III. Technology trends we are thinking about
I’m typing this letter on a big-screen desktop. Next to it is a digital frame that cycles through the last 100 photos I’ve posted anywhere online. At home this morning, I edited a few pages on my notebook computer. In a few minutes, I’m going to walk to the subway, listening to RadioLab on my phone; my hope is that I won’t receive a text from the Twine box that indicates alarming changes in room temperature at my place outside the city. Once aboard the 2 train, I’m going to read today’s Digg stories on a 7-inch Android tablet. Later this evening, I’ll catch a movie on my iPad (my wife having hijacked the Roku), right after checking my Nike app that shows my physical activity for the day, and checking TweetDeck to catch up on activity from the people I follow — and even from some objects and things I follow, like Tom Coates’s house.
As recently as 2010, the midpoint of the Obama presidency so far, my current day-to-day arsenal of connected devices would have sounded both a bit fantastical and absurd — who needs two computers, two tablets, and a smartphone? Now it’s boring, even a cliche, to paint that picture. Connected devices are everywhere. Indeed, if there was any news at this year’s Consumer Electronics Show, it was simply that every device on offer — every TV, refrigerator, thermostat, fork, teddy bear — was Internet-connected.
The proliferation of connected devices is no longer that interesting; what’s interesting is how the rise of the contextual Internet, ambient data collection, connected devices, user navigation innovations, and competition for user attention on the stack are all accelerating and evolving in multiple ways simultaneously.
A. Rise of the contextual Internet
For more than 50 years science fiction has painted a future stuffed with computers that understand context. From HAL to the Cylons to the Enterprise’s sick bay, sci-fi computers are aware of the basic context in which they operate. Yet today, if you stand in New York’s East Village, type a search into Google Maps, and you happened to be in another city the last time you opened your mapping app, it will initially default to your last known location. Every time you use it, the app has to run a bunch of often sluggish operations to figure out where you are; until then, your search for Greenwich Street will probably produce directions to Greenwich Street in San Francisco. WTF — #$%! Your phone isn't so smart about context today, but this is finally changing. It’s a huge shift because it’s going to change how we will use our devices, the services available on them, and the nexus of navigation to those services.
Remember Dory, the fish in Finding Nemo who remembered nothing? Or the old Start menu in Windows? Both were endlessly forgetful. The old Start menu remembered nothing about the previous state of its user. Quite the contrary, it placed that work on the user: tell me where you want to go today, tell me where to start. And so every day you started over.
Today, your phone is capable of understanding your state, location, and movement, and can collect and analyze ambient data — where you have been, how often you’ve been there, how long you usually stay, what searches you have conducted when you are in this location. With that data, the phone can get really smart. Now extend contextual understanding to networked devices, small single-purpose devices — Nike FuelBands, Fitbits, NODEs, Twine sensors — and beyond to wearable devices like Google Glass, where the computing experience not only is integrated into our current context but actually mediates our experience.
The whole world of how we use devices, the whole Start and search and navigate experience, is going to be redefined by a combination of contextual information, ambient data (e.g., usual activities in this location) and user-defined information and permissioning around state, opening up new ways for us to use and experience computing. In a talk at betaworks last year, Robert Scoble discussed contextual data and how and why latent or passive personalization works, noting that one size or one Start menu doesn’t fit everyone. The world is on the cusp of a huge change as this contextual data layer emerges.
There are several pieces to this emerging puzzle. Begin with data collection. Data is being gathered today via our phones and tablets; in 2012 we crossed an important threshold where improved battery management techniques, along with upgrades to the mobile OS’s, made latent data collection viable for apps on mobile devices. Apps like Highlight can gather basic location data, manage state, and match them to social graph datasets to offer users features like “Who is near me right now whom I might want to meet?” Moving beyond our phones and tablets, data collection is set to explode with the proliferation of uni-purpose networked devices and latent data sensors. Cisco estimates there are 8.7 billion sensors and devices connected to the network today, growing to 25 billion by 2015. With some different assumptions and definitions, IBM estimates the latter number will be closer to a trillion. Regardless, it will be a big number. These devices will feed data into the network and directly to our mobile devices via Wi-Fi, Bluetooth and other wireless protocols, offering users, the devices and app developers vast new stores of data and insight beyond geolocation. There will be enough real time and historical data to be able to ascertain individual intent with some degree of confidence; yet this data will be uneven for a long time to come.
Let’s drop out of these abstractions to a few examples. Maps are an easy and useful starting point.
Maps today look online very much like they used to look when they were printed on paper. Yet the reality is that there is an entire world of depth behind almost every object on a digital map. Take a look at this image:
It’s from a blog post last year outlining how these data layers are developed and managed by Google. You can see the satellite image but tiled on top of that you see the traffic flow, and key traffic signs. It’s worth looking at this for a second and thinking about the data layer and how it adds a totally new dimension to the map. Take direction of traffic, for example. A single ping from a sensor or phone doesn’t give any information about direction. Yet multiple pings let Google “see” the direction of a road, based on the sequential flow of traffic pings. These second-order, derivative data sets that emerge from large, dynamic datasets are fascinating. To be sure, the data sets aren’t even in quality or utility. The more densely integrated with active and latent sensors a space becomes, the richer it becomes in data. In many spaces, like cities, “data depth” is expanding at an exponential rate. The resulting possibilities are amazing. As Alexis Madrigal put it in his post:
“It's probably better not to think of Google Maps as a thing like a paper map. Geographic information systems represent a jump from paper maps like the abacus to the computer. "I honestly think we're seeing a more profound change, for map-making, than the switch from manuscript to print in the Renaissance," University of London cartographic historian Jerry Brotton told the Sydney Morning Herald “That was huge. But this is bigger.”
Another simple example of these secondary effects is the “data shadows” that emerge. Sticking with New York, take Foursquare as an example. In the early days of Foursquare the company was challenged to find an accurate data source for opening hours of restaurants and bars. But the team realized that their own data set offered them a highly accurate data shadow: their own users’ check-ins could reveal exactly when an institution was open or closed. When summer hours switched to winter hours, Foursquare was the first to see it. This is an example of the unintended uses that start to emerge from these very large data sets. More will emerge: take a click around this interactive map of foursquare data. Bitly generates a lot more of these; consider a few that Hilary Mason, the chief scientist at bitly, has observed:
“People in Brooklyn are more likely to read about food than people in any other part of NYC. People in Manhattan are more likely to read about business. People in Queens are more likely to read about sports.
The top Wikipedia article of 2012 was "Lunch"
People who read about fashion read about physics, but people who read about chemistry don't read about anything else (which is behavior we see in two other categories — religion and adult).
Sports was the one topic for which people actually only want to consume fresh information. Every other topic was roughly equal, with religion being the slowest.“
Or take something more radical. The Google self-driving car isn’t really a marvel of automotive engineering, rather it’s a marvelous application of existing Google data layers, matched with incredibly fast analysis of from onboard real-time sensors — it’s the data, stupid. It captures a big potential use of layered data, while simultaneously illustrating that, as with the Internet, many of the uses and applications of contextual computing will become evident only once the platform is in place. Uses of these data sets will be wide-ranging and significant. In one study it was estimated that government in the UK could save somewhere from £16 to £33 billion a year from the use of large-scale layered data.
As autonomous human beings, we hate the idea that our lives and choices are predictable, but it turns out they mostly are. A few years ago the Chief Marketing Officer at American Express asked me how long I had had my Amex card; I told him several years. With that depth of data, he wagered he could tell me where I was going to have dinner that night. His ability to predict was based on analyzing a time-series of transaction information, and is therefore limited to actions and activities that I’ve done before. That kind of predictive analysis is interesting, but it’s yesterday’s news. The new world we’re entering is different. I’m talking about a layer of data that exists over reality, one that is real-time and whose signals are highly diverse and redundant. One that has history, one that learns, one that can ascertain intent. Combining the data layer with simple prompt-based navigation, it becomes possible to tell a person exactly what (or even whom) she is going to want to know in a particular place at a particular time, before she even forms the thought. If you’re an Android user, you can get an early taste of how this kind of layered data from multiple sources works and how it can anticipate intent, through the Google Now app. In 2013 I hope to see some of our companies do fascinating things in this area.
Following are some interesting places to look for and think about innovation emerging out of this contextual data layer. I mean to frame these as broad zones; in each of them, there will be multiple opportunities for large-scale innovation:
i. New devices connecting the analog world to the network
ii. New Interfaces
b. Voice navigation
iii. Alerts and notifications
iv. New data and intentionality
i. Opportunity: New Devices
We are starting to see a rash of specialized-but-flexible devices to collect and disseminate data. Here again, we are in the earliest days of a Cambrian explosion. Want a networked fork? Pretty much anything you can think of will be connected to the network. There is a clear advantage that will accrue to these simple, single-purpose devices that we call “miniware.” The device represents the function, there is no need for an abstracted layer of navigation. We have seen mobile apps so simple and elegant as to be single-touch help buttons for our lives; yet miniware needs to be even more functional. If you are interested in connected consumer devices, head on over to Grand St., a marketplace for miniware.
Beyond the sensors are the devices we integrate into our lives that help us navigate and manage this contextual layer. These are the hub devices — today, that means phones, tablets and routers — that handle connectivity, navigation, and/or storage on behalf of our other connected devices. This hub-and-spoke model is a new, emerging local version of the classic client/server architecture.
Before delving into the coming changes in software driven by this new miniware architecture, let’s touch briefly on the commoditization of our hub devices. Let’s take the most sophisticated one in our current arsenal: the tablet. If you are interested in buying tablets in bulk, head over to Shenzhen where you can find a 7-inch iPad clone with capacitive touchscreen, Wi-Fi, two cameras, a speedy processor, and reasonable storage all for USD $35 per unit. Or even more disruptively: the Aakash tablet, which the Indian government is subsidizing from a $45 unit price to get into students’ hands for $20. It’s astounding how fast complex hardware is being commoditized. It is reasonable even to argue that price points are heading to zero given the migration of value to the underlying services.
Our experiences are going to center on our hub interface devices, not the sensors and other specialized devices that revolve around and communicate through them. The rise of new hubs will require some serious adaptation for some industries. For example, many businesses view the TV as a hub device in the home. It’s not. The TV is a second or third screen in our homes — in the best scenario for the consumer electronics companies it’s a source of rich data, but in the more likely scenario it’s just another terminal. The car is another platform that businesses often regard as a hub, as a primary user interface. It too is likely not. (At CES this year, Marc Benioff provoked this question and suggested the car might be an app.) There will emerge other hub devices, but they will likely be wearable, like Google Glass or a watch. One of the key properties of these hubs is that they have an interface that is increasingly tied to you. The iPod touch is more of a hub that the TV is. And Google Glass is going to be one of the most interesting hub platforms that will launch in 2013. If you haven’t delved into Glass, I suggest you watch this video. in which Diane von Furstenberg discusses how she expects Glass to affect fashion and integrate technology into our lives. While I wish the discussion went deeper into Glass’s potential, and the video were less marketing-heavy, it’s worth watching to see someone (who has built her life around a singular mission for women) describe how technology can integrate and serve that vision, and how technology will become a (fashion) accessory. Again, just like the Internet, its the unexpected applications of this data layer that will surprise us most. As Turing proved 77 years ago, there is no systematic way to determine what code or data will do. What you can bet on is that massive businesses will be born as our analog world becomes part of the network. As previously latent data, becomes accessible, readable, writeable in real time.
Let’s move up a layer and talk about the software on these devices, starting with navigational interfaces, and identify some of the areas where innovation will happen, where big ideas, big opportunities lie.
ii. Opportunity: New interfaces
a. Contextual: Over the past five years the start experience has been re-defined by stream-based navigation. In 2012, we saw the beginning of another re-definition as the notification stream combined with ambient data services to give us information that is timely and contextual. On the phone, the proliferation of apps, the abundance of storage, and the ease of download is together shifting navigation away from directories and menus and toward notifications and search. A new Start layer is emerging, and this layer is where developers are going to construct true contextual navigation. Instead of having to hunt and peck for a particular app, suited for a specific context, that information will be pushed to me, at the right time and the right place. As examples, try Highlight, Field Trip, or Google Now.
b. Voice navigation Voice is starting to emerge as a navigational interface. If you question the importance of voice, consider that Siri is the first non-"i" brand that Apple has launched in maybe a decade. While Siri's performance is still wanting, it is becoming part of vernacular — “Siri, tell me what is the meaning of life?” And Siri today is been complemented with editorial assistance, as searches like this by my wife suggest. And while Apple's voice service is getting pretty good for dictation of text (for SMS or emails, for example), Google has made even more progress in voice in 2012, having thrown prodigious measures of talent and computing power at the problem.
In the near term, while most of our navigation through contextual data layers will occur through the interfaces of hub devices (i.e, phone or tablet screens), voice is going to become a vital interface into single-purpose connected devices. You won’t have to keep pulling your phone out to find out why it’s alerting you to something. We are going to start seeing voice interfaces linked to embedded sensors throughout our everyday spaces — in wearable devices, in our homes and cars, on our doorbells, on our toys, on pretty much everything, be it consumer or manufacturing. And the rash of embedding these devices in our physical environments will accelerate the declining cost curves of the technology, in turn making it economically viable to embed voice interfaces and sensors into more and more objects.
c. Gestural interfaces Along with with contextual notifications and voice, gesture- and motion-based navigation is also starting to proliferate. The rapturous attention given to early innovations like Kinect, Leap Motion and Displair suggests that gestural interfaces will start to enter our lives. The Leap Motion demo got a lot of circulation; for something different take a look at Pinokio, a Pixar-like lamp at the University of Wellington in New Zealand. And then consider very simple gestural interfaces that respond to simple waves, as the Nest thermostat does. Or, lastly, check out this imaginative video by PrimeSense, which is producing the Capri, a cheap, miniaturized 3D sensor. Navigation of this data layer is going to get complicated and rich with innovation.
d. Humanized and zero-latency interfaces Over the past year, I have observed so many of my colleagues pick up and judge a new app initially by the criterion of the smoothness of interaction, how it feels to touch and swipe. As we integrate networked devices and their interfaces more and more into our lives, we want the hardware and interactions to feel natural. The technology needs to be humanized, and feel human. The interfaces need to be fast and the gestures natural. The services and products, hardware and software, that succeed will feel like seamless extensions of ourselves.
The importance of speed and zero-latency interfaces in this emerging contextual world cannot be overstated. Contextual data will often be relevant to me only for a very short period of time; getting it to me, via the right channel, on time, is vital. The Google team are starting to roll out knowledge search, search that uses structured entities to provide context to users in real time. From a discussion with Amit Singhal, you get a sense of how important speed and context is to the future of search as these interfaces become more and more integrated into our lives.
One of the questions I repeatedly ask myself in our business is: Where is the start state? What does the experience of turning on a device, and tuning in, feel like? Start state is still very much in the process of being re-defined. It’s no longer the Start button or finder menu, it’s not Google Search, it’s less and less the newsfeed, it’s not the app directory or screen icons. Will it be a contextual alert-and-notification interface, a user-defined agent interface, an augmented reality sidebar in glasses, Siri? The glib response is “all of the above,” but we aim to understand the relative value and importance of each of those components to the emerging contextual interface.
As navigation shifts toward contextual alerts and notifications, voice, gestures, and search, operating systems will lose the ability to drive distribution of default, OS-tethered applications. This is a lesson learnt by Microsoft. Once upon a time the Windows desktop was highly valuable real estate. During the 1990s, a shortcut on an OEM’s default Windows desktop configuration was regarded as a critical competitive advantage for websites. But as the browser and web search became primary navigation points, the relative power of desktop distribution weakened. Platforms are wonderful things, especially when they are open, but they can become rapidly commoditized as the non-native navigational layers on top of them become better and more useful than the platform’s own start state and metaphors. Apple iOS, for example, is clearly moving down that path towards commoditization.
iii. Opportunity: Alerts and notifications
Beyond navigation, the stratification of the alert/messaging bus is an area that is begging for innovation. Making this work effectively will be a vital building block for the contextual internet. Today, new devices and new sensors often have no option but to crudely hack their way to the top of your notifications stack. For contextual computing to become a part of our lives this has to change. We need fine-grained, contextually relevant interfaces for alerts — ones that tell me what I want to know when I need to know it, but don’t otherwise clutter up my already-overloaded consciousness.
An example: In the fall of 2012 I got a Twine device, a flexible, single-purpose sensor that was funded on Kickstarter. I set up the device to message me when it reaches a specific temperature (55 degrees): i.e., “Device, tell me when you are cold.” The problem is that the device uses the simplest form of messaging an alert, which is the SMS text. SMS, today, sits at or near the top of the notification stack for most people. Yet once a whole bunch of devices start to send text messages about worrying temperatures or low Christmas tree water levels or open garage doors, you are either going to turn off the devices or push their alerts far down in your prioritization stack, defeating the whole point of the devices.
Another example: Do you have any friends who send you texts instead of an email? They likely do this because they get an immediate answer, by pushing themselves to the top of your notification stack. Same for iOS or other badge-based visual notifications. In fact, once upon a time, it was true for faxes! This has to change, we can’t keep re-inventing new messaging layers to supersede the others; instead we need effective, multi-platform filtering along with smart passive alert management. Messages and content need to be accessible but also decay gracefully. This is hard but necessary, and in 2013 I believe we will see a better layer of alert management and notification infrastructures start to emerge.
iv. Opportunity: New data & intent
Contextual information will provide a rich strata of data, but in order to make them work for us we are going to have to find ways to map that data to intent. Determining intent has been difficult in most cases; when it’s not hard (e.g., intent-driven searches) it has proven hugely valuable for business/commerce. Yet intent is going to be more discernable in this world of contextual computing. Already today companies such as Socialflow are starting to use language to determine intent. Wedding this capability to the contextual data layer will, I believe, enable monetization to scale in a manner that we haven’t seen to date on the social web. Social media, in particular social media on mobile phones (where screen size is limited), is hard to monetize because it is devoid of data context and intent signals, and the existing tools to improve context (behavioral targeting, psychographic targeting) aren’t dynamic. Tools like Socialflow are starting to structure context in the real time stream and provide the ability to target offers and present highly contextual ads based on language. Take a look at this fascinating presentation the Twitter team put out re: second screen viewing and Twitter usage. The advertising section shows how Twitter users are connecting and discussing ads in real time. As a data link firms between the viewing platform (TV) and the discussion stream (Twitter in this example) you can see the opportunity emerge for intent rich, highly targeted — useful, interesting, advertising. Match that with location and the emerging data layer, you can see the beginnings of a big, big new business opportunity. (For more on this, see my recent discussion on Bloomberg about how Socialflow is delivering context into the real time stream).
Can we extract analogs from the four primary metaphors (pages, search, social feeds, and apps) for navigation that have emerged on the web? What can they teach us about this new data layer? What is the analog to an HTML link? A check-in? Or is that more akin to a click? What are the equivalents of web sites? Physical spots where we stop and spend time? While we have been building and dwelling in our networked world, we have been trying to absorb these metaphors into our lives. Over the past five years we have thoroughly deconstructed the two first navigational metaphors of the web (pages and search) and reformed much of how we think around real time streams and apps. As this new contextual layer emerges we are going to have to de- and re-construct our metaphors once again. It is going to be fascinating to see how we apply web metaphors to our physical space. Figuring out how to tell stories and discover things via contextual data layers is going to be an area of innovation and exploration.
Finally, in closing this section about the contextual internet it's important and fitting to add a point about openness. This contextual world, this contextual Internet, requires open interfaces to both data and devices. Services and products that are designed as islands will die; those that integrate with other systems will flourish. The logic is fairly simple, and familiar. We are creating a network of data, the value of which comes from its networked nature and accrues to those who participate in the network. Data islands will exist but they will, in all likelihood, be short-lived. Take a device as simple as the Swivl. Its value comes from its ability to connect to cameras or devices you place on it — without that it’s just a dumb robot, with that it’s a useful innovation. Fortunately, many device manufacturers are starting to realize that their devices have to be connected to be useful — it’s Metcalfe’s Law all the way down, and simple, open interfaces are the “cheapest” in every sense.
B. Default apps
Relative to Facebook and Google, Apple and Amazon are weak at app and web software development. This hasn’t affected Amazon’s business in a material way yet, but it’s starting to affect Apple. Apple’s default apps are regularly no longer best in class. From mail to calendar to books they have been replaced by many of Apple’s users. In 2012 I started using Sparrow for email, Google Maps for mapping, Chrome for browsing, Fantastical for Calendar, Brewster for contacts, Rdio for Music, Kindle for books and longform web pieces, and WeatherCube for weather. The only Apple apps I use on my iPhone are Phone and Contacts, Messages for SMS and Podcasts. My kids complain about the Apple iOS apps you can’t delete. And competitors — most significantly, Google — are moving fast, and succeeding, to build better replacements for the full suite of default Apple apps. In Google’s case, they go one step farther to leverage the value of their cloud and other networked state data. Coupled with that they are starting to introduce cross-app navigation that bypasses the Apple OS interface entirely. For more on this trend, see Liz Gannes piece, “The Year I Basically Stopped Using Apple’s iOS Apps.”, and Rich Greenfield’s “How Google Ties YouTube and Chrome Together Within Apple’s Ecosystem“ (yes, you need to register, it’s a broker dealer thing, and for Rich it’s well worth it).
As a zig to Google’s zag, Twitter is moving in the opposite direction, consolidating its experience into a single app and throttling third-party access. Yet it’s unclear whether and how much closed networks can prosper in this new emerging environment. Apple is the icon of a closed stack. Yet at the OS level, Apple’s platform is starting to fall behind, and it’s reasonable to expect Android’s advantage from its open software stack to widen its functional lead at the OS level in 2013. There are plenty of specific examples of this but most of them relate to device management features (e.g., do-not-disturb, parental controls, sharing buttons) that Apple has tethered tightly to the OS, so much so that there isn’t any way for developers to compete and offer better services. Match that with the fact that Apple has developed a vested interest in the app as container and with its High Street model for monetization (lots of successful small apps, few breakouts) and you have a system that is open on several flanks to competition and disruption. Track back over to Twitter, and as they consolidate the experience into owned and operated end points they are also launching Vine, a short form video app, as a separately branded experience, seemingly independent of Twitter, brand and integration wise. These inconsistencies reflect a reality based in uncertainty. Uncertainty re: mobile, uncertainty re: media types (eg: video) and uncertainty re: the extensibility of brand. This climate is offering a lot opportunity for startups.
Given the platforms that are in place today, the speed at which a startup can gain significant user traction is unlike anything we have seen before in technology or business. Developers can integrate everything from identity to payments in minutes. Users need invest no more than a few clicks to test and trial a new app or experience. Betaworks had front-row seats to an extreme example of this early last year when OMGPOP, a company that we seeded, launched a game called Draw Something that, as I’ve already noted, got 50 million users in its first 50 days. Then, for a brief period after its acquisition by Zynga, Draw Something’s fortunes flipped and the app reportedly lost a million users a day.
At the same time, many apps are integrating commerce and content into a super-simple flow that makes them essentially single-click experiences. When done well, these are the apps I consider “help buttons” for my life. And those are the apps I am happiest to spend money on. As Bruce Upbin observes: “[T]he mobile form factor drives a habit-inducing simplicity that will grab and take hold of more and more consumer spending.”
Yet despite all this, the most effective monetization models that are scaling in this world are commerce- or subscription-based. In 2013 we will see significant progress on mobile advertising. Facebook first and foremost has to demonstrate the breadth and scale it can achieve in its mobile advertising model. If Facebook can make it work and integrate its Open Graph so that the ~70% of auth’d-in users on iOS, and ~48% on Android get better-quality advertising than everyone else, then it could unlock a huge opportunity for itself, for publishers and for application creators. The recently announced Graph Search is clearly a big step in this direction. The potential is there and it's going to be fascinating to see its execution. If Facebook can make search a significant vector of navigation and use it to surface intent, it will become as central to the experience as the newsfeed and much more useful on mobile. However, today, it's unclear how useful and relevant the data set on Facebook is for search, and particularly intent-rich search; Facebook “likes” are fairly polluted as a signal and adapting dominant user workflow is never simple.
C. Personal device clouds
As noted above, concurrent with this shifting app world we are starting to see dramatic growth in small, single-purpose devices — miniware — coming to market. As I outlined above, many of these devices will hang off hubs for connectivity, storage, or user interfaces forming a personal device cloud.
For a long time, hardware represented a tall barrier to entry for startups. But several broad changes are conspiring to make today’s environment start to look very different. Rapid prototyping tools like MakerBot, flexible factories, visual communication tools like Skype and Google Hangouts, and collaboration tools like LayerVault have dramatically dropped the cost of developing beta products. Betas work, an insight that continues to spread memetically across industries. The relentless commoditization of processors, storage, memory, parts, and batteries has made it cheaper to build affordable miniware. Add in the new funding platforms (the various components of the emerging independent seed-stage stack I discussed above) and you get a rapidly growing marketplace of networked devices. A highly creative, product-centric maker community is forming, much of it here in New York City, around this new vision for hardware.
Let’s take a deep dive into one area of miniware that will see big changes in 2013: the TV.
The TV is not usually thought of as a peripheral device since the television has traditionally assumed such a hallowed position in our home and culture. Yet given this new device architecture, the TV is becoming just another device on your local network. Control of the TV and much of its data (e.g., your preferences and personalization) will come from our phones and tablets, not from the TV or its manufacturer. In 2013 there will be a massive acceleration in innovation as the TV collides head on with the Internet. Apple will strike a set of deals with the cable companies that will mirror its initial wireless deal with AT&T (my guess is Comcast and Time Warner out the gate). Apple will provide a box that is cable-ready and fully integrated so that it plugs and plays with any TV. There are arguments for and against Apple getting into the actual TV set businesses. I would bet it won’t, but even if Apple does strike a deal, it has to operate with an existing TV, given consumer expectations for quality.
On that box and its services, Apple and the cable companies will extend the data platform of the Internet into the TV screen, slamming together two massive businesses. It’s Apple and Big Cable, so expect the platform to be gated, with TV-based apps approved by Apple, similar in design and process to the iPhone. Concurrent with this data distribution business, the fight for multi-channel video (otherwise known as “the content that media companies are banking on that consumers still want”) will play out.
You can begin to see the outlines of this in the table below, compiled by BTIG from q data and showing the 10 biggest users of bandwidth in the US. It’s pretty interesting. Use of rich media exploded in 2012, with US bandwidth consumption up 190% for the year.
- Bandwidth usage is becoming somewhat less concentrated: the top 10 are using 5% less than they were in the spring of 2011.
- Netflix continues to dominate; if you are interested, dig into the Sandvine data: Netflix’s share of peak downstream viewing rose to 33%.
- YouTube gained a lot of share, up 30+% at 13.1%; consider the ancillary effects of that on the platform’s successful content creators, especially the channel providers on YouTube.
- Surprisingly Skype dropped out of the top 10; BitTorrent lost a lot of share: 17% down to 10%.
- And iTunes, which hosts video and music, streaming and downloadable, is still small, showing how much progress Apple needs to make. At 3.4% it is relatively insignificant.
- Facebook is very small at 1.5% and lost a fifth of its share in the past year; in other words, as its user base doubled, it lost bandwidth share, and thus has not grown its value as a platform for rich media.
These numbers reinforce that the collision between TV and the Internet is indeed producing a very different, and not yet stable, business landscape, where a new set of innovators can succeed both in content and distribution. Apple will enter this market. But it’s unclear whether they’ll be able to create a seamless experience for the user that is as controlled and as managed as their design and architecture demands today. Combine that with the proliferation of low-cost miniware connected through our hub devices and their interfaces, controlled by our smartphones and tablets, and we can see the outlines of a true personal device cloud.
D. Media and what storytelling in 2013
The world of media will be upended once again in 2013. As with other themes I have covered, there will be push and pull; this isn't a straight-line story of disruption. Rather, it's the tale of a massive existing creation and distribution system simultaneously evolving on its own accord and being disrupted from the outside. The broad shifts that continue to press against the creation and business of media include the transition of media companies to distribution over IP networks and the cloud, the decline of broadcast, the rise of narrowcast, the democratization of creation and distribution, the melding of previously distinct media types and forms, the rise of video over mobile, and the advent of new consumption technologies (everything from Apple TV to tablets to Aereo). But rather than focus on the technology let me use changes in the dominant forms of storytelling to frame this discussion.
Start with TV. Over the past five years we have seen an incredible resurgence of storytelling on television networks. Homeland, The Sopranos, Mad Men, Breaking Bad, BS:G, Friday Night Lights, Downton Abbey: we are living in an era of amazing storytelling on television. Why has this happened? Many predicted that with the rise of IP and collapsing windows of distribution, we would see shorter, cheaper programming — not higher-quality, long-form media. Instead, the expanded availability of scaled PPV and subscription models has unshackled long-form storytelling on television from the chains of a business model that depended on syndication and restricted the availability of archived content in order to sell network advertising in the future. Now something comparable is starting to happen with Internet storytelling.
Over the past five years Internet storytelling has become highly fragmented. As happened with TV, this was a function of the business model, or lack thereof, but it was also a consequence of the technology and the medium themselves. The web and its link structure gave rise to amazing new forms of media but it also resulted in a highly fragmented experience where everything was one click away from everything else. Writing or reading long-form media was hard when the temptation of a click to something maybe, possibly, hopefully more interesting was one finger touch away. All too often when I set out to read a long-form piece on the web I soon find myself three clicks away from the original article, having followed the thread of a hyperlinked thought pattern that’s interesting. And worth doing — but it’s not the focused, crafted experience the author labored to create. It’s no longer author-led. Attention became fragmented and the arrival of stream-based navigation (i.e., Twitter, Facebook, Tumblr feeds) only accelerated this dispersal.
The arrival of the iPad and the Kindle started to reverse these dynamics. Tablets, as well as phones, are far more intimate media experiences than desktop web. We hold them in our hands, we touch them and speak to them, they are impressive and they understand a bit about our orientation and movement; they are in a sense an extension, an element of our selves, as opposed to disassociated objects that sit on a desk at work or in a corner at home. While the bulk of apps on the phone or tablet today are simple containers of web content (80% by some estimates), they all must be developed, often at considerable cost, and distributed as encapsulated apps. And while the app format loses some of the benefits of web content (e.g., embedded linking), app creators and users clearly believe the benefits outweigh the costs. Steve Jobs's original and forceful conviction was that applications on the iPhone would exist only via web “apps”, as opposed to natively encoded apps; as it turned out, he was correct in his vision, wrong in the execution. While mobile apps are simple containers, they exist as discrete applications because as apps they can fully access the breadth of technology on the device and as apps they represent for users tightly designed, one-click functionality — that single-button "help" action I talked about earlier. The consequence of this is that mobile apps have created an entirely new canvas for creators and users alike, one that is intimate and immersive and well suited to storytelling. This canvas is now reinterpreting how we use the web.
Midway through 2012, we ran a small un-conference at betaworks called “Cleaning Up and Slowing Down”. The theme and discussion of the day was how to reform and reinterpret the web to create simple experiences that offer creators and users alike the possibility of longer-form, more immersive storytelling. I was amused to see that the team kicked off the day with the Gangnam Style video — the ultimate representation of fragmented, collage-based media. I saw this as a subtle acknowledgment that the emerging new long-form medium needs to embrace and extend the existing web – it’s a branch in the emerging tapestry, not a replacement – like a magazine was an extension of a book, or a... (Ahem.) Moving on.
This emerging new “slow” web ports the immersive visual expectations of apps to the web. Examples of sites that are doing this include Digg, Svbtle, Medium, Branch, Quartz, and tapestry. But you don't have to be a whole site to embrace this new form — even discrete features are starting to embrace this new clean web aesthetic. In December the New York Times released a long-form piece called “Snow Fall”; it’s a terrific example of emerging Internet storytelling. It offered a scrolling, unfolding visual experience akin to the tablet but matched with a reading experience that worked for users as well as the author. It pulled you through the experience even if you didn't read every word — you could flip through it, study it, and immersively experience it. Executive Editor Jill Abramson summarizes the results: "At its peak, as many as 22,000 users visited Snow Fall. ... Strikingly, a quarter to a third of them were new visitors to NYTimes.com. They were quickly hooked: users spent around 12 minutes, often more, engaged with the project. More than half a million visits went directly to Snow Fall, and more than half of those direct visits were from new Times users (or readers who hadn’t visited the site in a long while) lured to our journalism by this feature. In sum, then, people heard that The Times was doing something neat; they came in droves; and they stuck around for a while."
It’s interesting for a minute to dissect her comments. First, start with the content. She clearly appreciates what an innovative form this was (read the rest of her memo for further color). Second, she highlights that much of the distribution came direct. This highlights the power of the social web to end-run search and offer the ability for a large number of people to directly find something regardless of its domain. Third, she emphasizes engagement — the real engagement of readers/users spending time on the piece, and then sharing it. Engagement in excess of 12 minutes is very, very long for the web. Chartbeat has become the way people understand and measure this kind of engagement. And as Chartbeat is now serving 8 of the top 10 media — news, print, radio, TV — sites in the US, I can confirm from actual data that 12 minutes is a decidedly long time. Crossword puzzles average around 8 minutes. Abramson mentions pageviews, perhaps out of habit, but as the web and web pages become app-like experiences — essentially complex single-page canvases — engaged time will become the critical metric to understand, track and build against. Finally, note the emphasis on new visitors. If media experiences like “Snow Fall” can draw in new users who aren't frequent visitors to the Times site, it’s a big deal and well worth the cost of development.
Storytelling is part of what we do as humans, how we share, how we live, love and grow. Yet so many of the tools we have built over the last five years have fragmented one’s ability to tell stories, to develop narrative, develop identity and your sense of self. And your stories about self and others. That is changing.
At betaworks we have built and invested around storytelling in 2012 and we will continue to do so in 2013. We created platforms for understanding engagement like Chartbeat and on the creation side we invested in to be, Editorially, Checkthis, Svbtle, and Branch, among others. Likewise, the Digg version 1 that we launched in August was created very much in this vein — a single-page app, powered by both people and data. It’s designed to be a great single-user experience — a simple one in which you find that a visit to Digg will yield something wonderful, something interesting, something important, something curious you didn’t know. Many of the services I adopted in 2012 didn’t offer a step function increase in my productivity or workflow, but rather they let me create things differently. Tapestry is an example of this that we built.
Tapestry is a simple authoring platform and reading app for short, tappable stories. These can be read on the web or more often on the iPhone or iPad in an interface free of distraction. The inspiration for tapestry came from Robin Sloan’s Fish app and essay. I saw Fish and thought the content was great, and so was the medium. Six weeks later we had a working alpha of tapestry. We then worked with Robin to refine the authoring experience, to make it super-simple — the experience needed to be 100% about the content, not the tool. We launched tapestry into beta around Thanksgiving and since then a lot of wonderful stories have been created by bloggers, by writers, by kids, by cartoonists, by journalists. We think it’s a powerful idea that you can author — your words, presented just as you want them to be read, alone or with rich imagery — in 10 minutes on the web site and then immediately have that story appear as an immersive, distraction-free in-app experience. The emotional response that people have had to it has been fairly remarkable – take a read of some of them here. The phone and tablet need native, intimate, rhythmic reading experiences and that's exactly what we are building.
If you want to check out the tapestry experience, start with “The Italics” or “The Thank You War.” Then create your own.
In 2013 we will see these kinds of storytelling tools emerge and mature, encompassing everything from short stories and personal narratives to brands and advertisers. Advertising needs new storytelling tools as much as content creators do. While the word “native” already sounds like a tiresome buzzword to many, its ubiquity points to the need for new narrative and storytelling-centered advertising forms.
At betaworks we spent a lot of time over the past five years building tools and investing in the creation of the real-time web and the data tools to support it. These new storytelling platforms are a complement to that real-time stream. I always believed that the real-time stream was primarily a new navigational tool — a flow, not a destination. The destinations are the endpoints in the experience, the places you drop into. Some of them are transient and require little immersion, some of them require more. In 2012 this framework took a few big steps forward and some number of smaller steps backwards. The steps forward were around the creation of these new immersive authoring and experiencing tools, the steps backwards were about the stream platforms attempting to redesign themselves as destinations, a place for active consumption as opposed to navigation. We will see how this plays out in 2013, but I suspect this time next year the stream will continue to be primarily a navigational experience. It runs against its grain to be a destination, and individuals all have a conflicted relationship to the next hit of new information — on one hand, we desire and crave it, on the other, we become bored and exhausted by it when it starts to overtake our available time, crossing the boundaries we want to set in our lives. If the stream is the destination then its shallowness will start to work against it. It will seem thin and meaningless. Its depth comes from its links out, not its integration within.