The data management plan has a familiar life story, and for much of its history an unhappy one. A researcher writes a plan to satisfy a funder’s requirement at proposal stage, describing data that does not yet exist. The plan is submitted, the grant awarded, and the document filed away — never opened again. By the time real data begins to flow, the plan is already a work of fiction: the formats changed, the volumes grew, the consent arrangements were refined. A plan written once and never revisited describes the project that was imagined, not the one that happened. The living DMP — a plan that updates dynamically across the lifecycle of a project — is the response, and it belongs to the machine-actionable DMP domain of the CASRAI Dictionary.
The trouble with the static plan
The deepest flaw in the traditional DMP is one of timing. It is required precisely when the least is known — at the proposal stage, before the work has begun — then frozen at the moment it is most speculative. Research is not like that. Data management decisions are made and revised continuously: the instruments produce more data than expected, an ethics review changes how participant data must be handled, a new repository becomes the obvious home. A static plan cannot reflect any of this. It becomes, at best, a historical curiosity and, at worst, a misleading record nobody trusts. The energy spent writing it is largely wasted, because the document never connects to the reality it was meant to govern. The static plan fails not because planning is pointless but because a plan that cannot change cannot stay true.
What makes a DMP “living”
A living DMP treats the plan as a dynamic document that evolves with the project rather than a fixed deliverable. It is created at the start, as before, but expected to change — updated as decisions are made, as data is produced, and as circumstances shift. The aim is for the plan to remain an accurate description of how the project’s data is actually being managed, useful to the people doing the work rather than written only for an external reader. A living plan can answer real questions: where is this dataset stored, under what licence will it be shared, who is responsible. Because it stays current, it can guide practice, support handovers, and provide an honest record at the end. The shift is from plan-as-document to plan-as-living-record — from something written to be filed, to something maintained to be used.
Why machine-actionability is the key
Keeping a plan current by hand is a burden few will sustain, which is why the living DMP depends on machine-actionability. A traditional plan is prose: to update it, a human must edit text. A machine-actionable DMP (a maDMP) expresses its content as structured, machine-readable information, and this changes what is possible. Updates need not be manual: when a dataset is deposited, when an identifier is minted, when a project record changes, the plan can be updated automatically to reflect what has actually happened. The structure also lets the plan be checked — systems can verify whether stated commitments have been met — and lets it exchange information with other systems. Machine-actionability is what makes “living” sustainable: the plan keeps pace with the project without depending on someone remembering to rewrite a document nobody wants to maintain.
The RDA DMP Common Standard
For plans to update automatically by exchanging information with repositories, funder systems and institutional databases, those systems must agree on how a plan’s contents are represented. This is the contribution of the RDA DMP Common Standard, developed through the Research Data Alliance: a common, machine-actionable model for the information a data management plan contains. By defining a shared structure for the elements of a plan — the datasets, their characteristics, storage and preservation arrangements, licensing, costs and contributors — the standard lets a plan be created in one tool, understood by another, and updated by information arriving from a third. Without it, every system would represent a plan differently and automatic exchange would be impossible; with it, a living, dynamic DMP can flow between the systems that read and update it.
Integration with repositories and CRIS
The living DMP only realises its value when connected to the systems where the real activity happens. Two integrations matter most:
- Repositories. When data is deposited, the repository holds authoritative information — identifiers, formats, access conditions. A connected DMP can be updated from this directly, so the plan reflects what has actually been deposited rather than what was once intended.
- Current research information systems (CRIS). A CRIS holds the institutional picture of projects, people, grants and outputs. Linking the DMP to the CRIS lets the plan draw on and contribute to that picture, keeping data management visible alongside the rest of a project’s record — a concern of research administration more broadly.
Through these connections the plan stops being an isolated document and becomes a node in the research-information landscape — reading from and writing to the systems that record what a project is doing. This is what turns the machine-actionable plan from a clever idea into an operational reality.
A consistent vocabulary for plans that travel
For a living DMP to exchange information with repositories, funder systems and a CRIS, the elements it contains must mean the same thing in every system it touches. A licence, a retention period or an access condition recorded in the plan must be understood identically by the repository that updates it and the CRIS that reads it. That consistency is what the CASRAI Dictionary provides: a shared vocabulary so the contents of a machine-actionable plan are understood the same way wherever they flow. And because the people who steward a project’s data make a real contribution, that work can be described in the same shared framework — the CRediT taxonomy and its Data curation role. The static DMP described the project that was imagined; the living DMP describes the project as it really happens — and stays useful from proposal to completion.