[HOT] Task 661 (Ebola, DRC), road vs path

[Erik Walthinsen]

On 10/12/2014 11:23 AM, Charlotte Wolter wrote:
> Despite the fact that most hamlets are reached
> by something as wide as a vehicle. it doesn't seem to be the case for the
> houses at the upper left.
> Also it doesn't make sense for a road to become a path and then a road
> again (though, I guess it's possible).
> I think it's likely that the houses at the upper left are reached only by
> a path. The open areas in the front of them are probably ground beaten down
> by repeated foot traffic, not by vehicle use.
> That's my take. I wonder what others think.

I'm the other instigator of this thread (via private message), so here's 
part of my take on it, copied from the original thread:

----
I tag things as path when either a) I can very clearly see that there 
are not two visible ruts (e.g. @node 3044228914), b) I can estimate the 
width of the route to only a few feet, c) it's so narrow that the trees 
occlude the majority of the path and is sometimes only visible by the 
fact that the trees change appearance along the route (most of the 
routes in the task), or any combination of the three. I take into 
account how many dwellings are present along the route and the 
likelihood that a vehicle has ever traversed it.
----

I realize the 'two ruts' may not be particularly accurate, but that's 
almost never the sole determining factor for me either.  I also consider 
the 'jinky-ness' of the route: in many cases the visible path is so 
jagged from side to side that there's very little chance that it's used 
on any regular basis by vehicles.  Any degree of 'traffic' would cause 
that jaggedness to disappear, and straighten out the route to some 
extent.  Given the level of vegetation we're looking at, any road that 
hasn't been driven on in recent history has ceased to be drivable at all 
(without some serious effort).

I base this on my experience mapping logging roads in Oregon, which I 
have done mapping work on both before and after hiking them during 
hunting season.  With similar levels of imagery I can compare the roads 
and paths I've hiked to what I see on these tasks.

Additionally, I'm taking into account the current primary users of this 
data: aid workers.  I would rather be somewhat pessimistic and have the 
aid workers prepared to traverse these routes on foot, than have them 
assume that everything is drivable and find out when they get there that 
it isn't.