake is an industry that former NBA commissioner Adam Silver pegged at nearly $400 billion in a New York Times editorial from 2014.

Specifically, the case before the Supreme Court concerned the Professional and Amateur Sports Protection Act of 1992, which outlawed states from approving sports gambling. At the time, the court gave exemptions to states that already held sports lotteries and Nevada — because… well… Nevada . Other states had the opportunity to opt in, but they couldn’t get their acts together in time to approve sports betting in their legislatures.

Mainly, the current court was concerned with whether the law represented an instance of commandeering by the federal government (basically improperly stepping on states’ rights). At the heart of the case was New Jersey voters’ decision to legalize sports betting in the state in 2011 as revenues from Atlantic City’s casinos (the only places where hopelessness and schadenfreude converge) declined.

The New Jersey decision was challenged by every major American sporting league and the New Jersey law was struck down by the federal courts. When Jersey tried to change the law to make sports betting legal at at places where gambling was already permitted the leagues again swooped in to challenge the law — and they won again.

In the recent Supreme Court decision, the Justices decided that the Federal government could not commandeer states’ resources for national purposes.

The ruling MAY also have implications for another industry that has been trying to get on more secure legal footing. Indeed, the (very conservative) New York Post published an op-ed at the end of last year basically arguing in favor of letting states decide on the legality of both gambling and marijuana nationwide.  

14 May 2018

Didi Chuxing can now test self-driving cars in California

Quite a number of companies — 53, to be exact — have received permits to test self-driving cars with a safety driver in California. One of the more notable companies that has received a permit is China-based Didi Chuxing, which officially opened up its U.S.-based research lab last March.

Also on the list of permit holders is Udacity, which partnered with Didi last year to host a contest for the development of an Automated Safety and Awareness Processing Stack.

In April, the California Department of Motor Vehicles introduced some new regulations and permit application processes for self-driving car companies. The new system entails three autonomous vehicle permit options: testing with a driver, driverless testing and deployment. Most of the new elements of the regulations are around driverless testing and deployment.

For example, in order to conduct driverless testing, companies must have previously tested the vehicles in controlled conditions. The vehicles must also, among many other things, meet the definition of an SAE Level 4 or 5 vehicle. With deployment, companies need to ensure cars can detect and respond to roadway situations, meet best practices to detect cyberattacks and more.

Another notable permit holder is Apple, which in January expanded its fleet to include 27 cars. Missing from the list, unsurprisingly, is Didi rival Uber.

Last week, Uber CEO Dara Khosrowshahi said Uber will bring back its self-driving cars “within the next few months, I don’t know.”

To be fair, Khosrowshahi didn’t sound 100 percent confident in his answer. That’s just to say I wouldn’t be surprised if it takes Uber longer than a few months. He went on to say that when Uber does bring its cars back on the road, “it’s going to be in as safe of a way as possible.”

Uber pulled its self-driving cars off the roads following the March fatal crash. Later that month, Uber decided not to reapply for a self-driving car testing permit in California. Uber’s previous permit expired March 31.

If Uber wants to continue its tests in California, it will need to apply for a new permit, as well as “address any follow-up analysis or investigations from the recent crash in Arizona,” DMV Deputy Director/Chief Counsel Brian Soublet wrote in a letter to Uber in March. Uber may also need to set up a meeting with the DMV.

I’ve reached out to Didi to learn more about its plans for self-driving cars in CA and will update this story if I hear back.

14 May 2018

Apple’s App Store redesign improved app discovery, report finds

When Apple introduced its completely redesigned App Store last fall, one of its goals was to improve app discovery by placing a larger emphasis on editorial content – including things like “app of the day” picks, lists, how-to’s and even interviews with app developers, among other things. Now, a new study from Sensor Tower reveals those changes appear to have been working.

According to Sensor Tower’s findings, more apps are being discovered by way of browsing the App Store following the redesign launched in September.

Before, browse-driven downloads accounted for around 10 percent of all downloads. With the new App Store, they’ve grown to more than 15 percent. And that increase has held steady into 2018, even as the initial excitement around the App Store revamp has worn off.

Despite the growth in app discovery by browsing, searching for app by typing keywords into the search box is still, by far, the primary way consumers are finding and downloading new apps. Today, search accounts for 65 percent of downloads – well ahead of browse, referrals, or other methods.

Sensor Tower based its findings on data collected on app downloads between May 2017 and April 2018, it says.

The report also delved into the differences between how consumers discover apps and games.

As it turns out, browsing plays a much more significant role in game discovery than it does for non-game apps. Only 56 percent of game downloads came from search, compared with 69 percent for non-games. Meanwhile, browse contributed to 24 percent of game downloads, compared to just 9 percent of non-game downloads.

What this seems to indicate is that iOS users are turning to the App Store and its editorial recommendations in greater numbers to learn about what new game to try next. Plus, the fact that games can now include a video preview, and labels like “Editor’s Choice” are better highlighted in the new App Store also likely help people get a better sense of which ones to install, as they browse.

Sensor Tower’s findings about game downloads line up with research released last month where it found that games that were featured as the “Game of the Day” could see their downloads increase by 802 percent, compared to the week prior to being featured. Apps, by comparison, saw boosts of 685 percent.

The new report’s findings are good news for Apple which had a sizable challenge to tackle with its App Store redesign. Its app marketplace had grown almost over-crowded over the years. And even after the big app cleanup, it still stands at over 2 million apps. Finding a way to better introduce favorites and newcomers to iOS users at this scale was a tall order, but the growth in apps discovered by way of browsing indicates Apple has seen some success on this front. 

14 May 2018

Exploit puts popular web and mobile apps at risk

A new exploit could allow users to bypass security checks in Electron, a popular cross-platform development framework. The exploit, posted by Trustwave, has been patched and developers should update their apps as soon as possible.

The exploit could allow cross site scripting in some apps by turning on nodeIntegration, a method that allows the app to not only connect to its own modules but also Node.js modules.

From the announcement:

Electron applications are essentially web apps, which means they’re susceptible to cross-site scripting attacks through failure to correctly sanitize user-supplied input. A default Electron application includes access to not only its own APIs, but also includes access to all of Node.js’ built in modules. This makes XSS particularly dangerous, as an attacker’s payload can allow do some nasty things such as require in the child_process module and execute system commands on the client-side. Atom had an XSS vulnerability not too long ago which did exactly that. You can remove access to Node.js by passing nodeIntegration: false into your application’s webPreferences.

Many popular apps use Electron including Discord, Signal, Visual Studio Code, and Github . Slack also uses Electron for its apps.

The exploit depends on the nodeIntegration setting and the process of opening a new window. While in most cases nodeIntegration is set to false, in some cases you can set nodeIntegration to true and then pass other nefarious scripts including calling the child_process module which lets you make system calls like spawn which then lets you run commands in the operating system.

You can see Electron’s website here and here is their blog post on the update. Most apps shouldn’t be effected as long as you’ve upgraded the platform in the last few weeks.