In 5 years' time, the world's total population will be somewhere between 7 and 8 billion humans. At the same time according to Gartner there will be 25 billion connected objects in use. Even assuming every human has access to a computer or smartphone (unlikely to happen by 2020), there will be a lot more members of the Internet of things than there are members of the Internet of people!
Publishing for people
Today, the majority of your data publishing and consumption processes target humans. Of course, inside your organization, machines (and software they run) have access to data too. At the risk of oversimplifying, they collect data from sources, parse/cleanse/aggregate this data, and extract insight from it. This insight is then provided to its consumers, most of them human.
The form in which data is published varies: dashboards, reports, graphs and charts, textual content on Web pages, etc. But the commonality of these target formats is that they are designed to be consumed by human beings. The rendering of a report of chart matters as much, if not more, than the data it contains. Colors, 3D effects, scales, etc. all contribute to the readability of the data.
As far as websites go, an effort is sometimes even made to prevent data to be read by others than humans. Because you don't want your competitors to screen-scrape your website and automatically capture your product catalog, you go to great lengths to transform text into images, break down content in complex nested tables or interject useless HTML tags.
Publishing for machines
Just by their sheer number, connected objects will take an increasingly important space in the data management landscape in the next few years. A connected car needs to know about traffic or gas prices -- as its driver does. An insulin pump needs to figure out calorie and carb intake -- probably even more than the diabetic patient it is helping regulate blood sugar levels. An automated sprinkler system will be faster to react to a fast room temperature increase than the fire department on call.
It's actually not that hard to make data readable by machines. Proper Web design already separates content from presentation. Take it one step further and separate relevant data from content that is just there for context setting -- such as headers or comments. Then create Web APIs to serve the data -- the same APIs will be used by your website front end, and any machine that needs the data without the presentation layer.
Of course, without this presentation layer and context setting, there is a risk that the data becomes difficult to understand. This is where metadata is important. Be clear about what the data represents. And make it available in a machine-readable format.
Then connected objects will have the same ability to access the data than humans, and this data will be creating a lot more value.
It's not either/or
Providing machine-readable data won't preclude you to continue to make your data available to humans. If you design your data interfaces well, they will actually cater to both, at no incremental cost or complexity.
This article is published as part of the IDG Contributor Network. Want to Join?