Having just read Scarcity, I was reminded of how important working memory (or mental bandwidth) is for our productivity as professionals. I think this is an understated and, often differentiating, characteristic of great productivity software. In particular, software that requires us to make decisions benefits greatly from considering working memory as a design objective.
What is working memory?
Let me first define what I mean by working memory. According to wikipedia, working memory is the system that is responsible for the transient holding and processing of new and already stored information, an important process for reasoning, comprehension, learning and memory updating.
In easier to understand terms, working memory is the mental capacity we dedicate to a task we are working on, including processing the inputs, matching relevant patterns, reasoning about the task, making a decision, and acting on it.
Why is working memory important?
There are three reasons why I think working memory is, and will continue to be important for productivity software.
Decision making
For most of industrialized economy, our work was labor intensive. However, the characteristic trait of the information economy is ..well, information. This implies a lot of what we do is create artifacts based on decisions. Creation of artifacts is easy, thanks to software. As a result, most of our time is taken up in making decisions. Unlike labor intensive tasks, making decisions depends on how much working memory we have.
Influx of data
Data is pervasive. We are immersed in it. At surface, this makes decision making easier but that’s a trap. Almost always, the data that is most visible is the data that is easy to access, not the data that is most meaningful for decision making. It takes significant working memory to avoid getting distracted by easily available data, and doggedly pursue data that is meaningful.
Unintuitive phenomena
An intuitive decision is where you have a high degree of confidence in the outcomes of your action before you act. Where it isn’t obvious, instant feedback on actions helps form this intuition. That’s how we learn, and become good at things. However, there are many functions in an enterprise where this is simply not possible, especially when the feedback loop is long, and the most suitable agent is not making the decision (I call this decision distance elsewhere).
For example, hiring typically can take a few months from publishing a job spec to closing a candidate. The recruiter during this time, is making go & no-go decisions on hundreds of candidates, matching candidates to open positions, hiring managers are conducting dozens of interviews & grading candidates, before ultimately hiring a candidate. To make matters worse, the organization doesn’t know the final impact of their decisions till a year or more after the hire.
Similarly, closing a deal takes several weeks or months. It starts with the marketing team qualifying leads, sales reps talking to the prospect, doing demos, asking the right questions, talking to many people, furnishing collateral, etc. When a deal finally closes, hundreds of questions have been asked, many collaterals shared, and many small decisions made. The feedback loop is entirely absent. No, a sales playbook which relies on a human being (sales manager) for feature extraction, & is historic on arrival is not a meaningful enough feedback loop.
History of productivity software
The class of problems productivity software has solved involved reduction of effort & time. Ranging from accounting to writing documents to sending messages, software greatly reduced the effort & time required to carry out a task. Even today, software makers continue to optimize for productivity in the narrow sense of the word – the amount of time or effort taken to do X (e.g. marketing automation software).
However, as they reduced time, software started providing instant feedback as well. Excel caught our errors in formulae, and Word pointed out our spelling errors. These features liberated our working memory. When writing a document, I didn’t need to worry about spelling errors because the application would catch them for me.
Software engineering took this a step further. One of the huge benefits of continuous integration environments is instantly learning the root cause of a regression or failure. This frees up working memory to fixing the problem, instead of finding the problem.
Designing for decision making
With the amount of data and the number of situations where decisions are not intuitive & the feedback long, we need to start designing software that maximizes available working memory, and in many cases, amplifies it. There are two specific tools that will continue to further this.
User experience
Product managers have a hard time justifying or measuring user experience. This is difficult to do because good user experience doesn’t necessarily “save time” or “allow me to do X things”. Good user experience maximizes working memory of the user. For enterprise software, this will become increasingly important. Far too much memory is wasted trying to get software to just “work”. Well designed software creates a state of zen for its users, and allows them to get on with their work. Software teams that consider design as a first class citizen will win, & teams that sleep on importance of designers will lose. Not only will teams without designers lose, the teams that value design will see its benefits and continue to pull away by doubling down on design. I’m convinced one of the primary reasons for Slack’s success is that it amplifies working memory of its users
Machine learning
Techniques of machine learning are very applicable to problems where feedback is not instant or intuitive. This includes nearly all business functions, but primarily sales, recruiting, & people management. Human beings are terrible at feature extraction in noisy situations as well as recall. Our brain isn’t designed to be good at this but machine learning helps. This results in poor decisions & frustration for the work force. Machine learning, when combined meaningfully with good user experience, machine learning will amplify human decision making exponentially.
In summary, I think product teams need to start thinking in terms of working memory as an objective because a significant amount of work we do involves making decisions, and decision making requires working memory.