UX vs. AI

A review of Understanding Computers and Cognition

There are some UX books that a bit like candy: fun to read and light on the fundamentals. Then there are books on the other end of the spectrum. These are books that are heavy and out of date, but reflect the history of a field. I have more than a few of these in my library. I thought it was time to pick up one and give it a thoughtful review. I picked Understanding Computers and Cognition: A New Foundation for Design by Terry Winograd and Fernando Flores.

I’ll skim over most of the details about the history of UX to properly situate the book. Back in the 70s and 80s there was a big push for AI. If you think the push for AI or neural networks is big now, it was nothing in the 70s and 80s. The first UI was just being worked out at PARC, the mouse was being invented at SRI, and computer companies like Apple and Microsoft were just getting their feet off the ground. With these strange new machines with strange and interested interfaces a new ethos hit the scientific arena: all problems could be solved with a big enough and a smart enough computer.

Soon, funding started to be available for computer scientists to try and figure out how to make a smart computer. At the same time there were these psychologists who were saying, “hold on a second, no one can figure out how to use these darn things.” In particular, the sub-domain of psychologists who studied cognition and learning were saying that computers cannot “learn” and that to take a step back. Instead we should focus on making computers a tool that people can use — to make a system were computers and users can easily talk to each other.

This is where the work for this book starts. In this book Winograd and Flores take a formidable jab at AI and argue that there has to be a foundational shift in how we think about computers: users have to be first. These days that kind of thinking is much more common place but it wasn’t back in the 80s. Here is one of my favorite passages that does a good job of summarzing the whole book,

“This book, then, is permeated by a concern for language. Much of our theory is a theory of language, and our understanding of the computer centers on the role it will play in mediating and facilitating linguistic action as the essential human activity. In asking what computers can do, we are drawn into asking what people do with them, and in the end into addressing the fundamental question of what it means to be human.”

The book has two central parts. The first is a discussion on modeling a decision making tree. Back in the 80s computer scientists thought that if we just modeled enough, all of human action would make sense. Think of an endless flowchart or kind of what you saw on Westworld with modeling the robot interactions. Scientists thought that if we could just make a model complex enough that things would seem to make sense. And, because of this kind of thinking technologists were making software that reflected that strict cognitive bias.

The second part of the book frames what the psychologists were arguing about humans: that no one action is going to be same as the next because of the uniqueness of existance. Essentially, they argue that it is impossible to model human interaction. This means that when we create software we can create software to support needs in situ, but there is never going to be only *one* use or *one* way. They were arguing that context is the big player in any interaction; decision making modeling can never adequately account for context.

Ok, now I’ve just summarized some thicker concepts so I’m now going to take a step back and talk about whether or not the book is good. The book is terrible. It really drones on. And, it gets really into the philosophy of existence. I mean it is not a page turner. In fact, there aren’t any pictures. The cover doesn’t even have a picture, just more words to beat you to death by boredom. It is a difficult one to get through. The only way I got through was by breaking it down into 15 page chunks and making myself read it every day — like you do with things that are good for you but unenjoyable.

BUT, I’m glad I read it. After reading this book I was much more in touch with the foundational roots of a field I’ve dedicated my life to. I was able to hold my own in conversations about why social sciences are important and impactful on “hard” sciences. I was also able to see and remedy some of the tensions that still exist between decision modeling and user interfaces.

While this book is a slog, it is a well written argument. It is well thought through. It doesn’t try to be anything more than what it was: a very long commentary. It isn’t pretentious and I think anyone could probably read it and follow what is being said (with the background I provided above). It just isn’t going to be enjoyable.

Book Club Questions

  1. Have you ever made a UML diagram? How do you think modeling impacts the final design?
  2. It is well known that the waterfall method of software engineering fails. What do you think is the relationship between the waterfall method and formal modeling?