Privacy, equity, energy: the bills coming due
The technical problems in ubiquitous computing largely got solved. Three others were left on the table, and they are harder because none of them is only technical.
Wearables and networks moved the 2006 framework from paper into use. What they exposed was not a missing capability but three unpaid accounts: privacy erosion, digital inequity, and the energy cost of ubiquity.
Privacy: from defence to discretion
Continuous collection is the premise of the field, not a side effect. Phones, wearables and home sensors gather location, physiology and behaviour in real time and at high frequency, and analytics re-identify individuals from traces that were published as anonymous. Misuse is not hypothetical: it runs to discrimination, manipulation and loss of autonomy.
A useful way to audit this is dimensional. The 5-D analysis separates identity, location, digital footprint, query and intelligence — each leaks differently, and a system can be sound on one and wide open on another. Location alone is usually enough to infer a home, a workplace, a routine and a set of relationships.
The Berkeley formulation of why ubicomp is different still stands:
Protecting private data is a major concern for users. There are a few challenges that make data security in Ubicomp settings different from other system protection: the environment is often unfamiliar to the users… they will not have a trust relationship with the owners of the environment as they might with their local system administrator.
Duan & Canny, UC Berkeley
In a conventional system the user has some trust relationship with whoever administers it. In an instrumented space they typically have none, and often do not know the space is instrumented. Access-control lists are the wrong instrument for that situation.
The alternative is data discretion: granular, revocable, machine-enforced control over what is collected and who may compute on it, backed cryptographically rather than by policy text. Framed that way privacy stops being defensive and becomes a negotiation between producer and consumer of data — which is also the only version that survives contact with a commercial product.
Equity: the divide did not close
Ubiquitous computing was pitched as democratising. Participation data in the field's own professional bodies says otherwise: around 3% of ACM's global membership is from Canada and 4% from Latin America and the Caribbean, while participation from Africa, the Middle East and low-income countries is marginal. Women, racial minorities and disabled researchers remain underrepresented in the rooms where these systems are specified.
If the industry continues to source new hires through traditional pipelines without prioritizing workplace diversity, these jobs will be largely unavailable to an emerging workforce that is made up of women and underrepresented minorities.
Forbes, cited by ACM
There is evidence that this responds to design. When UbiComp/ISWC moved to income-based registration with travel grants and targeted programming, the 2023 edition produced its most demographically diverse cohort to date. That is a solved logistics problem being described as an intractable cultural one.
Sustainability: a linear lifecycle at planetary volume
Devices are still specified for peak functionality and a short life: non-renewable inputs, a couple of years of use, then e-waste. At the volumes ubiquity implies, embodied carbon dominates the footprint long before anyone argues about standby power.
The 21st-century computer should 'weave themselves into the fabric of everyday life until they are indistinguishable from it', envisioned by Mark Weiser… This vision, however, is overshadowed by a growing concern: the environmental impact of this rapid proliferation of devices.
University of Michigan, Circular Interactive Material project
Circular interactive materials are the most interesting response — substrates that sense and actuate but are designed to be recovered, recycled or to degrade. Biodegradable sensors and energy-harvesting materials are early, and they attack the part of the problem that efficiency gains cannot reach.
What responsible ubicomp would require
Addressing all three is not a technical challenge alone; it is a societal one. Concretely it requires:
- Empowering users with real-time, granular control over their own data — data discretion, not consent theatre.
- Broadening participation in research, design and deployment, especially for underrepresented groups.
- Designing for circularity, not just functionality, so the environmental cost is bounded.
Achieving real justice, equality, diversity, and inclusion… is an incredibly ambitious goal but, as highlighted in the introduction, supporting diversity in computing conferences is vital to ensuring adequate numbers of diverse computing professionals.
ACM UbiComp/ISWC best practices
None of that is expensive relative to what the industry spends on silicon. It is unallocated, which is a different failure.
The 2006 thesis argued that ubiquitous computing needed both technical and social science to work. That reads as the least controversial claim in the document and it is the one the field has done least about.