I don’t think that a canonical “driveway” and “track” are equal either - just that in the real world things are often a mix of both. It’s true that service roads are more likely to be accessible by regular cars than agricultural tracks, but again that’s not a hard and fast rule (the OSM wiki has an example paved track on it partly I suspect for that reason).
In England and Wales, what do you think is a “navigable road”? In the absence of explicit access tags, you’ve got exactly the same default access rights on a service road as on an agricultural track (none). If you’re delivering a parcel (for example) then in a sense you’re “invited to access” a property. In the absence of surface, smoothness etc. I can understand why a router would prefer a service road to an agricultural track, but that’s based on customary usage and likely vehicle appropriateness, not legal right of access.
Yes - and thanks for that. An example which benefited from this approach was here. There, there is a clear mismatch between on the ground data (although the signage isn’t great) and OS OpenData (that shows a change between 2016 and 2024, but still doesn’t match signage). The resulting “best guess” was only possible based on survey; there’s no way that some random MapRoulette user with no familiarity with the local area (or perhaps even the country) could do that.
To be clear, the first part of the sentence that I wrote that you omitted was “in the event of no action by Microsoft”.
Would you complain if 9.4m of your private driveway leading up to your house was changed to be a public road? I certainly would!
A good first step for Microsoft here would be to revisit all the changeset comments that are of the form "Reclassification of the roads in GB #MSFTOpenMaps mpr.lt/c/48118/t/229418319 " and add a disussion comment saying what they changed and why., and based on what evidence (which imagery or other sources did they look at, for example).