US Trails Stewardship Utah Campaign - Let's Hit Pause

I am asking that OpenStreetMap US put on hold the Utah Campaign, and any other campaign they might be anticipating. As noted before, my work has uncovered evidence that there are a tremendous number of errors in at least some of the government trail data. The risk is just too great that this defective data will leak into OSM if the organizers go forward with their campaign. While many of you reading this are probably very careful mappers, past experience has shown that not everyone is - even when explicit instructions are given.

When the announcement came out about the campaign, I was in the middle of assessing the USFS trail data for Dillon Ranger District near where I live in Colorado. I have now finished the first phase of that assessment, and I found a ~50% error rate! That is right, half the trails had errors! In many cases matching up the government data with the spatially correct data in OSM can only be guessed at as the geometry is so different. I realize that the Dillon RD is in Colorado, and the proposed campaign is in Utah, but based upon my previous work - which also uncovered many problems - I strongly suspect that at least some of the data in Utah will be just as bad. I made this video overview of the situation: This initial assessment didn’t cover trails missing from the USFS data, nor errors in names, trail numbers, or allowed usages - but I know from past work that these types of errors occur too. I would hate to have the tagging in OSM that was carefully and painstakingly collected through field surveys overwritten by this bad data.

After making that video, I also discovered that in some cases a trail that the USFS says is managed by one ranger district, is 90 miles away in another district! In some cases trails are duplicated with multiple ranger districts claiming them. In some cases the geometry is duplicated, but the tagging (“attributes” in GIS speak), don’t match.

If this organized editing activity were to be proposed by an ordinary mapper, it would probably be shut down due to the poor quality of the data. Why should this be different?

13 posts - 6 participants

Read full topic

Ce sujet de discussion accompagne la publication sur