Software developers and marketers are talking a lot about “context” these days. And yet, end customers are still presented with irrelevant ads and irrelevant software options. The truth is, contextual awareness is not even close to being "there" yet. Here are a few of the technical reasons reasons why the design hasn’t met its goals.
More developers are thinking about and talking about context or “contextual awareness” because, like GUIs and today’s touch-enabled UIs, it will change the way humans and devices interact.
“Contextual awareness” has been a marketing buzzword for about a decade. Before that, the goal was personalization in a more general sense, which required more technology and data points than anyone considered. The target marketing initially envisioned still has not reached its potential – as evidenced by the ads that search engines and social networks serve up. (Really, Facebook, you think I’d care about that?!) Behavioral awareness, a more recent addition, helps make target marketing more accurate because it adds another data element, but in the absence of context, behavior can be misinterpreted.
Meanwhile, developers have become user-centric. UIs have changed dramatically and interactions have become an art form. But still, software is not adapting to the needs and whims of users in the kinds of ways that will be possible in the future.
Context is a Difficult Problem
Who, what, where, when, why, and how questions are as relevant to software development as they are to marketing and journalism. The root of the contextual awareness problem is a shortfall of data aligning humans, their behavior, their histories, contexts, and predictive analytics in ways that truly reflect the motivations of individuals.
“Most software developers and firms strive to build straightforward products that provide the most flexibility to target users. Those two things become almost diametrically opposed as you widen your target audience,” said Jeff Kear, owner of Planning Pod. “There's a reason why people only use 10% of the features provided in most software. It's because those tools were built to encompass many contexts, and most users simply don't have the time or desire to learn all those features, even if they would use them.”
Context varies from person to person, company to company, device to device, Kear said. His company’s computer-based event planning tools require a fair amount of data entry upfront. When he and his team began to build mobile applications, they were peppered with questions about how much flexibility to provide knowing that most of the users would only be editing information or making quick entries on their phones as opposed to typing in volumes of information.
“Not every user is the same, which leads to more head scratching as to what to include and leave out,” said Kear. “So at the end of the day, you almost have to build for your most typical user because you can't possibly account for all the variations of usage for a wide population.”
Data entry is one way software developers and marketers have been trying to understand context. However, software designs and marketing campaigns are often limited by “if/then” thinking that oversimplifies what the context really is and more importantly, what end users really want.
“Anytime you find yourself having to remind a system about obvious information, such as you want directions from point A to point B, the software should know you’re at Point A. This type of thing is frustrating,” said Jeff Powers, CEO of Occipital, which develops mobile computer vision applications. (The company also offers a 3D sensor.) “We humans continue to carry out monotonous tasks such as looking at a calendar and trying to plan or typing in credit card and phone numbers over and over. The systems we use are remembering more information about us than they did 10 years ago so the context is becoming richer, but we still have a long way to go.”
Greater Intelligence is Coming
The context you build into your applications needs to consider the users’ circumstances and environment at a point in time but not in isolation. It needs to contemplate the present and the past. For example, a user’s behavior in a given environment and under certain conditions involves many data points such as location, weather, what the person is doing there and what he did previously, how others are acting or reacting, and how they acted or reacted in the past. Given these and other data points, the UI and a corresponding functionality could adapt automatically as the user’s goals, behavior, and context change. This is where sensors, machine learning, and cloud-based processing come in, not to mention the rethinking of UIs and device types.
In the absence of critical data points and the intelligence that must be derived from it, software and things running on software will continue to make false assumptions rather than true observations.
“Context is going to be an enormous market in the future. While we can sense activity and have computer vision moving forward, we will have all types of contexts including digital contexts,” said Xerox PARC scientist Oliver Brdizcka, who manages PARC’s Contextual Intelligence research area. “We just have a simple view of context now but it will get more sophisticated.”
The sophistication Brdizcka envisions requires innovation in every area that enables the understanding of context, from models and algorithms to UIs. Greater levels of sophistication will also require more processing power, greater amounts of storage, and the “always-on” recording and interpretation of what is being experienced by the end user. All of that will be part of a fabric that enables context to become more accurate and therefore more relevant.
“It’s a very hard problem,” said Brdizcka. “We need a critical mass of data interconnections and the machine learning methods to make sense of it.”
So the problem with contextual awareness is that we lack a critical mass of information, for now. If that were not true, there would be fewer false assumptions about lifestyles, work styles, tastes, preferences, demographics, and what is appropriate given a certain set of circumstances – that is, what should be presented in context.
Yes, technologies and methods have improved dramatically over the last decade but we have a long way to go when it comes to contextual awareness which is actually good news: It leaves a lot of room for innovation.