Human-Data Interactions
Augmenting our abilities to perceive,
understand and analyze
(A French version of this article is available here)
A year and a half ago, we redesigned our visual identity, freshened up our website and undertook to better explain what we do. Prior to that change, we used to introduce ourselves as a company specialized in data visualization interfaces, before we noticed that this denomination poorly reflected our multiple skills and the challenges we take on. We thus started using the concept of Human-Data Interactions to describe our activities.
This term, which refers to a movement that places Humans at the core of the data industry, is still rarely used, with the exception of some US/ UK publications (see Richard Mortier, Hamed Haddadi, Tristan Henderson, D. McAuley and J. Crowcroft on http://hdiresearch.org/)
Human-Data Interactions gather together all the ways aiming to improve how we understand and use the information contained in the data. Those interactions transform the way we learn and understand our environment, as well as our relationships to others and to ourselves.
Towards a better data literacy
Our societies have been producing increasingly complex information, and data has become a common way to encapsulate it. We generate huge amounts of data, referred to as Big Data when their volume, velocity and variety undermine the efficiency of traditional IT tools, and force us to rethink the way we approach statistics.
In this race towards ever-richer data, the limiting factor isn’t so much to do with technology, as it keeps improving, but rather with the abilities of humans that struggle to keep up with the pace. Computers and software have been designed to scale up, to produce and manage increasingly complex information. On the contrary, our human brains are not designed to keep up at the same speed. Our ability to understand and process data is limited, and already switched off from the continuous progress of our technological environment.
Just like there are sounds we cannot hear, we must now admit that there are logical concepts that we cannot conceive. Those include the algorithms that structure financial markets, the programs that drive our cars or regulate our cities, etc. Those complex systems lie beyond the understanding ability of the majority of humans. Therefore, just like we built tools to hear sounds we couldn’t hear, it is now time to build tools to understand those data systems we can’t conceive.
As a result, the true data revolution is not about running dozens of servers in parallel, but about how it can lead more people towards a better data literacy, i.e. a better understanding and intuition of data.
It is thus necessary to closely study how humans will access, understand and use data in the future. People will succeed through multiple interactions with connected systems: this is precisely where the scope of Human-Data Interactions lies.
The promises of Human Data Interactions
Human-Data Interactions seek to bring a more cognitive dimension to the multiple interactions people may have with connected systems, through design and underlying technologies.
Here are 5 promises brought along by HDIs to “augment” our ability to master data.
1 - See the invisible
The first challenge of HDIs consists in fostering trust around systems that leverage data. The purpose is to make the data feeding complex systems more visible. Smart cities, decentralized energy networks, autonomous cars, targeted advertising, etc.: an increasing part of our environment is structured by algorithms and data, that we often cannot see.
This creates a double challenge for our free will: refusing to be satisfied with an information environment that we understand less and less, and not letting third-party services influence our behaviors without us understanding how. We cannot live surrounded by algorithms and connected objects that seem like impenetrable black boxes.
At Dataveyes, this first promise refers to missions where we are asked to build applications that bring transparency to data, particularly personal data. It is about allowing a large audience to see large datasets as simply as they would perceive images and sounds, to bring trust in this new era of rich and big data.
2 - Easily understand large amounts of information
The second challenge of Human-Data Interactions lies in inventing a visual language that makes large volumes of data spontaneously understandable. It refers to the well-established field of data visualization.
In a nutshell, data visualization is about visually translating information contained in the data. It relies on the ability of our brain to process much larger amounts of information when it is presented visually. To do so, data visualization spatializes information and resorts to visual metaphors.
Data visualization is not a new subject, but it was until now largely limited to scientific or technical fields. Today, it must adapt to the needs of non-expert users who need representations to be more than a series of line charts, bar charts or pie charts to be meaningful.
At Dataveyes, this challenge is embodied by the missions where we must make rich and varied datasets do the talking. For example, our works can allow all citizens to approach large medical datasets, or help marketing professionals work with massive amounts of social data. We strive to invent the most meaningful visual codes to translate data.
3 - Allow non-experts to operate complex data
The third challenge of Human-Data Interaction is about allowing as many people as possible to maintain control over data. In other words, it consists in making the task of operating complex data straightforward and pleasant.
The goal is to replace heavy data requesting tools or computer code by interfaces allowing to execute operations such as filtering, sorting, selecting, cross-referencing, calculating etc., without expecting users to complete a lengthy training beforehand.
Seeking to facilitate interactions with data is closely linked to the need to reduce data complexity. Interactivity allows to break down dense information into fragments that are easier for a human to approach and explore. When zooming, navigating, or dragging and dropping, users make information their own at their own pace and depending on their needs. Users are placed in an active position that greatly helps taking ownership of knowledge.
At Dataveyes, this third challenge is materialized in missions where we imagine new ways to operate data. In addition to the now common touchscreen and mobile interfaces, we are led to explore contactless ways that introduce a much richer range of possible movements. This is typically the case with Kinect or Leap Motion technologies, or multi-device systems where familiar objects such as a smartphone enable users to interact with data from afar.
4 - Push the limits of our attention span
A fourth challenge for Human-Data Interactions lies in finding a better way to manage our attention span. More and more, we tend to delegate the processing of minor datasets to our peripheral attention, to better cope with the multiple sources of information surrounding us.
As we look to streamline the operations related to data handling, we witness situations where interactions with data completely fade to become more diffuse, even passive. Those frictionless interactions are the product of “ambient” interfaces that blend into our daily environment to bring information to our peripheral attention. Today, ambient interfaces consist of control panels or “social walls” in the reception lobby of companies, large urban screens, connected frames etc. Tomorrow, they could be mirrors, windshields, walls of public buildings, roads, public lighting etc.
At a time when information overload floods our concentration, those interfaces are discrete enough, and blend so well into our environment that we end up forgetting about their presence. Their purpose is not to monopolize our ability to analyze; on the contrary they do not require any effort of concentration. They disseminate information in a barely perceptible way throughout the day.
At Dataveyes, we address this challenge when working on projects requiring to combine algorithmic systems, data and information. For instance, how we will access information in the public transports of tomorrow, or how to anticipate the way connected cars will communicate with their occupants.
5 - Access key information, without necessarily transit through data
The fifth challenge of Human-Data Interactions resides in the automatic analysis of data. It is about building interfaces that do not display all of the data at a given time, but focuses on the useful information, disseminated in a synthetic way.
Even by mobilizing peripheral attention, the time a human can dedicate to the exploration and understanding of data remains limited. To free ourselves from information overload, we must rely on smart ways that are adapted to our needs, leverage useful information from the data, and deliver it to us at the right time.
All in all, the best Human-Data Interactions are the ones that allow us to take better ownership of our daily environment, without exposing us to its complexity. Those interactions will typically only focus our attention on prominent points, key information and significant variations.
Today such interactions are already at play in recommendation engines such as the ones of Amazon, Google, Netflix, or in smart electric meters. Those systems collect data about our behaviors, they learn from our habits to improve their analyses and provide increasingly relevant insight.
At Dataveyes, this fifth challenge comes to life when we build smart systems that display information to a user depending on specific contexts. This includes dashboards that emphasize the actions to take rather than data completeness, or applications that alert users when the data describing their environment varies significantly. In such cases, we work on each component of the data value chain — collection, structure, treatment, visualization, etc.- so that each step contributes to unveiling the information hidden in the data.
The title of this articles refers to « augmenting » Humans in its metaphoric, rather than transhumanist, sense. We wished to show how, through a better design of Human-Data Interactions, we could dramatically improve our ability to understand and leverage data.
Nonetheless, we cannot overlook the fact that those ways to interact with data are likely to deeply impact us, because they induce gestures, favor postures, and underpin ideas that will leave a mark on us. For this reason, it is crucial that those new smart interfaces come along with awareness and education.
Ultimately, our mission is not just about conceiving interactions with data. It also invites us to anticipate how to interact with systems that process our data. This thus opens our field to legal and economic topics that are as deep as they are fascinating.
