I believe that in a few hundred years, the introduction of the iPhone in 2007 by Apple will be considered the beginning of technological singularity. iPhone was the first device that successfully brought computing, practical and extensible computing, out in the hands of people. Having a device in one’s pocket, capable of searching the internet anywhere and anytime, ushered in the age of information. Intelligent and contextual information has ever since changed our perception of the world. And this is just the beginning.
Today, in 2013, information is still query based. Its flow is determined by curiosity and the tools chosen by the user. It’s inefficient and deprives people of information about their world that they might benefit from. This inadequacy is widely recognized and efforts to better the availability of information are certainly underway. Besides countless experiments in artificial intelligence, contextual search implementations like Google Now are a glimpse into a future with frictionless flow of information. I want to talk about the enabler of this vision: Context.
Context in the domain of machine learning and information is the idea about the circumstances that precede a query and follow the results. Context, like user location, allows a better understanding of the query and facilitates more relevant results. But that is a very limited use of context.
Technology is an extension of the human self, a bicycle for the human mind. Technology is a sense and contextual information, information that is beyond the perception of the 5 physical senses, is its input. But to get there, information has to be freed from the bounds of human generated queries. We need to understand our environment enough that this understanding can be used, not only to generate relevant results but also, to create relevant queries. Our devices need more context to ask questions about our world themselves and thus, populate the information we need without our conscious intervention.
In order to ask the right questions, our devices need to know us. The most essential aspect of context is the user, us. We humans have a state of being and a set of preferences. Our physical state, our mental pose, the state of the environment around us and our interactions with it are just a few of the data points that essentially differentiate every one of us. Pieces of information that are relevant to some are pointless to others. The first step towards better context is a better understanding of the users. Better understanding requires more data.
The rise of the sensors
Sensors are little widgets which make our devices aware of their surroundings, their users. The iPhone can sense ambient lighting conditions, proximity of the user to itself, its location, orientation and motion relative to the planet. The iPhone does not know enough. Proper context will require data in many more dimensions, dimensions which can define the user and his state. Their sheer number will placate strenuous constraints on their size and energy consumption but demand ever more accurate data. It is not difficult to realize that sensors will ride the next wave of innovation in computing. These minuscule and sophisticated apparatus will usher in the age of personal data.
The age of data
Billions of sensors will wake up thousands of times everyday to take a peek into the world they are in and generate massive, massive amounts of data. But data is not context. its interpretation is. Interpretation is the tallest step to context. It is an effort to make sense of the collected data and what an effort it is.
Algorithms are the tools used for interpretation of data. An algorithm is a logical process to solve a problem computationally. In the world of machine learning, human understanding of nature and behavior is coded into algorithms, which can then do so themselves. But the process of building an effective algorithm requires much training of the algorithm with immense sets of data. The human brain is effectively a cluster of algorithmic neural pathways. Aside from genetic clues serving as the instruction set for the brain, everything else, including the understanding of the world is based upon repetitive training which creates the neural pathways required for an effectively functioning brain. Just as there is no way (currently known to humans) to evolve an agile brain without training, there is no short route to context. It requires herculean efforts to collect data and the constant process of teaching algorithms the changing interpretations of things (until there are algorithms to interpret the factors causing those changes).
Intelligence is algorithmic and the only route to generating contextual and useful information is thru data.
Where do we stand today?
Context is the most prized commodity in the connected world. Today, it is known as its primitive form, big-data. However, sings of evolution are apparent in escalating endeavors by internet giants.
At the forefront of context today is Google and not coincidentally. Google originated as an algorithm to organize and make sense of the massive amounts of data on the internet. Its relentless pursuit for more relevant search results has taught Google more about data than any single entity could claim the knowledge of. Product decisions which might appear to be foresights from outside, demonstrate this.
- Google Voice, the “free” VOIP service from Google, has given them an unparalleled understanding of the spoken word. While Apple, the only other company with a successful voice-input interface, has to rely on licensed technology from Nuance and high-latency server-side processing, Google’s voice offerings are incredibly fast and accurate.
- Android, Google’s “open-source” operating system, is the most widely installed operating system on the planet and an unprecedented source of data.
- Google has long been known to design its own data centers and networking equipment. It is not merely an optimizing endeavor but a natural response to their realization that data interpretation is the foremost requirement for an intelligent world. Google landed on this realization long before others, including hardware makers.
Facebook and Amazon
Facebook and Amazon are the other two contenders which stand to benefit the most from context. They have access to some of the most personally identifiable data in the world. Facebook has, rightly so, initiated the Open Compute project and Amazon has expanded into IaaS, learning about data handling along the way.
Apple stands at a critical juncture today. As the owner of one of the biggest software and media ecosystems in the world, it is increasingly appearing as a services company. Apple has historically strayed away form intrusive data mining but they have access to one of the largest and fastest growing datasets of user preferences and behaviors. However, from the outside, it seems that Apple is culturally misplaced in this era of context. I will elaborate myself and analyze Apple’s state in a later piece.
Technology is not an abstract endeavor of the few, it is an effort to understand our world, its immensity and grandeur. It is not a divide between the real and the digital but a bridge between the abstract and the “normal”. Technology enables us to understand aspects of our world that we never fathomed and context is the very first step. It is a new dimension of awareness.