Good question. I'd assumed that connecting to a lake/reservoir would suffice - which would make all those extra lines unnecessary clutter - but now that I read the waterway=stream wiki page more carefully, it does sound like at least the larger streams should continue through the lake. (In the case of small non-navigable streams, a lot of them really do end at lakeshores.) I can put those back.

It probably WAS intentional, because the frequently-incorrect TIGER census boundaries tend to cause more problems than they solve, and back then the boundary=census tag was not in wide use.

I've just re-added it, but as boundary=census: way/38445684

(See: boundary=census)

This page has very general info: osm.wiki/United_States_roads_tagging — it says you should use tertiary, unclassified, service, and track for USFS and BLM roads, but doesn't say how to make the distinction. (However, it does give this useful bit of advice: "A road can be classified highway=track if the road is insignificant in the road network, for example: logging roads, agricultural roads, and roads that don't service anything of significant public importance." The roads on Steens Mountain lead to a lot of stuff that is of "public importance.") You mostly just have to use common sense and see how it's been done elsewhere, like in National Forests. "Track" really shouldn't be used for a road that people would be perfectly happy driving their Toyota Camry on. :)

And winter maintenance should definitely not be used to decide a mountain road's highway category. The majority of USFS/BLM roads above 4000' — including many of the paved ones — are inaccessible in the winter, but that doesn't mean they aren't important roads. If you really want to tag unplowed roads, you could use the "snowplowing" or "winter_service" tags: snowplowing=* winter_service=*

Note that the examples on that page are not really a "decision tree," they're just examples. The Steens Mountain Loop Road doesn't really fit any of the "no" or "yes" cases here (because it's not a "minor" road). It's a well-used, well-maintained road that is advertised to the general public as a scenic drive.

Aside from the bullet points, this is really more of a "you know it when you see it" situation. A rule of thumb that I use is that if you can see two ruts in the ground, it's probably a track. If it's a wide expanse of continuous gravel where two cars could easily pass each other and you can't see the double-track, it's probably "unclassified" (or possibly tertiary, although that's not common for gravel roads except in really remote areas that have no other roads).

An objective way to evaluate the road class out in the hinterlands is to check the raw USFS or BLM data: https://gbp-blm-egis.hub.arcgis.com/datasets/BLM-EGIS::blm-or-ground-transportation-gtrn-roads-line-hub/explore — in this case, the loop road's attributes are: RoadClass="Collector"; CartoRoad="Major road"; MaintInt = "MAINTENANCE INTENSITY 5." Anything with MaintInt of 3 or better is probably not a track.

"Not maintained in winter" is very different from "unmaintained track." The Steens Mountain Loop Road (along with the Kiger, Summit, and East Rim roads) is wide, smooth, and regularly maintained.

Oops! Those were completely unintended changes. I was trying to open a stretch of recently-completed freeway in SE Washington, and apparently I did a search for "construction" and accidentally selected some stuff in Nebraska because it was also part of the "Lewis and Clark Trail" relation. I just checked the Strava heatmap to verify that the new lane on 34/75 is not open yet, and I changed those 9 ways back to construction=trunk. If you see anything else that's off, feel free to fix it!

That's a good solution. If trails are completely deleted, someone might re-add them. The "abandoned" prefix keeps it in the database but lets future mappers know that there's a reason it isn't showing up.

OK, so maybe it isn't LIDAR-based. Definitely better than the boundary that was there before though. :)

I don't know... I was using the shaded relief layer from CalTopo, which is generally excellent and seemed to be a good match for reality. I think they use the latest available data they can find.

I assume you mean Cooper Meadows... at any rate, that was three years ago, so I don't remember for sure, but yeah, it does look like the path I drew follows the USFS data pretty closely. I see the current Strava line you're talking about... maybe that's a new trail alignment, but it could also just be one person taking a shortcut?

(Meanwhile, I'd love to properly map those private property parcels, but when we tried to do that a few years ago, the complexity of the Klamath NF multipolygon relation broke the OSM renderer!)

I just checked the Strava heatmap, and it looks like people are using one of the flood channels as the new "road." I traced the new route from Strava imagery and added it as an unnamed track with motor_vehicle=yes but motorcar=no.

I've edited that page in small ways, but I didn't write it. And if the OpenStreetMap wiki isn't "authoritative," what is? That's the only authority there is!

Because recreation_ground is defined in the Wiki as "An open space for general recreation, which often includes formal or informal pitches, nets and other recreation equipment or infrastructure" — in other words, a very man-made space, similar to leisure=park. But Swasey Recreation Area is 99% in a natural state. "Nature reserve" may not be the perfect tag for the situation, but it's the closest we've got. (See: osm.wiki/United_States/Public_lands#Bureau_of_Land_Management)

Oops! I started from raw NHD data for that one, and I forgot to remove all the extra data fields. I just cleaned it up.

Oops. I see what happened: I imported a bunch of tributaries of the Methow River from NHD data into my editor, but then forgot to remove the river itself before uploading. (Fortunately, it looks like I properly connected the new streams to the old river.)

I admit I may have been too hasty in thinking "heath" was not an appropriate tag in the shrub-steppe environment. I see now that they've included a picture of sagebrush on the OSM wiki page. (I think the text there needs be updated to de-emphasize English-style ericaceous heathlands, and emphasize "low shrubs" instead.) So, that was my bad.

However, in this area, there were also a lot of very large areas that I didn't think warranted sweeping natural=* tags. When they're applied to huge areas, then complicated multipolygons need to be created to add smaller features like ponds or scrub. Best to keep them smaller if possible.

I stand corrected. However, regardless of what it calls itself, it is not a "city" by the OSM definition; it's a hamlet. I've re-added the city limits (which are weirdly offset from the tax parcels) from Idaho GIS data.

I've added the Historic District boundaries as defined by the NRHP, with appropriate tags (osm.wiki/National_Register_of_Historic_Places).

Placerville isn't an incorporated city. It might make sense to put its Historic District boundaries on the map, but not as a "city," which is what was there before.

That road is snow-covered in the winter and spring, so it is only navigable for part of the year: usually June through October or November, but there are no specific dates.