Have you ever wondered if your perception that the Object Constraint Language (OCL), and more specifically, the research around it, is quietly fading away is actually backed by the data? Well we did, and decided to run a metascience study of the OCL research landscape to see whether our feelings were right. In what follows, we give you a summary of our findings. You can also read the full paper. The work will be presented at the OCL 2025 Workshop, so I’m sure we’ll be able to enrich this post with additional insights from discussions happening during the workshop.
Numbers Don’t Lie: OCL’s Steep Decline
We analyzed 2,376 OCL-related publications spanning from 1998 to 2024, and the trends are unmistakable. After peaking around 2006, OCL research has been in steady decline across virtually every metric that matters:
- Publication Volume: The total number of OCL publications has dropped significantly, with conference and workshop papers nearly disappearing. Today’s OCL research revovles primarily on a handful of journal articles and an even fewer remaining number of conference proceedings.
- Community Size: With more researchers leaving the OCL field than entering it each year, the community is shrinking. And new ones do not stay long enough to become recurrent authors. Indeed, only 54 recurrent authors remain active in the field.
- Research Impact: Perhaps most important, the citation impact of OCL papers has declined sharply since 2007. The most cited OCL research were published between 1998 and 2007. Since then, newer publications struggle to achieve comparable recognition.
Context Matters: OCL’s Decline in the Broader Software Engineering Landscape
OCL’s struggles don’t exist in isolation. The broader modeling community experienced similar challenges until recently finding new life through the low-code movement. This parallel trajectory raises important questions about whether OCL might benefit from similar revitalization efforts once the low-code platforms understand the needs to express business rules to generate precise software systems, something that we did add to BESSER.
Our study also highlights a persistent challenge: the lack of robust tooling. With B-OCL being one of the few recent OCL-focused projects, the community faces a classic chicken-and-egg problem—without good tools, adoption lags; without adoption, tool development stagnates.
The Community That Refuses to Die and a Path Forward.
Despite these sobering statistics, the truth is that OCL remains alive, supported by a highly collaborative community with opportunities for revitalization. Unlike other software engineering trends that burned bright and fast (the authors cite aspect-oriented programming as an example), OCL has demonstrated remarkable staying power. Some initiatives that could help in the long-term survival of the community could be:
- Cross-Pollination with Related Communities: The Semantic Web community’s SHACL constraint language shares conceptual similarities with OCL. Building bridges between these communities could expand OCL’s reach and relevance.
- AI and Machine Learning Connections: The current AI boom creates new opportunities for constraint languages in data validation, model verification, and ML pipeline management. OCL researchers could position themselves at this intersection by targeting AI-focused conferences and exploring applications in machine learning contexts.
- Lower costs for new tool developments: The emergence of AI coding assistants might lower barriers to OCL tool development, potentially addressing the long-standing tooling gap that has hindered adoption.
A Call for Community Action
This study serves as both a wake-up call and a roadmap. Our data confirms what many suspected—OCL research is indeed in decline. But our analysis also reveals the community’s resilience and identifies concrete opportunities for renewal.
The question now is whether we (the OCL community) will use these insights to adapt and evolve. We believe this is a critical moment for the OCL community to decide on its path forward.
What do you think? Do you agree with our analysis? Do you have other ideas on the reasons behind the decline or the path forward?
And have you observed similar patterns in other modeling language communities? How might we apply these insights to strengthen the broader ecosystem of modeling languages?
OCL has shot itself in the foot with its general form, which is still based on the predicate logic of the first order.
One of the secrets to the success of relational DB specification languages is the use of simple patterns. Instead of writing down a formula, the functional dependency is represented as a pattern. This appears to sacrifice universality. This was then spoiled by the introduction of more than 60 constraint classes, which also have to be handled separately. In practice, only a few remain. This may have satisfied mathematicians, but it shocked practitioners. In practice, the entire theory is then taken at face value.