chap, part of the problem, and one I've banged on about a number of times here, but it bears repeating, is that people seem to forget that cycling is predominantly used for short trips, so any additional deviation and delay on a route very rapidly reduces the number of cyclists likely to use it. 90% are under 5 miles, 55% under 2 miles, 20% under 1 mile (NTS 2006, Table 3.4).
This is of course well known, but seems to be frequently forgotten, partly I think because those actively engaged in lobbying for cycling, whether of the CTC or Sustans persuasion, are usually keen cyclists and are likely to undertake longer cycle trips than the typical cyclist. This means that far too many of those involved in planning 'networks' are doing so from an entirely unrealistic perspective of the types and lengths of trip that their target audience are likely to undertake. So forget long, tortuous routes round the back streets that add a mile and maybe another ten minutes in an urban area: this sort of thing rapidly eliminates any time advantage cycling might have over driving in particular, and won't therefore attract significant usage outside the minority who want to cycle for cycling's sake.
There's a NCN railway path near where I live, a very nice one in fact that I helped raise funds and lobby for and am very pleased to have. But as it enters the town it takes an indirect route to the main shopping centre, the most direct route being to leave the path at that point and ride about 0.75 mile down a fairly quiet but direct street. However, at the very place where you would leave the path there is a NCN signpost sending you via the NCN route, saying "Town Centre 1.5 miles". That's twice the distance of the direct route, a distance greater than probably 40% or so of all cycle trips... sadly that NCN signpost is actually discouraging people from walking or cycling to the town centre, but is a not untypical example of what happens on official cycle 'networks'.