The concept of granularity originally refers to a property of matter consisting of small parts or "granules". By extension, it may also refer to the choice of an observer to concentrate their attention on larger or smaller parts of an object of investigation, as appropriate for a certain purpose or in search of meaningful interpretations and connections, usually through abstraction.
Granularity concerns the ability to represent and operate on different levels of detail in data, information, and knowledge that are located at their appropriate level. The entities are described relative to that level, which may be more coarse-grained or concern fine-grained details. Devising these ordered levels of granularity in a granular perspective are either determined by the laws of nature or are a resultant of human cognition to divide the data, information, or knowledge. 
Granularity in complexity
Often used in the phrase decompose to the lowest level of coherent granularity is not the same thing as component analysis which could be a subset of it, as grains interlock, interact and clump. But we also need to aware of the context and in the complicated and clear domains the lowest level of coherence may be a process or component, but even here these may be composed of other tasks/items.
In the complex (and chaotic) domain, the choice of an appropriate granularity makes novel connections possible at a higher level of abstraction. This stands, for example, at the basis of exaptation.
The design principles for managing complexity refers to the need to adapt the granularity that we are dealing with, that is dealing with an appropriate level of abstraction. This generally means that working with finer-grained, that is smaller objects but we need to consider both the scale and modularity.
Granularity, resolution, and scale
The three concepts of scale, resolution, and granularity are distinct and related, Scale has to do with the proximity and breadth of the point of view. Resolution has to do with the ability to discern small parts. Granularity has to do with the choice of the relevant level of detail. Adequate scale and resolution are usually required in order to observe things at the right level of granularity.
- Maps (see picture)
- Beach, gravel, sand grain, mineral, crystal, atom, etc.
Granularity and modularity
Modularity as a property of things refers to the possibility of reuse in different configurations, and in a variety of different contexts and purposes. Modularity is usually possible at a certain level of granularity.
Depending on the nature of a system and its parts, the process of breaking it down into components and re-composing the latter into a new configuration may or may not be reversible. Within the complex domain, parts typically change their nature in the process.
- Individuals moving across teams and organizational units may only partially retain their identities and behaviors
- De-composition and re-composition of methods into other methods and assemblages (Note: not sure this is not aggregative)
Aggregative modularity applies when the whole is the sum of the parts and disassembly does not prevent reassembly. This happens when the relationships between parts can be established in repeatable ways, with predictable results. In terms of the Cynefin framework, this corresponds to the "ordered" domain.
- Bricks in a Lego™ construction toy
- Individuals in roles within crews (Note: not sure)
- Maps such as those shown above
- Patches in a patchwork artifact
- Wiki pages
Granularity and information
- Pixels and picture resolution
Other Related concepts
- Assemblage theory
- Dependency matrix
Related Principles and frameworks
- Anthro-complexity heuristics
- Design principles for managing complexity
- Principles for managing knowledge
- Principles for scaling a complex system
- Dave Snowden, Dependency matrix, Cognitive Edge Blog (November 6, 2014), Granularity and ASHEN in connection with the Dependency matrix
- "Granular", in Oxford Advanced Learner's Dictionary
- Encyclopedia of Systems Biology - https://link.springer.com/referenceworkentry/10.1007/978-1-4419-9863-7_65