Figuring out a routing problem

Here is the problem: An automobile route from Wickenburg, Arizona to Kingman, Arizona does not take the shortest path (about 130 miles) but rather detours to one over 200 miles long. Here is a link that replicates the problem: OpenStreetMap and an image:

I see this on a number of different OSM based routing engines so I think the error is in how some road segment is tagged rather than a specific routing engine.

I have loaded at the route relations for US-93 (the direct route) into JOSM to get see if there was an obvious break in connectivity or maybe an access tagging issue but nothing obvious jumps out at me. I have chased down some routing issues before that only involved a few road segments but the techniques I used don’t scale well to looking at the 400 or so objects along this route.

Is there a way to narrow down the search for why the routing is not working correctly other than brute force examination of each object along the way?

Here seems to be some road construction which has been mapped imprecise.
oneway direction should be checked.

oneway roads seem broken in this area

found by dragging start/end closer as long as they were producing bizarre results

opened Note: 5513653 | OpenStreetMap to not forget it

I think w0x0f tried to work on this already, see Changeset: 188599607 | OpenStreetMap

E.g. the opposite direction works: OpenStreetMap but the northbound directions had problems.

Sometimes the debug tiles by osrm also help, e.g. zoom and pan around the 93 here: OSRM Debug

if my “found by dragging start/end closer as long as they were producing bizarre results” is not helping - let me know and I will produce video of this method in action

i think that road was repaired from construction to normal state in Changeset: 188599607 | OpenStreetMap, e.g. if you zoom in more I see this Note: 5513653 | OpenStreetMap

Edit: damm, stupid me, I overread the one-way and that seems to be the problem here.

So pratically it is this part of the highway: OpenStreetMap ? @n76 If you have local knowledge and know that the northbound lane is open, a “change direction” on that oneway part there should maybe help.

No doubt way/1087768923 direction was wrong so I reversed it.

https://osmcha.org/changesets/188890481

I probably botched a direction in there. If it’s set so traffic can proceed in both directions on 93 between Kingman and Wikieup, it should be okay.

If you want a laugh, look at how it was before my changes, with one direction entirely broken!

Thank you all for your help.

I should have thought of the “move the ends closer together until the problem is isolated” method. I will try to remember it for next time I come across this type of situation.

bonus hint: first try big moves, only later move slower
do not drag at z20 for 200km

similarly, in more general case: when you have X items and they cause problems, and problem is likely with single one: it is better to initially check with about half of them and you know which group is causing issues

rather than checking one by one

(basically, binary search)