large (i.e. in complexity and requirements not tickness) standards are profoundly anti-open source and therefore anti-competitive.

Over the time individual communities have developed several open domain standards. As these standards tried to cover each and every aspect of the domain they became more complex, during standard evolution much attention went to converge and requirement rather than optimization. Developing large standards typically requires collaboration among multiple individuals or groups with expertise in specific areas, with each participant contributing only a part of the standard. Therefore, Modular standards are definitely attractive for distributed development. Instead of a single, centralized standard it will be better to have multiple distributed sub standards covering parts of the domain with a core standard. Although several standards adopted a modular or layered architecture to describe these standards by specifying core and secondary standards, but there was never a size constraint. Modular standards offer several advantages such as reuse and extensibility, every time a new technology or requirement arise it will be wiser to make changes or development in secondary specifications layer without touching the core. Many domain standards have already adopted this kind of modularization, most notably Chemical markup language (CML) which has been totally modularized starting with Version 2. The CML core components are now available as CMLCore, while there are various secondary specifications such as CMLReact, CMLComp, CMLSpect, CMLQuery to support the specific requirements. Similarly SBML community is moving towards a modular architecture and the next Level of SBML (SBML Level 3) will be modular, with a defined core set of features and optional packages adding features on top of of the core such as Layout, Rendering, Geometry. Requirement for modularization is not limited to standards only, there is urgent need to modularize the ontologies as well. Modularity is a key requirement for large ontologies in order to achieve re-use, scalability, maintainability, and evolution. Rick also suggests that any standard which is hard to implement for reason of size and complexity, should be just abandoned as that is not appropriate for an open technology.