TIGER source appears to be invalid, what to do?

While ensuring all roads are named properly on a road system near me, I noticed some oddities. Some parking lanes were given proper names, whereas proper streets were lacking… based on personal knowledge of the area, it also seemed wrong. I looked at the roads in OSM, and they’re associated with TIGER data.

I looked into the TIGER data, and it appears it’s wrong for a couple of roads, having offset them. I’m going to take some time and go to the roads in person to see what is actually going on. If I find that the roads are in fact wrongly named (based on bad TIGER data), do I simply move the TIGER tags and road name over to the proper streets?

It appears you are in a “TIGER Desert”, that is a place that hasn’t been cleaned up much or at all since the data was imported.

The good part of the TIGER import was that it got some sort of road network for the United States into OSM so that it became somewhat usable quickly. The bad part of the TIGER import was that the data quality was not very good:

• Private driveways and farming tracks often show up as unnamed residential roads.

• The geometry is often off.

• Road names are suspect.

• Etc.

• The experience with the TIGER import soured many OSM contributors on the whole concept of importing data. So while the import process has changed to put in place procedures to more carefully vet the data and to do quality checking after the import there are now many mappers who become quite upset at any possible future imports.

As a local to your area you will know far better than the bad, old TIGER data what is “on the ground”, your clean up efforts will be appreciated by lots of people you will never meet (more and more websites and navigation apps use OSM data).

I’ve cleaned up areas that I visit (fortunately other mappers did lots of clean up where I live before I started mapping). It turns out that even Google used either TIGER data or something very much like it and have areas with the same errors as OSM. I get a kick out of seeing areas that are better in OSM than in Google, especially if I was the one that cleaned up the TIGER data in OSM for that area.

In some cases it seems easier to delete large swaths of roads and just re-enter them. If possible, it is better to move a incorrectly located TIGER road to the correct alignment and fix the tags on it than to delete and replace: If you delete the highway it makes the history of the edits on that road very much more difficult to follow.

In the mean time take comfort in that while you are learning to map in OSM and likely will be making some mistakes, that even with your mistakes the map will be better than when you started.

And thank you for cleaning up more TIGER data!!!

Here is the link to the wiki: https://wiki.openstreetmap.org/wiki/TIGER_fixup

Regarding the ‘Wrong highway classifications’ that is also mentioned in the link (and we have recently stumbled over) I have a question: wouldn’t it make sense to re-import the Tiger data, where it is still unchanged and use highway=tertiary instead of the wrong highway=residential if it is outside of an urban area?

It would no longer make sense to re-import TIGER data, because there have been way too many edits, even around ‘TIGER deserts’. Rather for the case of mis-classified roads, it is better to review and update area roads in an editor.

Exactly. Generally the areas where there haven’t been any edits are those where the latest TIGER data is still terrible.

And definitely not highway=tertiary as a default - no no no. The default for the A41 class should have been highway=road in rural areas (effectively as a ‘fixme’), highway=residential in urban areas. But we are where we are and hindsight is a wonderful thing.

@n76: I would just dispute one comment in your post. TIGER data wasn’t usable for routing until at least the summer of 2009 as ways were not connected across county boundaries. Also roads crossing on bridges were connected to the road underneath (as well as to pipelines, power lines etc). (MASSGIS was worse because this was imported at a finer granularity (towns)and wasn’t interconnected either). See the old wiki page: https://wiki.openstreetmap.org/wiki/TIGER_fixup/250_cities.

@karussell: it’s not just the fact that a lot of stuff has been edited: no-one, has in reality, demonstrated a viable conflation method for this type of data.

@karussell: it’s not just the fact that a lot of stuff has been edited: no-one, has in reality, demonstrated a viable conflation method for this type of data.

Not sure if I understand this correctly. With the words from Richard I would just implement a big for loop over all highway=residential from the US, check if they are edited ‘recently’ and have a tiger ID and then change the highway to ‘road’. Also the conflation is not that complicated now, I think, as the tiger ID is stored in OSM.

And how do you know if nobody modified it in all this years because it is a residential road?

It’ll have “tiger:reviewed=no” on it: http://overpass-turbo.eu/s/qmM .

It was only recently that I started to delete all the tiger:* tags when I decide the road is fully cleaned up. In the past I’d simply remove the tiger:reviewed tag if I had field checked the situation. If I have only used satellite imagery, I leave all the tiger:* tags in place as I can’t verify the street name and so I can’t claim all major details are correct. I may have added surface=, lanes= and even turn:lanes=. In cases where the speed limit is painted on the pavement I might have added maxspeed=.

Most importantly, From satellite imagery I can often determine the highway=* classification at least well enough to distinguish between tertiary, residential, unclassified, service and track. I we were to follow your proposed rule, then many of the roads that I cleaned up years ago and left as residential because the Bing imagery indicated that was correct would be mechanically retagged as highway=road simply because I could not veryify the name. And even if I removed the tiger:reviewed tag because I had verify everything, the presence of a tiger ID would trigger your proposed replacement of residential with road. That is the type of thing that gives automated fix ups such a bad reputation.

Thatswhy I said ‘check if they are edited ‘recently’’. Or even only if the last change was from the tiger import

Similarly, there are many roads that I have given a ‘flash review’ for obvious errors, but never edited because they were nearly OK geometrically and my goal at the time was a different edit type of a nearby feature. Since I didn’t make any edits, they would be subject to the mass retagging. The counter-action to any mass retagging would be more mass retagging to regain usability of most of those roads.