Josh Lowensohn | The Verge

Former Apple engineers have built a $1,495 oven that can identify your food

2015-06-09T11:34:42-04:00

For the past four months, I’ve learned to operate in silence, moving about my house like a ninja. I pick up and put down each and every item as if it’s a delicate wine glass. Because if I don’t, I might not get any sleep.

I live with a baby now — which, among other things, means food has become a problem. Cooking makes a lot of noise. Time spent in the kitchen to create complicated meals, with their sizzling oils and clanking pans, has flown out the window in favor of takeout and things I can whip together in relative silence. Last week I actually went outside my house to remove Chinese food containers from a noisy plastic bag. Better to weird out my neighbors than mess with nap time.

It won’t be like this forever, but having a baby has kindled my love for silent, efficient gadgets, of which there are few in the kitchen — save the oven. Nearly everything else in my house beeps and bloops with reckless abandon. It makes the act of simply twisting a dial and walking away seem attractive by comparison. Turns out I’m not the only one with a fondness for this particular appliance; a team of engineers in San Francisco like it, too. They just think most ovens have become dated. So they’ve spent the better part of two years building June, a slightly crazy high-tech oven that has enough computing horsepower to run Half-Life 2.

June is not a gaming rig, but has all the trappings of a powerful computer. It’s a microwave-sized countertop oven with a 5-inch touchscreen, Wi-Fi, a quad-core Nvidia Tegra K1 processor, a high-definition camera that’s protected behind insulated glass, and a single control dial that’s its only real physical button. Paired with six carbon-fiber heating elements and a set of convection fans, it promises not only to cook items up to a 12-pound turkey in size, but to figure out what you’re making within a second or two of sticking a dish onto its racks. It’s a neat parlor trick, but also one the company is trying to combine with an estimation of each dish’s typical cooking time so that it’s harder to screw up your meal.

“Right now we are experts in steak, chicken, white fish, salmon, bacon, cookie dough, brownie mix, toast, bagels, and hamburger buns,” says June’s co-founder and CEO, Matt Van Horn, adding that the company plans to expand what the oven can identify over time with software updates.

An oven that recognizes your food

To accomplish the magic trick, June has a camera system built into the very top of the oven that pulls double duty. It’s running computer vision algorithms on your dishes to identify them, but it also lets you check in on what’s happening inside your oven with your tablet or smartphone. How does it tell the difference between two pieces of meat that look nearly the same? It’s all about the tiny details, says June’s co-founder and CTO, Nikhil Bhogal. “Natural foods all have micro-textures, which look different upon closer inspection,” he tells me. “With pork versus beef, they both have different-looking fat patterns. That’s what we train the computer to do.”

That system is combined with some juggling of the fans and heating elements to cook food in specific ways. Pop two sides of a bagel in, and the camera figures out not only that it’s a bagel, but also which side is up. It then adjusts its six heating coils and two fans to make the tops crispy and the bottoms a little softer. The same system works with June’s program for a roasted chicken so that it will blast the top of the dish at the very end of cooking to make sure the skin is crispy while the rest of the bird isn’t overcooked. The company is trying to build these kinds of programs itself, but also plans to let users program their own special cooking cycles on their phones and tablets, These can be shared with other June owners so you can begin to crowdsource ideal cooking patterns.

Bhogal was most recently at Path building its iOS app, but previously spent five years at Apple working on its camera technologies. “If you’ve ever taken a picture or recorded a video, done FaceTime, or taken a panorama, you’ve used a lot of my code,” he says. He left Path with Van Horn near the end of 2013, and the pair has been working on this ever since. Along the way, they’ve raised $7 million in funding led by the Foundry Group, and hired 22 engineers — about half of whom were previously at Apple. June is headquartered in a house in San Francisco’s posh Pacific Heights neighborhood. Van Horn says the spot was picked almost immediately for its massive and well-furnished kitchen, which has become the main testing area. They’ve also assembled a makeshift lab on the second floor where other prototypes are being trained to recognize food.

On my visit, I don’t have a chance to see June differentiate between a porterhouse and a pork chop. Instead, I witness a batch of cookies being made. Van Horn slides a small sheet pan of dough balls into a June prototype. The oven’s software is still a little buggy, he warns, but seconds later a small screen on the oven’s front screen pops up to alert us that it’s figured out we’ve loaded it with cookies; we just need to hit a button to confirm it. We do, and the oven comes to life, emitting a small, dull whoosh of fans while a glowing circle pops up that will let us know how much time is left.

Cooking probes measure internal temperature

As the cookies slowly melt into glossy puddles, Bhogal and Van Horn wax poetic about the things they’ve done with June that differ from most ovens. There’s a built-in scale, which helps the oven identify dishes and set cook times; it also zeroes out its own mass, letting you weigh anything set on top of it. There are two plugs in the top of the oven doorway for a pair of cooking probes (which are included), so that you can monitor the internal temperature of something like a steak or a roast. The oven has also been designed with an air gap nearly all the way around that manages to keep every side fairly cool to the touch. Likewise, the front has been designed with three panes of glass and a ventilated air curtain to keep people from accidentally burning themselves while fiddling with its touchscreen controls and selector knob. I’m invited to press my hand up on the glass mid-bake, and sure enough, it barely registers above room temperature. Inside, the cookies are roasting at 350 degrees.

“You can put it on the island and not worry about kids being around it, or accidentally touching it when it’s hot,” Bhogal says. “It’s happened to me. That was one of the motivations to fix [ovens].”

The pair acknowledges that with a just a cubic foot of space (which tops out with the aforementioned 12-pound turkey) and $1,495 price tag, it’s not for everyone. But they believe they’ve brought a number of professional-grade features (mainly the probes and high-end heating elements) down to something that costs far less and is easier to use.

“We feel that this small space works for 80 percent of these cases that have a small family,” Bhogal says. “If you’re making six cookies for dessert for a family of three or four, you don’t need to fire up a 5-cubic-foot oven just to do that. With constant use, this will work out to be much better, energy- and money-wise.” And after biting into one of our stunt cookies, the idea of being able to make more of these on a whim sounds nice.

Coming to everyone next spring

There’s still the question of making an expensive kitchen investment with an unknown appliance company. Warranties and customer service turn out to be vital, as recently highlighted by the horror story of one New York Times writer who went through five months of hell to get one well-rated Samsung oven fixed. June says it’s still working out the details on its warranty and service plans. It has a while to do that: today it’s taking preorders with a $95 deposit, and has plans to ship initial models to an early batch of owners in the fall. A mass-produced version aimed at everyone else will follow next spring.

At least one thing has been assured to me in the meantime: people who don’t like beepy kitchen appliances can turn off all of June’s sounds. Even so, you may still clang and rattle pans along the way. Technology can’t fix everything.

SoundHound’s new voice search app makes Siri and Cortana look slow

2015-06-02T09:00:02-04:00

Nearly a decade ago, SoundHound founder Keyvan Mohajer took an idea to a group of investors. He wanted to make a system that let people talk to computers casually, as if speaking to another human. That was not a new idea of course; 1968’s 2001: A Space Odyssey had a talkative computer as one of its main characters. But Mohajer believed such a thing was no longer science fiction and could become commonplace. The only problem? It might take 10 years to build it.

This is what SoundHound was originally supposed to do

Investors were enamored with the idea, but not Mohajer’s timeline. They said, “Ten years is a long time, can you show me something that will happen in three years?” he recalls. With that, Midomi was born, a service that would let you hum the tune of a song to identify it. Two years later, in 2009, he launched SoundHound, which did the same thing for music overheard on the radio or in the background of a TV show.

Now, nearly a decade after that pitch to investors, Mohajer’s original vision is here in the form of Hound, a voice search app that can handle incredibly complex questions and spit out answers with uncanny speed. Right now, you have to ask those questions inside the Hound app, but the company hopes to get the technology everywhere — even your toaster. That may never happen, but the company’s demonstration of Hound — which was fairly scripted in our case — is astonishing enough to make me believe it’s a possibility.

Mohajer started with a zinger. “What is the population of capital of the country in which Space Needle is located?” he asked briskly. It’s an oddly worded question, but intentionally so, meant to show how well it can extract and process what’s being said. Ask it on any other service (even Wolfram Alpha), and you’ll get the digital equivalent of a head scratch. But here, a robotic voice instantly replied, “The population of Washington, DC is 601,723.” There were two Washingtons there, and it got the right one. In another test, he asked, “How many days are there between the day after tomorrow and three days before the second Thursday of November in 2022?” The app nailed it again.

Hound feels a lot like Google’s Voice Search

Hound the app functions and feels almost exactly like Google’s Voice Search, but seems much faster at identifying words and delivering answers. In our demo, which contained several dozen scripted questions but also some impromptu ones, the words coming out of Mohajer’s mouth popped up on screen nearly as fast as he was saying them, and Hound would pipe back with an answer faster than seemed possible.

Mohajer says the speed comes from SoundHound combining two technologies that are typically separated on competing services. Hound is doing both voice recognition and natural voice understanding in a single engine, whereas rival services break them up into separate steps, first transcribing your question, then extrapolating what you were asking about. That said, our test also took place over Wi-Fi, and in a perfectly quiet room, making it impossible to tell whether Hound maintains these speeds in the real world.

This is a personal assistant without a personality

Unlike Siri or Cortana, Hound doesn’t have a personality. Instead, it’s a sass-free robotic voice. One other area where it’s different is the number of sources it’s pulling from. From the outset, Hound will have about 50 domains, or services it’s tying into through APIs; things like currency converters, news sites, flight status information, and navigation. Mohajer says the plan is to ramp that up into the millions. “Siri launched with 10 domains, and three years later it’s at about 22 new domains, so it takes a long time,” he says.

For example, with Hound’s deal with Expedia, you can ask Hound to find you a hotel in Seattle that costs less than $200 a night, that has free Wi-Fi, parking, and a continental breakfast. It’s the same information you could get on Expedia’s site, of course, but here, there’s no need to click on a bunch of filters. There are other simple tools it’s linked up to as well, things like a mortgage calculator (from a real estate site Mohajer would not disclose) and a speech-based game of Blackjack where you can place bets with your voice.

For everything that doesn’t get picked and assigned to one of Hound’s sources, the app defaults to Microsoft’s Bing. That means web results, including videos and images, are all shown in an integrated browser. Sometimes that’s just fine, but in similar tools like Siri and Cortana, web results are a sign the system couldn’t keep up with what you’re asking of it. Mohajer contends that by kicking people to web results, nobody ends up feeling disappointed, though I’d argue that if it happens enough you’ll just stop using the app entirely and forget about it. I wasn’t quite able to push the boundaries of Hound beyond our demo, something users will get a chance to do once the service launches today.

This has been designed to replace Google, but it can’t just yet

That brings up one of the weaknesses of Hound in its current form: it’s not available as a replacement to other voice assistants. Developers will be able to integrate it into their own apps and hardware creations through a development platform called Houndify, something Mohajer believes will be widely adopted.

“Our vision is that everything can be enabled to have this interface, from millions of phones to billions of other types of devices like consumer electronics and cars,” Mohajer says. “We can’t be the company to build this for every company — we need to enable them to do this for themselves.”

But until that happens, most will know Hound for its app, which will be available only as an invitation-only beta on Android to start, followed by iOS where it will exist as a stand-alone app. That’s a lot like how Siri was a third-party app before Apple bought it, and how Google still is on iOS. It also means that you need to have a very specific reason to use Hound over those built-in options on both platforms.

You still have to go out of your way to use this

It’s worth noting that Hound is arriving at a time when Google and Apple are stepping up efforts to add context to the things people are looking at on their phones, often using voice interfaces, which could almost entirely remove the usefulness of Hound for simple searches. Last week, Google unveiled Now on Tap as part of its upcoming Android M release, a feature that brings its Now service inside of every app and gives the company an incredible amount of context for why you’re looking for something. It hopes it will be good enough that you never even need to leave an app to pull up something you might search for. Apple is also rumored to be working on a feature called Proactive that attempts to put relevant apps and information in front of users without them having to search for it in the first place.

That hurdle of having to find and launch Hound could change if app developers build the voice search into their apps, or if SoundHound and its technology get snapped up by one of these larger players. In the meantime, Mohajer believes that Hound’s performance and experience will be enough for people to go that extra step of launching it before they ask, what they’ve been doing with the company’s audio recognition apps for years.

“Just because it’s easier to get to something is not enough for me to choose it. I don’t use Siri for food, I use Yelp, even though Siri uses Yelp data, because they have a better experience. I use Google Maps on iOS instead of Apple Maps, even though Apple Maps is more integrated,” he says. “I think if you deliver something that is substantially better, people will use it.”

Uber is now lobbying to win your heart

2015-06-01T16:25:30-04:00

Excited teens packed into an SUV safely make their way to prom. A kind-looking father waits patiently in front of his Toyota Prius for a daughter who comes running out excitedly to greet him. After she jumps happily into his arms, he coos, “Good jump!” A second later, we see a woman helping two 20-somethings get an awkward houseplant into the tight confines of their apartment building. That same woman ends up at band practice later on, driving the same car. Everyone is smiling like we’ve just won another World War, or have not seen one another for months.

But actually, it’s because they just got a car ride.

Everyone is happy in Uber world

This is Uber and the world it’s opening up for people, according to a new commercial that popped up on the company’s YouTube account and business blog today. It could honestly be an ad for just about anything — prescription drugs or a smartwatch, even — but the message here seems to be that everyone can be a driver, and they’re happier with Uber around. Also, that your driver is not this person. Or this person. Or this person.

The ad falls along the lines of what Uber has attempted to preach to its drivers and customers in its battles to keep its service alive in places where local regulators have either shut it down or kept it out. The company has launched numerous grassroots campaigns asking for people to call local officials and sign petitions in areas where it’s run into resistance. Perhaps we can view this as a similar form lobbying — just about letting Uber into our lives.

It’s unclear what spurred the ad, but not where it’s going to end up. The company tells The Verge that the commercial will exist only online, meaning it’s not something people will catch while watching prime time TV.

Smart jeans, tiny radar, and other crazy inventions from Google’s ATAP lab

2015-05-29T19:41:05-04:00

I/O is a show about the future of Google, and one of the best places to see what that could look like is from the company’s Advanced Technology and Projects group (ATAP). It’s the skull-and-bones, emblem-toting team behind modular smartphone Project Ara, and up until January was in charge of all-seeing 3D tablet experiment Project Tango. Unlike the rest of Google, ATAP is defined by its deliberate want to create and complete projects in short bursts, or else abandon them. It means that there’s a constant flow of new products and ideas that could either be the next big thing or end up too complex to make it out the door.

ATAP’s latest crop of ideas here at I/O follow that ethos to the letter. We saw touch-sensitive fabrics, tiny radar sensors that can pick up the smallest of hand movements, and a MicroSD card that’s actually a computer designed with a dash of subterfuge to keep private data away from phones, computers, or anything else you stick it into. It might be years before we see these things, or maybe we never will. The allure is less about the gadgets though, and more about ATAP’s ambitions to get them into our hands, pockets, and even suit jackets sometime in the future.

I just bought a soda with someone else’s phone and Android Pay

2015-05-28T16:50:05-04:00

Stop me if you’ve heard this one before, but Google’s got a way for you to pay for things with your phone. “What?” you ask. “That’s been a thing since 2010.” Yes, I say back to you. Yet here we are in 2015 and here comes a new thing called Android Pay. It lets you pay for things at wireless payment terminals using NFC, but it’s also been designed to work within apps so that your payment credentials can follow you around.

Google’s own showcasing of this here at its annual developers conference in San Francisco was just about as limited as when Apple introduced its own payments service (Apple Pay) last year. There are two things you can try to buy with a test phone: a bottle of soda and small, themed Android figurines, purchased from inside of shopping app Wish. I obviously went for the chemical bomb that is a 20-ounce bottle of cola. Doing that required pressing an unlocked phone up to an NFC terminal. I didn’t know their code, so they had to unlock the phone then handed it over to me. I held it timidly toward the NFC reader, and it made a little beep. A small screen on the machine let me know I’d paid in full and with which credit card (it was an American Express), so I hit a button on the front, and caffeine dropped down with a satisfying thud.

The other test felt less viscerally exciting. I went to a special Google I/O-themed page of shopping app Wish, and paid for everything with someone else’s Android Pay credentials. The checkout information was already filled out for me, I only had to tap an on-screen button once. There was no fingerprint needed, just like using Google Wallet.

You realize you’re still just buying small things

That’s really the thing with Android Pay: even nice features like hiding your payment credentials from vendors, and getting to use a rewards card membership at the same time, ends up feeling just like the system it’s emulating. You’re still just buying things.

But that experience won’t seem so painfully familiar forever, at least if Google has its way. Google’s plans to integrate fingerprint sensors into the next version of Android, Android M. Some phones like Motorola’s Atrix, Samsung’s Galaxy S5, S6, and Note 4 smartphones already shipped out with readers. Google’s hope now is that there will be many more phones with that in the future, all to let you use your fingerprint to do things like unlock your phone, pay for in-app purchases, and authenticate yourself inside of apps. For soda, and everything else.

Google I/O 2015: Introducing Android Pay

Hands-on with Google’s new Photos service

2015-05-28T15:44:34-04:00

It’s no secret that Google+ failed to catch on like Google wanted it to. Despite efforts to shoehorn it into other parts of Google — though mainly the dark realm of YouTube comments — people are still flocking to Facebook to stay in touch with friends, and to be entertained with a non-stop barrage of news stories, videos, and photos. Yet while Facebook boasts that it’s the place where the most people are sharing photos on the entire internet, there’s little arguing that it’s one of the worst for organizing them. Facebook’s done a ton of work to overhaul how it stores photos, but very little in the way of letting you find the one you’re looking for in a sea of snapshots.

These things, combined with the general sluggishness among other big photo services, made Google+’s photo service so exciting. It turned out to be a great product with powerful features that would turn busts of photos and related videos into surprisingly-polished things you actually wanted to share with other people. The only problem is that it lived inside of Google+, where your friends might not be.

That’s changing today with Google Photos, a brand new service that’s the spiritual successor to the photo tools in Google+, but that lives on its own. It’s fast, smart, and principally designed to try and help people deal with the oppressive problem that is dealing with the piles of photos they’re shooting on their phones. I’ve just spent the past few minutes playing with it here at I/O, and do not want to leave the building without it. Luckily for me, and soon you — that won’t be a problem.

The first thing you notice about Photos is how fast it is. Even over some very questionable conference Wi-Fi, thumbnails popped up quickly and the full image streams in just a second or two later. I’m told that it’s quite a bit speedier when you’re on a solid Wi-Fi connection. This has kind of become the standard of any photo service, but it’s more impressive when you keep pinching out to see your entire library and it’s all just there. Search was equally snappy. I was able to search for shots by location, and the results popped up faster than I thought they would. All in all, a very nice first impression.

The editing features, which is where I tend to spend a lot of time making tweaks in places like VSCO Cam, are highly simplified. Adjustments can be made either with a filter, or with sliders. On a small screen, that’s about as much fiddling as you want to do, and there’s not much more to it than that. The message seems to be that the auto-enhance button should do most of the job, and if you need much else, you should go edit that photo in another app.

I only used it on a Nexus 6 and iPhone 6, so I’m not able to stack up how the experience compares with the web version or on tablets. But my guess is that most people will be pawing through their shots on phones. In that regard, Google’s added a few nice things here to help you sort through it all with search. You can see your photos sorted just by people and places, or by “things.” That last one is one of the most impressive, since it’s identifying what’s in your photos — or at least trying to. The lead image in a test album with 16 photos of “cars” showed up with a stack of eight boxes of Oreo cookies. There were no cars to be seen. Others in the group nailed it, or were comedically off. But this is absolutely the kind of feature you want with a growing library. Google’s given us several ways to deal with an ungainly number of shots, and it’s all very straightforward.

Dump everything into Google

The big bold idea from Google here is that you should just dump every possible thing into its servers, and access it about the same way from every device. Between the new Flickr and now this, paying for storage as we know it is dead, and with it the need to worry about how much room you have on the device you’re on. Using it with just the small collection of photos I’ve stored with Google thus far, that vision holds up pretty well. But the real test is dumping in your thousands, or even hundreds of thousands of shots, something I wasn’t able to try just yet, but plan on doing now that it’s arrived on both the iOS App Store and Google Play.

Correction 5:29PM ET: This article previously contained a video depicting the former Photos app. Updated 6:18PM ET with further impressions from the app on iOS.

Florida says former Uber driver is an employee, threatening its business model

2015-05-22T20:27:57-04:00

Uber promises big things for its drivers, from plenty of work to flexible hours, and even discounts on vehicles that can be used when drivers aren’t on the clock. However one thing that is not promised when someone becomes an Uber driver is the legal consideration of being called an “employee” versus an “independent contractor.” Anyone who’s worked in a company with both roles knows the difference: the former comes with numerous benefits, including expanded legal protections and maybe even a dental plan. The latter does not. And if you’re a professional driver, whatever camp you fall into could also define who’s footing the bill for things like gas, maintenance, and car repairs.

That very issue came up when a Florida-based Uber XL driver named Darrin McGillis got in a car accident last spring and could no longer pick up passengers. While dropping off a passenger, a scooter rammed the vehicle he himself had purchased. As BuzzFeed reports, McGillis tried to get Uber to pay for the repairs with its insurance, and the company told him it was actually his problem to deal with. Out of work, he decided to file for unemployment with Florida’s Department of Economic Opportunity, claiming that Uber’s job description and modus operandi for drivers shared more in common with Internal Revenue Service’s definition of an employee than an independent contractor. The state agreed, marking McGillis eligible for unemployment.

The difference between employees and contractors makes legal issues murky

If upheld, Florida’s decision could signal trouble for it and other companies that want a large, dedicated, and sustainable workforce without the overhead associated with traditional transportation companies. Under the current arrangement, Uber doesn’t need to worry about paying for things like paid time off, healthcare, and things like Social Security and Medicare taxes. That’s allowed it to hire an army of drivers that can be scaled up or down at a moment’s notice, but also created situations like McGillis’s where things get murky when there’s a snag. A potential driver for Uber may also take a look McGillis’s situation as a cautionary tale, and factor it into whether they want to work for the company.

In a statement provided to The Verge, Uber said it disagreed with the decision and planned to appeal it. A separate report on McGillis’s case in The Miami Herald notes that Uber has managed to get similar state decisions overturned.

While this is just one specific case, and only in one state, the battle over the independent contractor versus employee designation has been underway for decades, and extends beyond ride-sharing companies. It’s been a long-running issue at FedEx, which operates with a similar contractor setup with its ground delivery drivers. That’s brought class-action lawsuits, and efforts to change state laws to put liabilities on the companies. Uber and its competitor Lyft now face similar suits over whether the two companies (and others like them) should be footing the bill for things like gas and vehicle maintenance.

The science behind Netflix’s first major redesign in four years

2015-05-22T11:32:11-04:00

Back in March, a developer named Renan Cakirerk wrote a small piece of code that made a big impact on Netflix. Cheekily named “god mode,” it addressed one of the most annoying aspects of trying to use Netflix in your browser: scrolling through the company’s ever-growing list of movies. Once enabled, it would simply give you one, big list. Instead of sitting there, holding your mouse in anticipation, you could simply find the title you wanted and get on with watching.

The web lit up with stories hailing it as a crowning achievement of little-guy ingenuity. TechCrunch called it a fix for a “wonderful problem.” CNET dutifully noted that it “saves you from the slow monotonous horizontal scrolling.” Mashable, meanwhile asked, “Are you listening, Netflix?”

Inside Netflix, though, god mode was old news. In fact, the company had already tried it out on thousands of unsuspecting users years earlier. And it was a total disaster.

“Who the heck has time to go through every title?”

“It’s the difference between what people say they want, and what they actually want,” says Todd Yellin, Netflix’s VP of product innovation. “Consumers say they want to see every title in a catalog, but who the heck has the time to go through every title?”

Netflix began testing it on small portions of its user base — typically new users, so as not to confuse existing ones. It was as if they’d been given a 60-page restaurant menu. Instead of making people watch more, they ended up watching less. “We’re guessing that people were just overloaded,” Yellin explained to a group of reporters at Netflix’s Los Gatos, California headquarters this week.

But Cakirerk wasn’t wrong in trying to improve that aspect of Netflix, and Netflix wasn’t oblivious to the fact that its old system was just plain bad. In fact, a fix for the issue is one of a few subtle but substantial changes people will notice when they start using a new version of Netflix.com in browsers over the next few weeks. The company began rolling that update out to some of its users this week, and will have it out to everyone else by early next month.

At first blush, the new design doesn’t seem markedly different. It’s got a black background now, to match what people are used to on their phones, tablets, and TV sets. Those little arrows that scroll through the service at the speed of a glacier are still there, but now they jump between entire rows of choices. And the service does a much better job of letting you see information about a show as you click around, instead of accidentally playing something you only wanted to know more about. Under the hood though, the changes are the culmination of years of research aimed at gleaning every nuance about how humans hunt for things to watch. Netflix has been tossing out breadcrumbs in various configurations, and seeing how we gobble them up. This is the newest handful for us to taste test, and it comes with the hope that we’ll feast.

To aid in that ongoing psychology project, Yellin and his team are gathering and analyzing a mountain of data, then twisting the dials on what we’re all seeing in real time. By the time you see the cover for the next season of House of Cards, it likely will have already gone through several rounds of virtual focus groups to see which design drew the most intrigue. And you yourself could be an unknowing participant in countless other tests designed to get you to watch more, without ever making you feel like you’re being led astray.

“It’s not good enough just to see what they click on, because you can show the most sensationalistic thing: show a lot of sexiness on the cover, even though there [are] no sex scenes. Then they’ll hit play, but they won’t watch it,” Yellin says. “We’re not just looking for clicks here, because that’s not a good metric. We’re looking for finding the right people to watch the show, because we want to promote our shows to the right people who will actually play it through.”

Netflix has snuck fake shows in research emails

Changing covers is one of the least crazy things Netflix is doing to gather data. Last year, the company sent a dozen researchers out to conduct more than 1,500 interviews in people’s homes to understand how they use Netflix. They also sent out 15 million email surveys, sometimes with fake shows and star ratings to determine which is more effective in getting someone to watch. The exercise has the potential to tell Netflix any number of things, like if people really care about half star differences in ratings, down to if they’re still checking their email.

Yellin says that while Netflix is paying attention to what people ask for in these surveys and feedback requests, it ultimately spends far more of its energy watching what they’re doing on the service. “Most of our personalization right now is based on what they actually watch, and not what they say they like,” he says. “Because you can give five stars to An Inconvenient Truth because it’s changing the world, but you might watch Paul Blart: Mall Cop 2, three times in a few years … so what you actually want and what actually say that you want are very different.”

The stakes in all these experiments are high. There has never been more competition in online video; between HBO Now, Amazon Prime Instant Video, and Hulu, retaining customers has never been more challenging. Netflix’s goal is to keep its 62 million existing users there month after month, while also converting everyone who completes a free trial into a paid subscribers. By changing just one aspect of the service across groups of as many as 100,000 new members, Netflix tries to spot differences that get them to pay. Which should make it clear — Netflix’s new design may be the culmination of past research. But it’s also the beginning of another test.

Inside the labs where Netflix is trying to make televisions suck less

2015-05-21T13:17:50-04:00

In a nondescript room in the center of Netflix’s headquarters, you can walk into a small, paper white chamber called the Shu. It’s an homage to the solitary confinement cell in Orange Is The New Black, one of Netflix’s most popular original series. And just like the Shu in the show, Netflix’s version is a place of punishment — only of TVs, not humans.

The Shu is actually a faraday cage, a shielded enclosure where outside signals can’t make their way in or out. It’s here where the company tests the latest TV sets as part of a new program that rewards TV makers that have designed sets on which Netflix shines. That means things like turning on quickly, and remembering what show people were watching the next time they open the Netflix’s app on the TV.

The designation program was unveiled back in January, but it’s only now beginning to make its way into this year’s crop of TV sets, something the company showed off to a small group of reporters during a day-long tour around its headquarters in Los Gatos, California, yesterday. Right now, only about 20 sets have been blessed with the designation, a figure Netflix hopes will grow throughout the rest of this year. Looking ahead, Netflix also plans to make its requirements a moving target, not only to keep up with new TV technologies, but to help push the ones it thinks should become new standards.

Getting Netflix’s seal of approval is not a simple affair. The sets need to pass certain benchmarks, many of which are based on speed. For instance, Netflix’s app needs to start up within a certain amount of time, as do its videos once you hit the play button on your remote. Netflix is also counting how long it takes when you come back to its app after doing something else on your TV. Does it take too long to get back to your show after jumping out to watch sports or catch the local news? Better try harder next year.

While those are examples of software, Netflix is also judging TV makers on their hardware. Its program requires TVs to turn on instantly, and it rewards sets that come with remotes with a dedicated Netflix button. Netflix considers these things necessary additions to help bring TV watching in line with what modern viewers have become accustomed to on mobile devices like smartphones and tablets.

“A lot of this was inspired by innovation that was naturally happening in the phone and tablet space,” says Brady Gunderson, who is Netflix’s director of product development. “When I turn my phone on, I never really turned it on because it was never off — it just comes right back where I left off, in the app I’m on. If I’m reading a news article, I’ll be right back to that article. TVs, meanwhile, when you turn them off and turn them on again, there’s some time to reboot, they lose all context, and the network has to come up.”

“They weren’t purpose built for internet TV,” he adds, saying that all that’s changing this year. “There’s some quantum leaps in terms of improving that experience. We think of this as the year the smart TV got smart.”

To illustrate how quickly things have progressed, Gunderson — who is joined by his colleague David Holland, a director of business development at Netflix — stage a competition between last year’s crop of smart TVs to this year’s, some of which have been designated with the new standard. In each case, the newer set handily beats the old one on things like turning on, locking onto a Wi-Fi signal, keeping users from digging to find the Netflix app, and eventually starting up a show. In one particular matchup between Sony’s old Bravia TV set and its newer one running Google’s Android, we were already watching the opening credits of a film while the older TV had only just put some text on the screen to let us know it was warming up.

The exercise is painful to watch: the older sets lag and struggle by comparison. Gunderson is careful to say that the program isn’t meant as a threat that people should just give up watching Netflix on older TVs, or set their busted old 4K TV from 2014 on fire just to put it out of its obsolete misery. “The message is not that this is bad,” he says, pointing to the older set. “It’s just about how much better these newer TVs are.”

Much of the improvement has centered around a real sea change in the software TV makers are using. New in this year’s TV sets are things like Mozilla with its Firefox OS, and LG’s WebOS. There’s also Samsung with Tizen (which really only this year became ready for the world), and Sony embracing Google’s Android TV platform.

Netflix’s recommendation could carry a lot of weight with hardware and software makers. It commands a paying army of ravenous viewers 62 million strong worldwide, and which is set to grow as Netflix expands into new territories like Japan. Not only is the company bringing its customers movies and TV shows from studios, but it’s also making its own programming that can’t be had elsewhere (unless you’re willing to wait for the DVD version, that is). And some of those shows are being shot in 4K and with high-dynamic-range, things that TV makers like because it means that people who weren’t even thinking about buying a new TV set a year or two ago now might suddenly want one to experience these things.

While the side-by-side tests seem like a remarkably clear-cut way to test things, it’s not very scientific. That brings us back to the Shu. New TVs are bought from stores once they’ve been released, a process that like Consumer Reports’, involves sending out employees to make the purchases unbeknownst to the TV maker or any particular store that might try to weed out dud units. It’s then brought back to the cage of horrors, and run through a simulated home Wi-Fi network (they don’t do wired tests since most people with these TVs simply hook them up over Wi-Fi instead) that’s been throttled to match up with both good and bad speeds that can be expected in homes in America and elsewhere.

The entire recommendation program was born out of what Gunderson says became a common, and often bothersome question from people at barbecues and other social gatherings. “Because you work in TVs, they say ‘what TV should I buy?’ My answer was always ‘don’t buy one this year, wait until next year ‘ and it was never quite there. [Now] I do feel like it’s there — buy a TV this year.”

That may be a fair assessment, but even some new model sets from major TV makers have not managed to make Netflix’s cut for one reason or another. Of the aforementioned requirements (things like the fast start and fast resume), Netflix is looking for at least five of seven criteria. That gives manufacturers some wiggle room to be approved, which Netflix points out does not cost set makers any money or require a fee for them to advertise their designation. But there have been cases where even high end, flagship sets from some TV makers are not approved. Right now that notably includes Samsung, which is the top TV maker in the entire world, along with budget set maker Vizio, neither of which have had any sets approved. Gunderson mentions both of these examples, then points to a small, inexpensive TV from HiSense that’s sitting in the Shu. It’s just been added to Netflix’s recommended list, despite its low cost and lesser picture quality compared to the sweeping Sony and LG sets mounted to the walls in the other room.

Netflix’s lab is outfitted with common consumer routers that have been throttled to simulate what people have in their homes.

That brings up one shortcoming of Netflix’s big thumbs up for TVs: it’s not looking at things like image quality, sound, ease of use, ecosystem, and myriad other factors that can be really important when you’re buying a new TV set. Instead, Netflix’s program has been designed only to address what the company believes are things that make using its software on those sets the best possible experience. If you don’t even plan to use the TV’s built-in Netflix app, and believe you’ll use it through a set top box, or HDMI stick like Roku, then this will have little impact.

But it might not stay that way for long. Netflix has already begun pushing the technologies it’s using to present its programming. First it was 4K video, and now its high-dynamic range (HDR), technology designed to make the image on screen look more realistic, with brighter brights and darker darks, all intermingling in a way that is more pleasing to the eye. Holland says the company is incredibly wary about requiring these features in its evaluations given how new they are.

“We have a principle that in order for something to be included in the Recommended TV program, it needs to impact most of our members most of our time,” he says. “For the same reason that 4K is not part of the spec today, [HDR is] being limited to very high-end TVs and there’s very little content available.”

Two things that are likely to be part of next year’s requirements? More stringent requirements for speed, and making sure that users can easily return to Netflix if there’s an interruption. “We think the TV instant on feature could be faster. They take several seconds now,” Gunderson says. “Next year, we think they could be that much faster.” It’s a similar story for when you have to take a sudden break from watching something. “Someone’s calling you downstairs and you turn the TV off, ‘did someone call me?’ [you ask] No. You go to turn the TV on, and the app comes back to snow, or live TV, or your row of apps.”

The unspoken genius about all this is that Netflix is very carefully pushing TV makers to design what it believes to be the perfect Netflix experience while avoiding the need to become its own hardware maker. That’s something the company briefly contemplated in 2006 in the early days of its streaming service, before realizing it was better to focus on making software for computers and mobile devices made by others. While no money is being exchanged in this new program, the benefits are being pitched as a two-way street: TV makers who play ball might snap up sales from consumers who trust Netflix and believe they are getting what Netflix deems an optimal experience. Netflix, in turn, has made sure that the buyers won’t get so fed up with the TV software that they give up trying to use Netflix. In fact, if Gunderson and Holland are doing their jobs right, Netflix will be the best part of using your TV.

QuizUp is trying to reinvent itself by turning into a social network

2015-05-21T08:00:02-04:00

QuizUp as a place to dig into coffee culture? It is now.

One of the strangest things any company can do is suddenly shove a social network in your face. It’s even worse when a service you’re used to suddenly gets socialized in a way that is unavoidable. A recent example of that is Google requiring a Google+ account for anyone to comment on a YouTube video or sign up for a new Google account. It seemed arbitrary and artificial, and the company’s at least partially relented on the rule for people who were making new accounts. But what if the social networking features that were seemingly tacked on ended up being the thing that kept people coming back?

You can now play on your computer too

A year and a half into its existence, QuizUp — a trivia app that lets you compete with strangers from around the world — says that’s just what’s happened. As a result, it’s introducing social networking features to keep users around after they’re done playing. It’s also laying the groundwork to expand beyond trivia into other games. But for starters, it’s debuting profile pages, the ability to follow other users and topic pages around quiz genres that feel more like tiny Reddit communities than something associated with casual gaming. QuizUp is also launching a desktop version of the service that you can play inside your web browser.

From the outside, QuizUp looks like one more faddish game that grew quickly and raised a ton of money only to fade into obscurity, forcing it to come up with this new strategy. In 2012, a Pictionary-like game named Draw Something rocketed to 50 million downloads in less than two months, and was snapped up by Zynga for $180 million. It promptly fell out of the App Store charts, and soon Zynga’s stock price had lost half its value. More recently, King.com went public on the strength of its signature hit, Candy Crush. Investors are so skeptical of its future that King is now the fourth-most-shorted stock on the New York Stock Exchange.

As a private company, QuizUp has a bit more room to maneuver. According to the company, its move into social networking is simply a reaction to the way QuizUp‘s users were already behaving. Fans have started spending as much time chatting with one another and on the service’s group discussion boards as they are challenging one another to rounds of trivia, which only run about a minute to a minute and a half in length. Those places that were once envisioned as a side feature have morphed into the places where people gravitate. Now the company wants users to stick around even longer — even if it’s not playing the company’s games.

Facebook is “kind of creepy.”

“When you think of the social networking landscape, there isn’t a good place where you can meet new people. Facebook is all about connecting with existing friends, so it’s kind of weird when somebody [you don’t know] adds you — it’s kind of creepy,” says Thor Fridriksson, QuizUp‘s founder and CEO. “Here, you can safely get to know new people, and we think doing that through interests is the way to go. We’re trying to create communities where people play each other, and interact with each other.”

This is Kimberley, another player I got matched with based on the two of us liking similar topics.

Fridriksson likes to point to one instance of this happening when two users from completely different parts of the world began chatting with one another after trading victories in Lord of the Rings trivia. They got to talking to one another and eventually decided to meet in real life. They later got married. “For the new QuizUp, we thought, ‘Could this become something more than a popular trivia game?'” Fridriksson says. “We could use shared interests like games and challenges as an icebreaker, and make it easy to connect to each other.”

To help users begin exploring the new features, QuizUp is taking people’s existing topics of interest and putting them in a feed. If there’s a new item posted to one of those topic pages, you see it in the home feed. The same goes for if you start following people.

Trivia could soon be just one of several games

The new platform will allow QuizUp to expand beyond quizzes and into other types of entertainment, Fridriksson says. “The trivia itself is an evergreen thing. We will start to add more game modes, and other parts of entertainment connected to the topics,” he says. “I think that what we want to do, and our vision going forward, is connecting people with shared interests. We have built a platform we can iterate on.”

That could eventually include things like a Facebook-like news feed, and what Fridriksson only referred to as “procured, good quality content.” The company’s also ditching in-app purchases inside of QuizUp in favor of creating sponsored trivia and topic pages that Fridriksson says everyone will be able to use, instead of just allowing people with disposable income to improve their score (the company previously sold experience point boosters to users, which ranged from $1.99 to $5.99).

QuizUp became an overnight success in 2013, gaining 1 million users in its first week. At the time, its parent company, Plain Vanilla Games, had just 12 employees. In the year and a half since, it has raised $27 million and increased the size of its team to 90. Most of those employees work in engineering or moderation — user profiles were overwhelmed with lewd photos after they launched. Moderation will only become important as the company invites users to share photos and links.

QuizUp’s popularity ratings have plummeted

While Fridriksson says it still manages to bring in about 30,000 new players a day, QuizUp has rarely broken into the top 500 most-downloaded apps on iOS this year. Meanwhile, fellow trivia app Trivia Crack has been a top 10 mainstay for much of the year. Fridriksson insists that it’s unfair to compare the two given QuizUp’s move into social networking. “There are lots of cool trivia games, but Trivia Crack is just like the opposite to us,” he says. “It has five or six categories. They’re not about communities, and not about getting to know people. But [it’s] a great game, and there’s absolutely room for more than one trivia game.”

Whether QuizUp’s network features will help it climb back atop the charts remains to be seen. In the interim, Fridriksson believes the most important challenge is keeping people coming back in a way that is not annoying or scummy. That’s been a tall order for just about any game that are not named Minecraft — even some of the more well known ones.

“You don’t see everyone playing Angry Birds. In some ways, that’s why we’re making the change,” he says. “We want to be more of a daily habit.”