Poll: Unpacking tracktype=* - how should it be used?

General talk Tagging general discussion
, , , , , ,
1

Let’s start with an essential question:

What should the destiny of the tracktype tag be?
  • It should be deprecated, existing tags are sufficient
  • It should be deprecated in favour of new tags, each representing its different criteria independently
  • It can continue to be used, but it needs a major redefinition (e.g. dropping some criteria and selecting dominant ones)
  • It can continue to be used with minor changes (e.g. aligning criteria that mostly give similar results)
  • It can continue as is, no changes are needed
0 voters

If you chose “no changes are needed”, please ignore the rest of this poll.

The recent discussions allow us to better describe how tracktype has been used and what the current data might be capturing. The rest of this poll, however, is an attempt at improving how the wiki prescribes its use for future mapping and review of existing data. For the next 24 questions, I tried to avoid terms that caused confusion in discussions, a sign that these terms may not be understood equally by all. Instead, I tried to decompose them into less abstract descriptions, aiming at eventually better defining the abstract terms. Though long, despite my best efforts, this poll is by no means necessarily complete, and some options might still need clarification. Given its apparently irreducible complexity, a follow-up will likely be needed to address any issues. Here we go:

When assigning a tracktype value, how relative should the mappers’ assessment be?
  • They should apply a fixed standard regardless of what tracks in the area look like
  • They should apply a relative standard based on what tracks in the area typically look like
  • It depends; it may be something like fixed standard for material composition and engineering, and relative standard for overall condition
0 voters
How regionalised should tracktype values become?
  • Ideally it should be uniform globally: the grades in one country should be comparable to the grades in another country; the resulting distribution will naturally vary by region, reflecting their real differences
  • The grades should be adapted to local variation; grade5 in one region may be similar to grade3 in another; the resulting distribution will tend to be more uniform within each country
  • It should accommodate some local variation relative to a desirable global standard, within some limits
  • It should drift freely, no need for inter-regional compatibility
0 voters
How should existing data be reviewed?
  • Slowly, as possible and as needed
  • Quickly, count me in!
0 voters
Should tracktype be based on characteristics easily verifiable by other mappers?
  • Solely
  • Mostly
  • Partially
  • Not necessarily
0 voters
How variable can the characteristics determining tracktype be over time?
  • They should typically vary slowly
  • They can vary at any rate
  • It depends on the characteristic, more variable characteristics may be interpreted less strictly by the mapper
0 voters
How should the main criteria for tracktype be weighted?
  • Approximately averaging all criteria; for example, a compacted road (grade2) covered by grass (grade4 or worse) would be mapped as grade3
  • Choosing the criterion producing the best grade (grade2 in the example)
  • Choosing the criterion producing the worst grade (grade4 or worse in the example)
  • Defining an order for the criteria, continue the next criterion only if the previous criteria don’t narrow down to a single grade value
  • Setting specific rules for deciding which criterion dominates in each situation
0 voters
How should variable conditions be represented?
  • tracktype should reflect the worst conditions
  • tracktype should reflect the typical conditions, tracktype:seasonal and tracktype:conditional should reflect worst/variable conditions
0 voters
If tracktype should reflect the typical conditions, how should the typical conditions be determined?
  • Recording the result of multiple surveys
  • With a single survey under apparently typical conditions, assuming it reflects a mostly stable condition, with no follow-up required
0 voters
How should the worst/variable conditions be determined (whether for the main tracktype value or for tracktype:seasonal and tracktype:conditional)?
  • By surveying during these (worst/variable) conditions
  • By inference while surveying under typical conditions
  • Using alternative sources (local news, word of mouth, community input)
0 voters
Before assigning tracktype, should mappers drive, walk or cycle on the track first?
  • Yes, interacting with it is a must
  • Interaction is preferable but not required
  • Not necessarily, observation is sufficient
0 voters
Which parts of the track’s surface should mappers evaluate along its cross-section? Select all you think apply.
  • Wheel tracks
  • Center strip
  • Edges
0 voters
How should those track parts be evaluated?
  • tracktype should represent the worst part
  • tracktype should represent an average condition over these parts
0 voters
How should variability along the track’s length be reflected in the value of tracktype?
  • It should reflect the average for each section between intersections
  • It should reflect the worst value for each section between intersections
  • The track should be split along major changes affecting tracktype, as long as those changes are fixed or predictable
0 voters
Should grade1 still be associated only with the paved surface types (asphalt, concrete, chipseal)?
  • Yes
  • No
0 voters
Should grade2 still be associated only with imported or engineered material (gravel, compacted, fine_gravel)?
  • Yes
  • No
0 voters

grade3 and grade4 have no strong anchors and can only fully be addressed in a follow-up once the criteria questions have been answered.

Should grade5 still be associated only with completely unimproved natural surfaces (no imported material, no engineering work, e.g. natural bedrock)?
  • Yes
  • No
0 voters
Is track classification hierarchical, as highway classification is?
  • Yes, lower grades branch off from higher grades topologically; grade3 should connect to grade2 or better on at least one end, and grade3 should not be completely surrounded by grade4 or worse
  • No, grades can be spatially arranged freely
0 voters
If mappers should consider vehicle usability (travel comfort/ease and if the track’s length can be traversed) directly when mapping tracktype, which vehicles or modes of transport should be the reference? Select all you think apply. If you think mappers shouldn’t consider this directly, rather treat it a consequence of other characteristics, you can express that in the last question of this poll.
  • Agricultural machinery
  • Trucks
  • Loaded logging trucks
  • Standard passenger cars
  • High-clearance vehicles
  • 4WD vehicles
  • Motorcycles
  • Standard bicycles
  • Mountain bikes
  • Horses
  • Loaded carriages
  • Walking
0 voters

This last part of the poll has many criteria to select from. The first three lists are about acceptance, and the last three lists have the same options but they are about rejection. Abstention can be expressed by simply not selecting a criterion in any list. Selecting the same criterion across two list pairs (acceptance and rejection) will be treated as a mistake and considered a form of abstention.

When assigning a tracktype value, which of the following should mappers observe or consider to determine its value? Select all you think apply.

Group 1: composition and engineering

  • The surface’s proportion of hard and soft particles (hardest: rock; softest: organic matter)
  • The surface’s particle sizes (coarsest: boulders; finest: clay)
  • The resulting firmness of the assembled surface (bulk combined materials)
  • Superficial presence of imported aggregate (gravel, crushed stone)
  • Whether the surface is paved (asphalt, concrete)
  • Track width relative to vehicle width
  • Each wheel’s track width
  • Presence of drainage infrastructure: ditches, culverts, camber (raised surface)
  • Signs of surface compaction by engineering or maintenance
0 voters

Group 2: condition

  • Degree of surface compaction by traffic
  • Presence of tire imprints
  • Presence and depth of ruts
  • Presence and clearance of exposed rocks
  • Presence of potholes
  • Presence of cracks
  • Presence of erosion gullies
  • Presence of roots
  • Signs of surface subsidence
  • Signs of crumbling or collapsing edges
  • Presence of vegetation on the wheel tracks
  • Presence of vegetation on the center strip
  • Overhanging or encroaching vegetation from the sides
  • Signs of vegetation clearing (leaves or branches have been cut)
0 voters

Group 3: interactive, inferred and additional knowledge

  • Deformation under traffic: how much it sinks or ruts when driving, walking or cycling on it
  • How loose or unstable the surface particles are when driving, walking or cycling on it: whether they shift, slide, or do not hold together
  • Inferred surface changes when wet: how much it softens, becomes slippery, or loses bearing capacity after rain
  • Surface changes when wet verified directly
  • Additional knowledge about the surface when built (e.g. if it was compacted by a roll or paved and later degraded)
  • Additional knowledge about the subsurface: levelled, graded, or compacted base layer; or knowledge of load bearing capacity independent of visible top layer
  • Additional knowledge about construction/improvement effort level to build the track
  • Usability (passability/comfort) for one or more types of vehicles
0 voters

Are there any characteristics (same list as above) that you think should NOT influence the value of tracktype, even if mappers notice them? Select all you think apply.

Group 1: composition and engineering

  • The surface’s proportion of hard and soft particles (hardest: rock; softest: organic matter)
  • The surface’s particle sizes (coarsest: boulders; finest: clay)
  • The resulting firmness of the assembled surface (bulk combined materials)
  • Superficial presence of imported aggregate (gravel, crushed stone)
  • Whether the surface is paved (asphalt, concrete)
  • Track width relative to vehicle width
  • Each wheel’s track width
  • Presence of drainage infrastructure: ditches, culverts, camber (raised surface)
  • Signs of surface compaction by engineering or maintenance
0 voters

Group 2: condition

  • Degree of surface compaction by traffic
  • Presence of tire imprints
  • Presence and depth of ruts
  • Presence and clearance of exposed rocks
  • Presence of potholes
  • Presence of cracks
  • Presence of erosion gullies
  • Presence of roots
  • Signs of surface subsidence
  • Signs of crumbling or collapsing edges
  • Presence of vegetation on the wheel tracks
  • Presence of vegetation on the center strip
  • Overhanging or encroaching vegetation from the sides
  • Signs of vegetation clearing (leaves or branches have been cut)
0 voters

Group 3: interactive, inferred and additional knowledge

  • Deformation under traffic: how much it sinks or ruts when driving, walking or cycling on it
  • How loose or unstable the surface particles are when driving, walking or cycling on it: whether they shift, slide, or do not hold together
  • Inferred surface changes when wet: how much it softens, becomes slippery, or loses bearing capacity after rain
  • Surface changes when wet verified directly
  • Additional knowledge about the surface when built (e.g. if it was compacted by a roll or paved and later degraded)
  • Additional knowledge about the subsurface: levelled, graded, or compacted base layer; or knowledge of load bearing capacity independent of visible top layer
  • Additional knowledge about construction/improvement effort level to build the track
  • Usability (passability/comfort) for one or more types of vehicles
0 voters

Thank you for your interest and patience!

Poll: is tracktype about improvement?
Tracktype challenge 1
2

Poe’s law wins. I genuinely can’t tell whether this is supposed to be satire or not.

12 Likes
3

I gave it a try but gave un in one third or so. I would actually drop firmness as an elevated criteria and have tracktype just be a general quality tag like sac_scale and mtb_scale. I think that it already mostly works like that despite what the wiki says. The polls by the way assume a way too detailed and overcomplicated way of assesing it. There can be a simple table along several axis, as I think @Hungerburg suggested somewhere.

2 Likes
4

At least it was quite a slog.

5

treating as viable option or proposing to redefine tag used over ten million times is deeply unserious and waste of time

maybe tag is terrible, maybe it should be ignored, maybe it should be replaced, maybe new one should be invented

but redefining is nonstarter, impossible without time travel or other magic, would only make data worse and make mockery of any process used to do it

redefining tag would mean that values are mix of old and new and just make any existing problems worse

if you would have resources necessary to redefine tracktype= and succeed then using them on improving tagging schema in OSM would be outrageous waste of them

6 Likes
6

There was a poll that would have received 75% majority if it wasnt for somebody from Southern America that did not participate in the discussions here at all. I guess I throw in the towel (is this English?). Obviously the users of a tag have no say in how it is documented. My favourite phrasing from a post in the forum that matches what I perceive of being the most approved use of tracktype by people actually using it got reverted quickly when put on the wiki. Obviously the custodians of the wiki develop the documentation along different trajectories. More scientific, I suppose? I cannot even read all that what the OP here posted, plain too much.

Unfortunately, surface=fine_gravel is another such effort that invalidated lots of mappings, when it got changed from an alias of compacted used a lot on track roads into something never found there but on playgrounds or outdoor seatings.

2 Likes
7

I had to enlist AI assistance just to confirm there are indeed 25 polls in this post, which I believe is a legitimate Guinness World Record candidate. :star_struck:

Fully agreed. The most sensible path is to acknowledge that tracktype is, and has always been, per the iD preset, about firmness, document it as such, and leave it alone.

If we want a proper general-purpose track quality classification, we invent a new tag for it.

And in true OSM naming tradition, since tracktype got hijacked to mean firmness, the logical name for a new general track classification tag would obviously be… firmness. :upside_down_face:

1 Like
8

is what taginfo calls it, a quality:

in German “Güteklasse eines Wirtschaftsweges”, in English “Provides a classification of tracks.”

Funnily in Portugues it is calld “Esquema de classificação da firmeza/solidez de trilhas/trilhos e caminhos, maioritariamente rurais.”

Should I make a poll out that to gather if this became the global definition? So mappers outside of Brazil get to think over what they mapped?

Regarding the presets: The key is still called tracktype, not firmness. It is all in how somebody parses the values. Mind you, in the beginning it was about “usability”. and infrastructure,

2 Likes
9

TagInfo takes this from the wiki. The description was expanded from the original by the two translators, but, like much of the Portuguese wiki, the article is mostly an out-of-sync translation of the English wiki, last synchronized with it in 2020, and 2016 for the values table.

Why not? It’s up to you. Perhaps highway=track is more relevant in Portugal, in Brazil it is quite insignificant, its last mention in our Telegram channels was in 2023.

10

Noting this here as the relevant question (number 18) currently has 4 votes, all against a hierarchical interpretation. I couldn’t find any other forum post, mailing list message or wiki discussion claiming the same, and since this isn’t part of the original proposal either, I highly doubt it’s a widely adopted mapping practice.

11

Let me try to be charitable here. Your description of a “hierarchical interpretation” (in your question 18 that now only 3 people have voted in) is “Yes, lower grades branch off from higher grades topologically; grade3 should connect to grade2 or better on at least one end, and grade3 should not be completely surrounded by grade4 or worse”. The original proposal (in 2006, talking about rendering) says “but the outline should be a fainter grey, the further down the list it is.”.

That is what I (and the original proposal) meant by “a simple hierarchy” - a hierarchy of quality implies nothing about what is connected to what.

2 Likes
12

Given that tracks are roads, a road hierarchy is typically a topological hierarchy as I described, where each level branches off from higher levels. I assumed this was the common understanding in this context. A “hierarchy” where topology is absent is more accurately described as a scale or a rating. The original proposal does not mention “classification” nor “hierarchy” but it refers to “grade” which is a level along a “scale”.

13

I remember when I read the proposal some time in the past, I got reminded of the primary, secondary, tertiary tags that openstreetmap uses for public roads. Perhaps some kind of envy that mappers of tracks felt, having only a single tag for what is quite richly layered in nature.

Although grade1 was said to be “mostly” (IIRC) the ways from the public network to the dwellings of the farmers, it never occurred to me to describe such highbrow philosophic notions as you bring to debate. I understand that you want a rigid ontology. But then, please choose another key to force that on the community, you have been late back then and are even more late today.

Did I say that already? The pictures from the proposal tell a different story, quite convincing, at least for those that know areas as the ones depicted. Tracktype is all about the physical side. That the concept not developed with scientific rigour, too late to change that.

Mark the key legacy? OSM-Carto certainly would love to get rid of it. Smoothness is going strong. but I do not think it is up to what Carto maintainers expect, in the sense of verifiability certainly not. Only recently I learned, even surface not so easy, unless paved/unpaved/unknown is all that is required.

2 Likes