My main interest in OSM is attempting to map public rights of way in the countryside such as bridleways and footways. I have read advice that a way such as a cycleway should only be mapped if someone else can verify its existence somehow – such as physical signs on the route. However, as anyone knows who has tried to follow footpaths in the countryside, many PROWs have no visible presence at certain times of the year – after ploughing or sometimes through fields of tall wheat or rapeseed. There are often no field edge markers which indicate which route you are supposed to take. If you are lucky, those who have gone before will have started to make a trail across a newly planted crop, which then tends to become the accepted route for that part of the footpath. Very often walkers consult their Ordnance Survey map before they strike out across an unmarked field, although the OS map may not be up to date with the legally-binding Definitive Map held by the County Council.

So hopefully you can see where I'm going with this. The aim, as per the Wiki UK_Countryside_mapping is to represent ways with a legitimate “foot=yes or foot=designated” tag. In the absence of signposts, in practice the guide for UK walkers is the OS map, but if I walk a route which I believe follows the OS map for the purposes of a GPS trace, is this not using derived-data? Many people will walk round field-edges where they find the PROW blocked by impenetrable crops – but the arbitray nature of this clearly makes it unsuitable for OSM.
If you don't map bits of a footpath because of lack of signposts, then foot-routes between villages etc lose their function as a route, and become disconnected isolated snippets of path, which seems to me pretty pointless mapping.

Is there a solution to this?

The answer is NO.

It is pointless tracing footpaths from NPE data. Over 50 years, footpaths have been closed, diverted, amalgamated etc and new ones established in line with the changing features of the countryside.

Because of the poor upkeep of footpaths in many parts of the country, it's often impossible to tell which side of a stream, wood or hedge the path travels. If not completely overgrown, the physical trail is often no more than a rabbit track! Because of the ambiguity, it doesn't seem logical to suggest that seeing other walkers on it means that it's 'permissive'. They could be trespassing too. One UK OSM'er suggested that if a farmer with a shotgun wasn't there telling you to get orf his land then where you were walking could be mapped as a public footpath. An unbelievably crass and glib response.

However, my question was not so much about where you can see a path but can't tell if it is a legal right of way or not without looking at the OS map.
It was about when the physical evidence of the path fizzles out and the only way you know which way to proceed is to look at a copyrighted map. Another experienced OSM mapper suggested drawing a straight line between the last known physical evidence point, and the next one, and let the landowner object if it's wrong! This doesn't seem satisfactory to me – that line could cut through crops, private property, industrial plants, protected habitats etc. It isn't reasonable to declare that a particular way exists and put the onus on the landowner to refute it. Assuming we expect the map to be used in practice to navigate, then putting straight paths across countryside without knowing that they exist or have ever existed along that line, is simply inviting people to trespass.

I'm coming to the conclusion that OSM will never map GB rural walking routes at a level of accuracy required by a hiker. Even where legal rights of way exist, there will be significant breaks in the route where it is impossible to provide non-copyright verifiable evidence of the PROW.