Consolidating values of tiger:reviewed=

Over the past 15 years folks have been quite creative with values in this tag. It would be great to get a consensus on which tag values are most useful and proceed in cleaning up the rest. It is possible I have missed some place where this is already written down, please let me know! My hope is to get things reduced down from the 170 values currently in the db.

As a starting point here’s my proposal for acceptable tag values, I will update it as discussion commences:

  • tiger:reviewed=no - original tag value
  • tiger:reviewed=yes - has passed tiger review but the reviewed did not want to remove all tiger tags for whatever reason
  • tiger:reviewed=aerial - the position of the roadway has been review using aerial layers as noted in the sources in the changset
  • tiger:reviewed=name - the name has been corroborated with another source, position has not been reviewed

Examples of current values that will be coalesced mechanically:

  • Yes, YES, y, ye: these will become yes
  • No, n, no; : these will be converted back to no
  • arial, aerial;no, aeriel, aearial, position: these will become aerial

Examples of current values that will be reviewed and updated to aerial:

  • Bing
  • yahoo
  • USGS OrthoPhoto
  • geometry
4 Likes

This sounds sensible.
Having reviewed name and geometry (by current aerial) then be “t:r=aerial;name”?

In the case that you have verified the name and the geometry, you can probably just go ahead and remove the TIGER tags entirely. Also might be good to tag the road surface if you know that as well.

1 Like

Looks like a reasonable plan to me.

I’d say let’s just delete these ones. Is there a reason to keep them around?

1 Like

I would treat any value as equivalent to tiger:reviewed=no, so as far as I’m concerned, they could all be consolidated to a single value. In particular, tiger:reviewed=yes doesn’t give me much confidence in the prior review of the feature.

However, preserving the information that someone previously considered some aspects of the feature might be useful. So, I think your proposed values look good.

For the “yes” values. I know some folks really liked keeping the status after they reviewed it. I find it a bit silly and would remove them (and the rest of the tags on those), but I think that’s a separate proposal.

2 Likes

I just delete them all especially if the road is fundamentally different. There is no one at TIGER ingesting OSM data back and doing QA on this.

4 Likes

I think that’s find but have made another topic for this exact question: https://community.openstreetmap.org/t/proposal-for-mechanical-edit-remove-all-tiger-tags-for-items-with-tiger-reviewed-yes/

Trying to keep this topic focused on cleaning up values of tiger:reviewed

1 Like

I started working on this a bit. The big changeset is: Changeset: 148080213 | OpenStreetMap

I didn’t touch the “position” value yet as it has ~22k uses (go Maryland!).

I’m working through the slew of smaller buckets… maybe get that in tomorrow. Anyone want to advise me on:

  • “no;yes” (275 uses) - I think this should be aerial looking at a sampling of them

No-one at the Census Bureau is consuming it, but others might be (and in fact are).

tiger:reviewed=no on rural highway=residential is a signal that the road hasn’t been significantly touched since being imported from class A41. It’s a very useful factor in developing a heuristic for “passable rural roads”.

If you haven’t reviewed the road - which essentially means checking that it is indeed a road, that the geometry is mildly accurate, and that its character doesn’t contravene what you’d expect for that highway type - please don’t remove the tag. Thank you!

I don’t see anyone advocating for removing tiger:reviewed=no mechanically. Perhaps I have missed it?

As a data point, I (personally) delete the tiger:reviewed=no tag after I have given what I consider “comprehensive review” to the way. I advocate that it not be changed to something like tiger:reviewed=yes or something even stranger (and not easily parsable) like tiger:reviewed=aerial. See the quote below.

To wit, I have written (and keep updated) United States/Railroads - OpenStreetMap Wiki , a section specifically on “Editing Railroads starting from TIGER data,” since TIGER imported literally hundreds of thousands of miles (>400000 km) of rail data. These have been and continue to be rather nicely cleaned up since 2007-8, though work to complete this (easily tens of thousands of miles of rail) remains yet to do. Estimates are “about 75%” of TIGER-imported rail have been Reviewed like this; some might say “not bad” for 15 years of effort. (Thanks to all who do this, “it takes a village.”)

Specifically for rail, this wiki section suggests:

If a tiger:reviewed=no tag exists and you “reviewed” the rail (e.g. comparing against aerial/satellite imagery and better aligning where necessary, assuring tags are all correct…), delete this tag — don’t change it to the superfluous tiger:reviewed=yes or something like tiger:reviewed=aerial.

I continue to delete tiger:reviewed=no tags after such comprehensive review by me (including on roads / streets, especially local to me / in my county), but it is unknown to me whether I should similarly remove the remaining TIGER tags.

I do not recommend / advocate for removing tiger:reviewed=no mechanically: this would frustrate and actually deeply confuse our ongoing (slow) TIGER Review process. Finally, I only briefly, once again, mourn the loss of the excellent ITO World visualization tool for seeing how much TIGER Review HAS been completed — a similar tool / relatively simple renderer would really help such efforts.

Oh Steve have I got good news for you! It’s still very rudimentary but, through the power of vector tiles, I have a map showing tiger:reviewed=no. TIGERMap. I want to make is more useful to folks interested in working through things etc. I am happy to discuss that more in another thread or in the issues tracker for the site. (ex: if you want a rail layer I can absolutely make that)

I am happy to have a split between the guidance for rail and for roadway.

Speaking for roadway TIGER tagging only (as I don’t yet know much about the rail tagging) is that it would be nice to have a way to say “I have reviewed the geometry, it is fine but I don’t know about the name”. Or similarly “I have reviewed the name but reviewing geometry is too much of a bother for me personally”. Making those necessarily be the same person can only slow us down.

If we can consolidate values then I can make maps showing each of the classes and the invariably Lots More Progress that has happened. Right now it looks like an insurmountable sea of “no” and I really do think that makes people not bother at all.

4 Likes

Awesome, even as a first draft! Rail…can be elsewhere / otherwise discussed, as there are ORM and OHM, ORM with newer and newer (some beta) layers, OHM with that handy-dandy time slider…OT queries.

The magic thing about ITO World’s was the colors and their tags (it was four classes / colors, I believe, some with timings about “since reviewed”, those filters can be re-discovered). The colors and heuristics of filtering were a real sweet spot for “focusing in” on things, in my opinion. I can do (do do, actually) TIGER Review with OT queries (my county gets closer and closer to “done”), but sweet visualization with colors and time filters, yowza!

Doable vector tiles: yeah, baby! It’s like we just grew new muscles or something.

Did everybody see what we just did there?!

That’s amazing! What does the “redlined” option do?

The “redlined” layer is the neighborhoods that were marked by HUD as “grade D” for purposes of insurance and loans. It was explicitly racist and caused a huge amount of inequity that persists to this day. I’ve wanted a redlined map layer for a while and the TIGER map happens to be a place where I could put it.

Here’s a good resource for anyone wanting to read up: Mapping Inequality

1 Like

Cool TIGER map! It’s been a while since I’ve been able to highlight our projects in the are to review all roads in a region in detail and remove tiger:reviewed afterwards. TIGERMap The only remaining bits are pipelines, tracks, and paths which cannot be determined from the aerial images.

It’s so fun to see the little holes in the map folks have carved. Thanks for sharing!

I have pushed the UI update onto the main branch and started a post about the map Announcing TIGERMap - tiger:reviewed=no

Update on consolidation

There are now ~100 different values and I continue to grind through them. This mainly entails doing the “aerial” review of the roadway and marking it as “aerial”.

One thing I have learned, is to be more dubious about values like yes;no as this seems to have a decent hit rate of “I merged 2 roadways together one with yes one with no”.

@Richard to clarify, I delete these on roads that I’ve edited and are completed. I don’t do any mass deletion. Just when I happen across them. My own personal rule is not to remove tiger tags until the surface type is added.

1 Like