Homework 4 - Assessing maturity

Andrew J. Ko

In homework 3, you learned to do generate a detailed, near-expert level understanding of exactly how a particular class of interfaces works. I argued that this level of understanding is key to being able to judge the limitations and uncertainties of an interface.

The next skill I want you learn is how to make a judgement about those limitations and uncertainties. Why a judgement? In the future, you're going to be in an organization where someone asks you as a designer, an engineer, or a researcher to provide advice on a user interface paradigm and whether it's worth investigating in. Imagine you work for Honda and they want to know about the maturity of driverless car interfaces. Imagine you work for Ikea and they want to decide whether to invest further in augmented reality applications. Imagine you're work for a non-profit doing micro-lending and they want to decide decide whether machine learning is good enough to offer a recommendation feature that would actually increase lending. All of these situations actually happen in real companies, and they all require you to use your expertise in precisely how a class of technologies might impact user experience.

I wish I could say there was some process you could follow for making these recommendations. I'm not certain there's just one. Instead, I'll offer an example of a judgement process and a resulting opinion.

Let's pretend I work for Apple and they're debating whether to invest in the R&D necessary to add FaceID to unlock Apple Watches. Here's the process I followed to make a recommendation:

I then apply the limitations and uncertainties to the scenario. To do this, it's easiest to create a table with limitations in the rows and scenarios in the columns:

Limitation Puffy rain coat user Sweaty jogger Pre-workout news reading
FaceID won't work for surfaces that don't reflect infrared light The user's face would still reflect infrared light. The user's face would still reflect infrared light. The user's face would still reflect infrared light.
FaceID won't work in environments with rich fields of infrared light If the user also had an iPhone with FaceID, it might not work if both devices were projecting light on the face at the same time. It also wouldn't work if the FaceID sensors were covered by the puffy winter jacket. There shouldn't be any other interfering infrared light. There shouldn't be any other interfering infrared light.
FaceID might not work for faces that don't look like faces in Apple's training set The hood might obscure some of the user's face, preventing a face from being recognized. The jogging might cause the watch to rotate, preventing the face from being fully within the sensors' view, requiring the user to readjust the position of the watch and try again. Nothing is obstructing the user's face.
FaceID might not work for people without two eyes that can gaze at the camera sensor (unless the user turns off gaze-detection). The rain on the glasses might interfere with gaze-detection by warping the sensor's image of the pupils. Because the user is jogging, they may not be able to gaze at the watch in a stable enough manner to allow gaze detection to work, and if the user tried to stabilize their gaze, they might run into something because they're not watching where they're running. Nothing is preventing attention on the sensors.

Should Apple adopt FaceID for Apple Watch? I recommend that rather than investing in solving these edge cases (and having some unfortunate news stories about injured joggers), it might make more sense to invest in R&D for other wrist-based biometrics that could passively authenticate a user without any user attention at all.

Note that the result of this analysis isn't necessarily a yes or no (though it could have been, if the evidence was strong), but rather a recommendation or set of recommendations with rationale grounded in the analysis.

The Task

Your task is to take the limitations and uncertainties from the last homework, describe some fictional organization context in which you're asked to make a judgement about the interface you've been studying, and then follow the analysis process above.

Your submission should follow this template:

In practice, many analysis like this might be more informal. I'm making you be more structured and systematic about it will not only scaffold your process, but it will probably also lead to better reasoning.


This homework is worth 5 points: