Hi Gert,
I have seen my share of network plans made totally without understanding for bits, hierarchy or actual *networking*, resulting in "oh, for these 500 sites, we definitely need a /24!" (and "oh, for all the electronic passports for 100 million citizens, we must have a /19!") - and thus it is good practice to have someone more experienced in addressing review the plan and see whether it makes sense.
I think that is the general point that Silvia is making... It is very difficult, if not impossible, to come up with a one-size-fits all approach to assessing requests for address space, particularly when trying to cater for organisations that don't quite fit the usual mould. Are approaches such as 'up to one extra bit per hierarchical level or geographical segment' compatible with this premise and are they even necessary? I know from personal experience of assessing hundreds of requests for IPv4 address ranges over the years within my own organisation that there is no substitute for experience when it comes to performing the task effectively. Whilst rules of thumb are useful I think that attempts to 'proceduralise' the task with more specifics can end up being unhelpful and, in any event, are not necessary prerequisites for consistency. In my view what is more important is general oversight and capturing of experience garnered over multiple requests and it is noted from the IA that >/29 requests will continue to follow the escalated evaluation process which ought to help provide this. To be clear about where I stand; I am still satisfied that the revised policy and its proposed implementation will meet the needs of the UK MOD however I would nevertheless support a more 'liberal' approach to the consideration of the varying requirements of other organisations who will undoubtedly have different - yet equally valid - priorities and needs. Regards, Mathew