The Sharply Decreasing Disruptiveness of HCI
Zhilong Chen, Yong Li · 2025 · Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3706598.3713917
Summary
This large-scale bibliometric study by Chen and Li from Tsinghua University investigates the creativity of HCI research itself by measuring its "disruptiveness" — the degree to which a paper's contribution is recognised independently rather than merely alongside its references. Analysing 17,476 papers from four premier HCI venues (CHI, CSCW, UbiComp, and UIST) spanning 1982 to 2023, the authors use the disruptiveness index D to quantify whether papers disrupt existing knowledge or consolidate it. The index ranges from -1 to 1, where higher values indicate that subsequent work cites the focal paper alone (disruption) rather than citing it together with its references (consolidation). The study traces the evolution of disruptive themes across six time periods, revealing a shift from technical system development topics (bubbles, user interfaces, software) in the 1980s-1990s toward social concerns (social media, privacy, politics) and emerging technologies (machine learning, augmented reality, wearable computing) in recent decades. The authors also examine knowledge use patterns, finding that disruptive papers tend to build on fewer, older, and less popular references than non-disruptive work, suggesting that drawing from overlooked or unconventional sources may foster disruption.
Key findings
The central finding is stark: HCI's disruptiveness is declining at a rate even faster than the global average across all scientific fields. In 1982-1983, 81.4% to 87.7% of papers from the four premier HCI venues were disruptive; by 2021-2023, only 17.0% to 21.8% were — dropping below the global average for all science. The absolute number of disruptive papers has grown, but their growth rate lags far behind the explosion in total publications. Notably, disruptiveness and citation counts are nearly uncorrelated (Pearson's r = 0.020), meaning high-impact and disruptive are fundamentally different qualities. Disruptive papers are more likely to achieve exceptionally high citations (top 1%) but do not have higher average citations. Author freshness is strongly associated with disruptiveness: teams with more new authors and shorter career spans are more likely to produce disruptive work, with papers where 80%+ of authors are new having a 72.6% probability of being disruptive. Industrial research labs (Xerox PARC, IBM, Apple) that once dominated disruptive HCI output have largely disappeared from the top contributors, replaced by universities like the University of Washington and Tsinghua University.
Relevance
While not directly an accessibility study, this paper has significant implications for accessibility research within HCI. Accessibility has appeared as a top keyword of disruptive HCI papers in both the 2016-2020 and 2021-2023 periods, suggesting it remains a fertile area for paradigm-shifting work. The finding that disruptive papers build on older, less popular references and that fresh authors drive disruption is particularly relevant for the accessibility field, where interdisciplinary newcomers (from disability studies, rehabilitation science, or lived experience) may bring the unconventional perspectives most likely to produce breakthroughs. The paper's broader warning — that HCI's emphasis on extensive related work, context-dependent contributions, and conservative peer review may be systematically suppressing disruptive research — should concern accessibility researchers who often challenge established paradigms. For practitioners and researchers, this study argues for welcoming unconventional approaches and supporting fresh voices rather than optimising for citation counts.
Tags: bibliometric analysis · research methodology · creativity · human-computer interaction · disruptiveness · citation analysis · research trends