User-Authored Personal Sensing Systems: Who Gets to Define the Boundaries of Technology?

A World Shaped by Design

Every technology product we use is more than a neutral tool. Design carries an ontological quality — it quietly shapes, and sometimes constrains, what we can imagine and achieve. How an app defines "health," how it categorizes "emotions," how it decides which data is worth recording — these choices invisibly draw the boundaries of users' cognition.

This research approaches the question from exactly this angle: designed artifacts are ontologically significant. They shape — and sometimes narrow — what becomes possible or even imaginable. In other words, when you use a system to understand yourself, the system's designers have already made countless assumptions on your behalf about what matters and what exists.

The term ontology originates from philosophy, referring to the basic assumptions and classification systems underlying what exists and what is real. In technology and design, the concept is borrowed to describe how a system implicitly models reality — how many stages a health app divides "sleep" into, which emotion words an affect-tracking tool offers, how a scheduling system draws the line between "task" and "event." These are all ontological choices made by designers. Once baked into a system, these choices quietly influence how users perceive and describe their own experiences. Information science researchers Geoffrey Bowker and Susan Star articulated a similar argument in their landmark work Sorting Things Out: classification systems are never neutral. They carry specific values and power structures, which are continuously reinforced through everyday use.

Giving Users the Role of Designer

One way to ease this "pre-closure" is to return to people the power to design and construct the systems they use. This idea is far from new in academia — decades of research have accumulated around systems that grant users authorship. From end-user programming to highly customizable tools, researchers have long explored how to enable non-experts to shape their own technology.

Yet the researchers behind this study point to a long-overlooked problem: these systems, when evaluated, are almost always assessed solely on whether they are "usable, useful, or technically feasible." These metrics matter, but they leave a deeper question unanswered — how exactly do users negotiate with the ontological boundaries built into such systems? When a system allows customization, can users genuinely break free of the designer's framework, or are they simply making choices within a pre-given set of options?

End-User Programming (EUP) refers to the design philosophy of enabling ordinary users without professional programming backgrounds to create or modify software behavior. Canonical examples include spreadsheet formulas (e.g., Excel), visual programming environments (e.g., MIT's Scratch), automation rule platforms (e.g., IFTTT's if-then trigger logic), and the no-code/low-code tools that have proliferated in recent years. EUP's core claim is that the people who actually use software often understand their own needs best, and should therefore have the ability to shape their tools rather than passively accepting the tools' built-in logic. Critics, however, note that most EUP systems still operate within a "possibility space" defined by the designer — users can combine existing modules, but rarely define entirely new concepts or categories. This is precisely the tension this research sets out to examine: the fundamental contradiction between empowerment and pre-framed ontology.

Probing the Experience with a Wizard of Oz Method

To open up this neglected question, the research team designed two open-ended probes using the classic Wizard of Oz (WoZ) technique. The core of this method: participants believe they are interacting with a real intelligent system, while in reality, researchers are playing the role of the system behind the scenes.

The advantages are clear — it bypasses current technical limitations, allowing participants to genuinely experience the process of "training a personalized machine learning system" without being obstructed by engineering constraints. The study focuses specifically on personal sensing systems: technologies that use sensor data to understand and represent users' own states. Within these probes, users could experience firsthand what it means to teach a system to recognize categories and concepts that are uniquely their own.

Why This Method Matters

The Wizard of Oz technique is not just a convenient workaround here. It allows the research to shift attention from "can the technology do this" to "how do people think and negotiate within it." When a user tries to get the system to understand a concept it was never pre-programmed for, the friction, adjustment, and redefinition that emerge in that moment are precisely where ontological boundary negotiation happens.

The Wizard of Oz method (WoZ) is a classic prototyping technique in human-computer interaction research, named after the behind-the-scenes wizard who controls everything in the story. It can be traced back to IBM researcher John Kelley's 1983 speech interface experiments — at a time when speech recognition was far from mature, researchers had typists covertly transcribe users' spoken input in real time, simulating a system that could "understand natural language." Participants were unaware, yet could genuinely experience and evaluate the interaction. The method's core value lies in decoupling user experience research from the difficulty of technical implementation: researchers don't need to wait for the technology to mature before exploring user behavior, expectations, and cognitive processes within a given interaction paradigm. WoZ is especially common in AI product research, because fully implementing many AI capabilities (such as natural language understanding or personalized learning) is extremely costly, while insights at the user experience level are often needed much earlier.

Implications for AI Product Design

The value of this research extends well beyond academia. As personalized AI and sensing technologies grow increasingly embedded in everyday life — from health monitoring to emotion tracking, from smart homes to wearables — the question of ontological negotiation between users and systems becomes ever more critical.

If a system only offers the categories its designers imagined, then no matter how "usable" it is, it may quietly constrain how users understand themselves. Genuinely giving users authorship means more than letting them operate the system — it means giving them the capacity to challenge and reshape the classification logic underlying the system.

For product designers, this raises a new dimension of evaluation: beyond usability and functionality, should we also measure the degree to which a system allows users to negotiate — or even break through — its pre-defined boundaries? This may be a crucial step toward building truly human-centered personal sensing technology.

Conclusion

Technology has never been a value-neutral container. When we hand the power of design to users, the real question is not whether the interface is friendly, but whether users can participate in defining what is possible. This research, conducted through Wizard of Oz probes, invites us to revisit the deep questions in human-computer interaction that have long been left unanswered — questions that are growing more urgent than ever as personalized AI continues to proliferate.