When there are many strands, the cost of understanding goes up. The cost depends on various factors: how large and complex the topic is; the prior knowledge of the people trying to understand; the availability and power of relevant tools; our confidence that we have the right information; and much more. In one sense, this book is about how we can manage, shift, or reduce the cost of understanding. How then, do we manage this cost? And who does the bulk of this work?
If we return to our privacy policy example, the cost of understanding is traditionally addressed in one of several ways. Assuming we care enough about the topic, we will either:
• Call an expert: A legal professional could bear that cost for us, as they’ve been trained to make sense of the information (granted, this option brings with it a different kind of cost!). If we’re lucky, we might have a friend who is also an expert and willing to help us.
• Figure things out on our own: Given the time and motivation, we could make sense of all the legalese. While some people will undertake this effort, most people give up, concluding that “I’m not smart enough.”
Of course, many more of us opt for a third “nonunderstanding” option; we conclude that the cost of understanding is not worthwhile. Given the cost (money or time) of the two options above, we give up or trust things will work out, hoping we haven’t agreed to anything of consequence (like our first-born child).
Generally speaking, these two (or three) responses summarize how we respond when most things get confusing. Are there other ways to manage the cost of understanding?
When it comes to those obfuscating privacy policies, another approach is to distill the documents into something that is concise and easy to grasp. This is the approach taken by the group Terms of Service; Didn’t Read (ToS;DR). It’s a nonprofit group, staffed by volunteers, who read terms of service documents and convert them into a list of bullet points (Figure 1.6). Each item is given a thumbs-up (your rights are protected), thumbs-down (your rights are not protected), or a star (neutral). Each service also receives an overall rating, from Class A (terms that treat you fairly and respect your rights) all the way down to Class E (the terms raise serious concerns).
FIGURE 1.6 The group Terms of Service; Didn’t Read converts lengthy terms of service documents into a series of bullet points, rating each item and classifying the service as a whole.
ToS;DR does to privacy policies what I did for the diabetes chart: transforms something complex into something simple. But where I neither added nor removed any information, ToS;DR removes almost everything. They have to do that since the original documents are overwhelming. They carefully work through all the gory details, identify the essential details, convert them into plain English, and assemble the result into a clear and readable list. They distill and translate, taking on, as volunteers, the bulk of the understanding. They do the hard work, making your life easier.
Projects such as Polisis take a different approach to the same problem. They use machine learning to scan and summarize legal contracts, displaying information back to users in a consistent, visual representation (see Figure 1.7). Where ToS;Dr relies on humans, Polisis relies on algorithms.
FIGURE 1.7 Polisis uses machine learning to translate policy documents into a visual form.
Polisis produces an interactive, visual summary of a specific contract. Notice the difference between these two approaches: ToS;DR shifts the cost of understanding by doing work for us, then asking us to trust their conclusion. Polisis shifts the cost of understanding by making the document easier for us to figure out—there’s still work to be done. With Polisis, there is no easy recommendation, but rather clarity. From their project’s web page: “You don’t have to read the full privacy with all the legal jargon to understand what you are signing up for.” Polisis makes the information understandable, empowering you to make a more informed choice; it’s a tool that facilitates understanding. It’s probably not as easy to understand as the lists provided by ToS;DR, but it provides more detailed information about what information is collected and why.
This kind of solution excites us, not necessarily for the use of technology, but for how this technology taps into natural human abilities. As you’ll soon see, we learn through interactions. Our sense of vision is powerful. Humans are great at spotting patterns. We value learning that is active and self-directed. But, we’re getting ahead of ourselves ...
When we begin to view information as a resource, we open our eyes to not only the many problems of understanding, but also the many ways we might transform information to create understanding.
Who takes the time or effort to facilitate understanding? When should we delegate this work to other people, or algorithms, and when should we do it for ourselves? If we do take this on, how might we create our own understanding? What role does technology play in how we understand? These are central themes of this book. We hope that after reading this, you will be fully aware of when something is an understanding problem and what can be done about it, whether that means fixing things yourself, or being able to articulate what needs to change so that understanding can take place.
A Distributed System of Resources
Whether we are making sense of something for ourselves, or others, the questions are the same: How does understanding happen? And—more fundamentally—where does understanding take place?
Think of cognition as a distributed system of resources, resources that include mental perceptions but also much more. Where understanding takes place depends upon what resources we use.
Here is a brief explanation:
In some cases, what you think is in your head, based on prior associations. If you multiply 32 by 11 in your head, the cognitive work happens—in that moment—in your brain.
But where does the cognitive work happen when you use pencil and paper? Not in your brain and not in the pencil either. It happens everywhere, spread across brain, hand, pencil, and graphite symbols etched on the page. In this case, we introduce interactions with external representations.
When we introduce more resources into the mix, things get more interesting. Now, what we think of as cognition is spread through a collection of people, resources, tools, interactions, and so on.
To say that cognition is distributed across people and things doesn’t mean we’re talking about a new kind of cognition. It’s about seeing how all forms of cognitive activity—remembering, planning, reasoning, deciding, analyzing—depend on the ways people use their available resources. Moreover, these cognitive resources can include everything from neurons and whiteboards to laptops and airspeed indicators. Figuring it out is the art of bringing all this to bear.
This shift in perspective, away from just objects and representations and interactions and toward distributed cognitive systems, is vital for taking on complex understanding problems. This also requires coordination, whether this is simply coordinating the cognitive resources at your desk or the resources at play within a globally distributed team of people.
So