Smart Things, Dumb Data
Just because devices collect data doesn't mean we can use it. Why data from IoT devices needs to be standardized.
- By Barry Devlin
- June 17, 2016
Bright, shiny things are everywhere. Like magpies, we collect them, carry them home, wear them, use them for a while, and finally forget them when the next bright, shiny thing is pushed at us by the Internet marketing machine. The Internet of (bright, shiny) Things -- IoT -- is eating the world, to paraphrase Marc Andreessen's 2011 Wall Street Journal article about software. However, it is the world that is in danger of getting indigestion. These bright, shiny things may be smart, but their data architecture is pretty dumb.
The smart-things-dumb-data problem is particularly evident in the Smart Homes market. Over the past couple of years, a plethora of connected, IoT-enabled devices has been promoted mercilessly to home owners. A recent addition to the list is a frying pan that communicates with your smartphone, guides you through the cooking, and can even control the temperature if you fork out for an electric countertop burner and a WeMo switch. Ashley Clark Thompson provides a glowing review of her salmon searing experience. At $200, plus another $100 (or more) for temperature automation to tackle just one part of preparing a meal, the return on investment seems doubtful. ROI is but one part of the problem.
The story clearly illustrates the approach to market taken by suppliers of "smart, connected" devices. Individual, separate parts of real-life processes are smartened by the use of sensors and connected to the nearest computer -- usually a smartphone -- without any thought of what might happen before or after, either in the specific process or in the broader context of the user's life. Is there salmon in the refrigerator and is it fresh? Was it sustainably sourced? Do ingredients need to be bought, weighed, chopped, or washed? How many calories does it carry?
Some of the steps implied by these questions can already be automated or augmented by IoT approaches. However, even the most tech-savvy home owner would be hard pressed to join the all dots into a complete picture.
It's not only our homes that are being instrumented with smart (but one-trick) devices. An army of wearables are invading our bodies too, quite literally in some cases. Joanna Stern lists examples from hydration to feminine hygiene where smart devices offer to record and take over responsibility for every personal activity. The disempowering marketing message from one smart product startup is that "remembering to floss your teeth is hard." When you do as prompted, another data dot is added to another disconnected and disparate data set.
Despite the preoccupation with the bright, shiny, one-trick things, suppliers such as Apple and Amazon are already creating "islands of integration" of different Smart Home devices. Amazon's Echo (and its follow-on Dot) are probably the most advanced, offering the physically embodied and friendly Alexa that (or should that be who?) enables voice-activation in a range of devices from different suppliers, including security systems, connected thermostats, Philips Hue light bulbs, and Samsung SmartThings. It can also extend outside the home to offer integration with services such as Spotify, Uber, and Domino's Pizza. Such process-building requires understanding of the different languages that devices speak. The ecosystem is being built one device at a time. It is a competitive and perhaps sub-optimal approach, but winners will emerge.
That covers people and process, two of the three thinking spaces I defined as a conceptual architecture for Business unintelligence. (We can consider Things in this context to be proxies for people, sensing and operating in the real world, essentially on behalf of humans.) As is often the case in IT, the foundational thinking space -- information -- gets the last and least attention. The bright, shiny Things catch the eye first. The actions and activities that they support or perform attract our attention and admiration, especially as the processes become more inclusive and apparently intelligent.
However, it is the underlying information -- we might reasonably call it smart data -- largely silent and invisible, that will be the real foundation of the revolution occurring in the Smart Home and beyond.
The current situation with IoT data is similar to that seen in the earliest days of business application development, where every application used its own definition of the data used. As every data warehouse and master data management developer knows, this creates significant challenges whenever such disparate data must be combined or used together. The solution was data modeling and governance, particularly at the enterprise level.
However, the IoT situation is far more complex and extensive. Millions of different IoT device types with different data "models" are being developed and distributed into every imaginable environment, where data is used and shared in unpredictable ways within and across enterprise boundaries. For business applications, the problem was limited to a handful of systems with relatively well-defined use cases, mainly within a single enterprise. The data modeling effort was neither simple nor cheap. We are still doing it on a case-by-case basis. Extrapolating this effort to the IoT, this Dumb Data problem will become unmanageable unless we take early action.
The onus is now on manufacturers of IoT devices to define and standardize the definitions of the data being collected and made available externally. These definitions must be agreed, documented, and made mandatory for all devices connecting to the Internet of Things. We are already late in doing this, with billions of devices already in the wild, speaking in thousands of tongues. According to Peter Sondergaard at the Gartner Symposium last October, by 2020, one million new IoT devices will go online every hour.
It is time we took care of IoT data governance if we are to avoid a Dumb Data environment in Smart Homes, enterprises, cities, and beyond.
Dr. Barry Devlin defined the first data warehouse architecture in 1985 and is among the world’s foremost authorities on BI, big data, and beyond. His 2013 book, Business unIntelligence, offers a new architecture for modern information use and management.