Knowledge that Transforms

Expectations propel the evolution of emerging technologies, yet existing research generally assumes that expectations are gradually displaced by concrete outcomes as technologies mature and become widely accepted. We revisit this assumption by examining how technology adopters' temporal framings of technology performance, their expressions of concrete outcomes of technology adoption, and the broader technology discourse reflected in trade-press coverage relate to one another over time. We illuminate these patterns by tracing 21 years (2002−2022) of radio-frequency identification (RFID) adoption and analysing adopters' statements in 407 case stories published in The RFID Journal, coupled with a longitudinal corpus of trade-press coverage from the same period. Manual coding of adopters' expressions of concrete realised benefits and challenges in the case stories is combined with a purpose-built dictionary that detects five temporal positions in adopters' framings of RFID performance. We find that associations between concrete outcomes and temporal framings are selective and stage-dependent. Concrete outcomes do not simply replace expectations, instead, they become associated with specific temporal framings over time. We also find that adopters' temporal framings are associated with trade-press coverage in stage-specific ways that change over time, often amplifying but also sometimes tempering broader attention to RFID. We contribute to research on technology framing and expectations dynamics by foregrounding adopters as co-constructors of technology discourse, challenging the displacement assumption with a stage-sensitive explanation of expectation–outcome dynamics, and introducing a replicable, dictionary-based approach for large-scale analysis of temporal framings in technology narratives.

Artificial intelligence and other general-purpose digital technologies often diffuse unevenly: large firms adopt early, while smaller firms delay adoption, contract, or abandon the active adoption option. This paper develops a survival-constrained theory of technology adoption in which firms choose when to adopt an irreversible technology while financing operations under uncertain implementation costs. Abandonment/exit is endogenous: firms leave the active adoption race when the value of preserving the adoption option falls below the passive legacy fallback (the value of abandoning the active adoption option, normalized to zero). The key object is an adoption survival frontier , a boundary in financing-cost–implementation-uncertainty space separating environments in which followers survive long enough to adopt from environments in which they abandon the active adoption option first. The Cournot block disciplines the price-pass-through component of follower payoff erosion; additional non-price appropriability losses, captured by a reduced-form term 𝜒 𝐴 (data accumulation, platform lock-in, switching frictions, and reduced access to post-adoption rents), remain reduced-form. Numerical characterization of the corrected nonhomogeneous stopping problem shows that a leader-induced regime shift moves the survival frontier inward and raises the risk that followers abandon the active adoption option before adoption. A welfare decomposition clarifies when this selection is efficient and when it reflects an accounting externality: a leader-induced payoff shift that the follower’s private stopping problem does not represent before the regime shift. We do not solve a dynamic adoption-timing game. Public EU aggregate data show that large-minus-small AI adoption gaps are measurable, but also illustrate that aggregate cross-country regressions are confounded by development gradients and cannot identify the firm-level survival-to-adoption channel; we treat this as a measurement-feasibility map for future linked-microdata tests rather than as a test of the mechanism. The paper contributes a computational theory of adoption survival and a transparent measurement-feasibility map.

We develop a dissertation-based methodology for measuring PhD populations and use it to present new evidence on who funds STEM PhD training in the United States, how many graduates are trained in areas of strategic national importance, and the effects of public investment in PhD training on PhD production. The U.S. government is by far the largest source of financial and in-kind support for STEM PhD training in America. We identify universities and fields where PhD training has high rates of government, industry, or philanthropic support, and the organizations and universities that fund and train the most PhDs in critical technology areas such as AI, quantum science, and biotechnology. Leveraging variation in government support across agencies and over time, we provide evidence suggesting that increasing government-funded PhD trainees increases PhD production additively or with modest crowd-in. To support further research, we provide public data at multiple levels of aggregation. These data and methods complement existing data collection efforts by national statistics agencies, producing information which is otherwise hard to collect or not systematically observed, and can be extended forward in time and to other countries.

Academic labs are central to the creation of radical-generic technologies. Yet translating these inventions into successful innovations poses a significant challenge as academic scientists often struggle to identify and prioritize viable markets in the early stages of commercialization. This study examines how individual scientists' opportunity beliefs evolve into the shared commercialization strategy of an academic spinoff. Using a longitudinal single-case study of the technology-market matching (TMM) process of a radical-generic technology, we analyze individual and team assessments of eight technology-market combinations over 15 years. We show that opportunity beliefs evolve from an initial emphasis on functional aspects of structural similarity toward an increasing emphasis on commercial viability, eventually integrating balanced similarity considerations. Within TMM, three mechanisms – productive cognitive dissonance, expanded similarity awareness, and team reflexivity – explain how heterogeneous individual beliefs evolve and coalesce into the commercialization strategy of the spinoff. We introduce “simmering-with-the-lab” as an organizational strategy through which latent technology-market options are maintained in the academic lab while the spinoff pursues a narrower set of commercial opportunities. The study advances theory on TMM and organizational learning for deep tech academic spinoffs and has practical implications for science entrepreneurship training, innovation policy, and university IP strategy.

Whereas prior platform research has mostly assumed standardized interfaces that are platform-specific, we examine how platform strategies based on third-party standards have become prevalent in a nascent ecosystem. Based on a longitudinal analysis of 97 smart home platforms, we analyze the value creation and capture benefits of four platform modes defined by two dimensions: the source of complements (first- or third-party complements) and the standardized interfaces used to connect these complements (first- or third-party standards). Our analysis demonstrates a trend from closed strategies (only first-party complements based on first-party standards) towards strategies that (also) rely on third-party standards. We find that startups and diversifying entrants make different strategic choices at entry and that firms increasingly combine platform modes over time to leverage their distinct value creation and capture benefits. Our findings contribute insights into the ecosystem conditions and pre-entry capabilities of firms under which strategies relying on third-party standards become viable and how this contributes to the proliferation of standards, platforms, and strategies in an ecosystem. We also theorize the time-dependent strategic choices of entrants navigating a nascent ecosystem and the complex complementary and competitive relations that emerge in an ecosystem with interconnected platforms as a result.

We report findings obtained by comparing two innovation surveys in Japan conducted in 1994 and 2020. A comparison of statistical means at two points in time, controlling for firm size and industry as covariates, reveals that the following three changes have occurred for appropriabilities and technological opportunities of innovation during the past quarter-century. First, the effectiveness of almost all the mechanisms by which firms' appropriate profits from innovation declined. Second, the imitation lag has lengthened considerably. Third, while universities and government research institutes have become more important sources of technological information, internal sources and competitors have become less important. These results suggest that Japan's innovation system has undergone a fundamental transformation between the two time points.

The absorptive capacity construct derives its functions, dimensions, and internal relationships from the implicit assumption that humans are the only learning agents underlying knowledge absorption. I posit that the emergence of artificial intelligence (AI) challenges this assumption and requires a re-examination of the construct. Through a theory-building abductive study combining conceptual mapping with qualitative interviews, this work investigates the complex relationship between AI and knowledge absorption. It theorizes new dimensions of absorptive capacity pertaining to AI-driven processes and analyzes the interactions between these new dimensions and the traditional facets of the construct. Ultimately, the paper seeks to reaffirm the importance of absorptive capacity in the age of AI and develop a nuanced understanding of AI's role in reshaping the construct's antecedents, mechanisms, and implications.

Innovation is cumulative and interdependent: successful inventions build on prior knowledge within technological fields and may also affect success across related ones. Yet these dimensions are often studied separately in the innovation literature. This paper asks whether patent success across technological categories can be represented within a single dynamic framework that jointly captures within-category reinforcement, cross-category spillovers, and a set of aggregate regularities observed in patent data. To address this question, we propose a model of interacting reinforced Bernoulli processes in which the probability of success in a given category depends on past successes both within that category and across other categories. The framework yields joint predictions for success probabilities, cumulative successes, relative success shares, and cross-category dependence. We implement the model using granted US patent families from GLOBAL PATSTAT (1980–2018), defining category-specific success through a cohort-normalized forward-citation index. The empirical analysis shows that successful innovations continue to accumulate, but less than proportionally to the growth in patent opportunities, while technological categories remain interdependent without becoming homogeneous. Under a mean-field restriction, the model-based inferential exercise points to a positive but non-maximal interaction across technological categories.

In this paper, we examine the role of hype in the discursive construction of emerging quantum technologies. Our findings reveal that, as these technologies transition from academia to industry, they are the object of ‘hyping’ in which the discourses of risk and responsibility figure prominently. We find that hype – a collective vision around which attention, excitement, and expectations escalate – emerges from a process of discursively constructing emerging technologies as risk management solutions to highly salient geopolitical and economic risks to state actors, while acknowledging that they may pose less salient, novel social risks to citizens and social groups within a state. We also find that the former construction overwhelms the latter, which has important implications for what actions associated with developing these emerging technologies are considered responsible. As a result, and building on the sociology of expectations, we develop a discursive model showing how hyping mobilizes risk and responsibility to shape the meanings attached to emerging technologies and, in turn, the expectations that form around them – with important implications for the direction and pace of technological development.

This paper evaluates existing approaches to identifying artificial intelligence (AI)-related patents and introduces a novel, scalable framework for improving classification performance. Motivated by growing reliance on patent data in innovation research, we assess widely used methods, including patent class-based approaches and the USPTO’s Artificial Intelligence Patent Dataset (AIPD), with an independent, human-expert-annotated ground-truth dataset. We document substantial performance limitations in existing approaches, particularly in terms of precision and generalizability. To address these challenges, we develop a CPC-informed, iterative positive-unlabeled (PU) learning framework for constructing high-quality training data. Our approach integrates hierarchical patent classification with data-driven refinement procedures to reduce label noise and improve representativeness. Using this refined dataset, we train a range of machine learning, deep learning, and transformer-based models. Our results show that models trained within our framework significantly outperform existing methods, including AIPD, achieving improvements over AIPD in F1 scores of approximately 18–21% on the same benchmark dataset. These gains are primarily driven by enhanced precision without sacrificing recall, highlighting the central role of training data quality in classification performance. We further demonstrate the empirical value of improved AI patent identification through two applications, showing that the release of ChatGPT increased both the market valuation of AI patents and firms’ allocation of innovative effort toward AI technologies. To support future research, we release our training data, source code, and patent-level predictions, with ongoing updates to reflect the evolving nature of AI innovation.