144. For inter-nation interoperability purposes, it is only necessary to standardise the definitions of information exchanged between systems in different nations and, even then, it is sufficient to standardise external schema representations only.

145. There are other reasons why more extensive standardisation of data might be advantageous (e.g. application portability) and these concerns are relevant to NCF systems.

146. To support unanticipated exchanges of information (for which provision was not made at design time), it would clearly be beneficial if the format and meaning of that data had been agreed in advance. However, such universal standardisation is highly costly and its benefits are at best difficult to quantify (and at worst might never materialise). Speculative standardisation also has the disadvantage (demonstrated repeatedly by previous 'open' computing initiatives) that the standards produced are seldom usable; by its very nature the process for production of such standards is pre-emptive and therefore unable to take into account specific requirements. Subsequent failure to adopt those standards leads to their being undermined and eventually displaced altogether by de facto substitutes.

147. Nevertheless, the increasingly unpredictable nature of military operations, together with changing concepts of operations demanding increased interchange of data from the strategic to tactical levels, suggests that local interchange agreements (e.g. within a nation, bilateral between systems) will become increasingly unsustainable. Emphasis will need to shift towards NATO-wide agreements, encapsulated within the NISP.

148. These issues, and the wider role of data management within NATO must be reviewed. The eventual scope and form of the NISP will therefore be influenced by this work.

149. It has been suggested in some quarters that object-based approaches providing data encapsulation ('data hiding') can obviate the need for the widespread data standardisation to support remote data access, and thus that any other approach to data management may be inappropriate once extensive use is made of object-based mechanisms. Both of these assertions are true to a degree, but need qualification. This Data encapsulation does not remove the need to standardise the data that is exchanged between systems. Moreover, encapsulation only limits the degree of inter-system data agreement needed if the specific requests for information exchange (the 'methods' in object terminology) can be determined in advance. Therefore, even if object mechanisms are widely employed it will still be necessary to standardise core data elements relevant to inter-domain interoperability within the NISP, and to pre-determine the nature of requests for information exchange.