Uncorrectable wrong assumptions?

Long in the past, some farangs started mapping in Thailand. Without a local community and without proper knowledge of Thai language, they set up some rules. Of course, they got some things wrong.

So what? It is possible to learn from experience, and to adjust rules to fresh knowledge. At least theoretically possible.

The original focus of OpenStreetMap is roads. Hence, the road network needs to mapped and structured, and some classification schema is required. Mappers decided to count the digits of the reference number, as this seemed to be a good fit for the roads they knew back then.

Relatively early, some mappers found out that some road reference numbers contain a prefix indicating the province (changwat) that was previously ignored (and is still ignored very often). They decided to categorize them at the same level as 4-digit-roads without prefix, so they could continue with the number counting classification.
And still nowadays, roads which do not have a reference number at all are not differentiated from roads which do have a number which just has not yet been mapped (and those are many).

Originally, there was a strict assignment from number of digits to the highway tag. In the current version of the wiki, one might assume that there is now some flexibility to one level upward or downward. But it is strictly not more than that, and even that one level is an extreme exception, as we see from several discussions.

But in fact, that system is fundamentally wrong.
Other countries managed to do things better.

Examples from France
Near Haguenau, road D 263 changes from secondary (south of town) to primary (in town, and north of town): https://www.openstreetmap.org/#map=15/48.7963/7.7687

Nearby, you can find tertiary roads with „D“ numbers (as for “Departement”, meaning “province”), like D 139. Near Strassbourg, there are roads with “M” numbers, also at secondary or primary levels.

The French wiki page https://wiki.openstreetmap.org/wiki/FR:Voirie just describes the meaning of the different highway levels, without using the administrative numbering.

Examples from Germany
While the administrative numbering schema generally fits well to the OSM highway levels, there are exceptions.
Road B19 is a primary north of the junction with motorway A7, and trunk south of it: https://www.openstreetmap.org/#map=14/49.8443/10.0097

But my favorite example is the South of Karlsruhe. Road K9657 is a “Kreisstraße” (county road, and county is one administrative level below “changwat”), so that would typically be a tertiary. But it is a trunk: https://www.openstreetmap.org/#map=16/48.9927/8.4237
South east of it, there is K9652 also at trunk level, and north east of it, K9652 at primary level, and a little farther to south east, K9652 at tertiary level.

Consequently, the German wiki page https://wiki.openstreetmap.org/wiki/Attributierung_von_Stra%C3%9Fen_in_Deutschland always admonishes that the administrative number is merely a „Richtschnur “ (“guideline”) and the actual “Verkehrsbedeutung” (“traffic importance”) is the actual criterion.

Oddly, while Germany has several “secondary” roads which do even lack a central line (not wide enough), Thailand has dual-carriage tertiary roads…

“Objective” criteria?
At OpenStreetMap, we prefer measurable criteria, which can be reproduced by other mappers, over “subjective” criteria where different people may reach different conclusions. A reference number is hence “objective”: you can go the signboard, and read it, and others will be able to read it in the same way (if they know the Thai alphabet, in some cases). But “traffic importance”? How can we measure that?

While one might argue that it works in Germany because of the generally high coincidence with the administrative numbering, the example of France proves that it does also work when the administrative numbers do not reflect traffic importance.

Edit wars do not seem to be common there.

In contrast, a small group of loud mappers in Thailand, aggressively discourages other mappers who follow international rules. They get accused of “cavalier mapping” or “vandalism”, and will see their contributions reverted. E.g.

Several of them did give up.
RocketMan (https://www.openstreetmap.org/user/RocketMan ) was last active some 7 years ago; he even removed his forum posts when he left.
Autoisme (https://www.openstreetmap.org/user/Autoisme ) was just recently discouraged.
Natthawee Chutianusornchai (https://www.openstreetmap.org/user/Natthawee%20Chutianusornchai/ ) also.
Yot Chaiyana (https://www.openstreetmap.org/user/Yot%20Chaiyana ) also.

The atmosphere here has long been toxic for new users, and also for some long-term members. A small group of people manages to enforce their style which is different from the rest of the world.
And because others refrain from using this forum (which uses English language, which many Thai people do not know; and which uses a farang communication style which is not at all compatible with typical Thai conversation styles), this small group thinks they are the majority.

No. The majority is not here on this forum.

I’ve long admired your knowlege of and contributions to OSM. But to expect the same quailty of road classification in Germany to apply to Thailand makes no more sense than expecting the quality of concrete here to be as good as in Europe (It’s terrible). I’ve often changed classifications based on signs that did not agree with the map. I just assumed the sign was right, but who knows? I use OSM on every bike/car trip and the only thing that matters to me is if the road is right.

If the atmosphere here is toxic to new users, it’s because there is too much concertration on minutae. I say get the basics right and the future will take care of itself.

is not what I expect.
I just want to get rid of wrong assumptions which are now virtually exclusively used for highway classifications. To get a little closer to international standards of classification.
Without edit wars, as is exemplified with France.
The way is long.

(On a side note: take care of the quality of roads in German villages… - they are often neglected by the authorities)

In Malaysia, that’s how you get primary roads everywhere there is heavy traffic. Secondary would’ve been adequate enough but there’s no going back now. At least the routing works adequately most of the time.

If traffic importance is defined as the volume of traffic, then this would solve the problem.

By the way, the wiki page on highway used the term “importance of the highway within the road network”, “highway=* distinguishes roads by function and importance rather by their physical characteristic and legal classification.”, and also “Only highway=motorway and highway=motorway_link indicate quality. Other road types, …, do not imply anything about road quality, only importance and intended main use.”. So, I believe the traffic volume is not the case.

If we only focus at National Highway (maintained by the DOH), the official classification is meant to separate by importance. The 1-digit is the main road from Bangkok to each region, the 2-digit is the main road inside or between each region, the 3-digit is the main road connecting each province, and the 4-digit is the main road linking each district in the province. Because the DOH’s classification is similar to the OSM aim, tagging 1-2-digit as trunk, 3-digit as primary, and 4-digit as secondary should be the best way for general rules. There may be some few exceptions, such as when a 3-digit national highway is exceptionally long and should be considered the region’s main road, or when a bypass road outperforms the old one and the classification should be switched between them.

By the way, the Rural Road and Local Road is NOT separated by their importance. It is only about authority, and we should not rely solely on it for OSM classification. If we’re going to make a decision based on “importance,” the only way we can do is to compare between nearby highways.

For my suggestion, what is practical is to have clear general rules for national highways, and if there are any roads with exceptions, we list all of them in the wiki. But for other types of roads, I have no idea but let it remain relative.

Though a high importance often comes with high traffic volume, these are different concepts. Also dense local traffic is different from an important role in the country-wide road network.

Unfortunately, that was not achieved. When a new bypass road was built, the situation must be re-visited, and tagging adjusted: the new road may become more important than the old one. Even when the old road had 1 digit only, and the new road 4 digits (e.g. Phangnga town to Thap Phut village: 4311 is more important than 4), and in other places the bypass may even have a DRR number and become more important than the DOH road.

Far too many exceptions. Look at the example from France I mentioned above. You could try to establish such a system there too, but the French did not. In South Thailand, e.g. some sections of 4, most of 42 is not trunk, while e.g. 401, 403, most of 406, 419, some sections of 408 are.

and beyond that, there may be DRR roads which are quite important. In the South, secondary is quite common, primary rather rare; in the North we saw some examples of even trunk which were denied that trunk classification only by virtue of their DRR number - not at all by their importance in the road network.

We had many discussions in the past, and in 2013 (?), after a discussion here on the forum, I had the impression that we really came to that decision and then adjusted the wiki accordingly.
But later on, I had to learn, that other people still only looked at the ref number thingy which was still mentioned in the wiki (as typical example, not as absolute mapping law) and insisted that the decision must be based on the number…

That’s why I would follow the French example, and get completely rid of the road numbers in the wiki section on the highway tag (a section on the ref tag, explicitly mentioning that it is for the ref tag only, should describe the ideas behind the Thai numbering system: that helps to detect wrong ref numbers or to find connections to other roads; e.g. road 1023 is wrong in the South, all DOH numbers there start with 4; might be a DRR road where the changwat code was forgotten, but then it must connect to a 1 digit DOH road etc.).

According to France’s wiki, https://wiki.openstreetmap.org/wiki/FR:Cartographie_des_routes_en_France , the official classification is still used for OSM classification, with at least a N (National), D (Department), M (Municipal), and Fast Lane (with specific sign). They are not avoiding an official classification, but exceptions may occur, with clear guidelines.

Of course, things aren’t perfect, but it’s undeniable that the DOH road classification criteria is the importance. We can see some attempts to improve it, such as the National Highway No. 12 is used to be shorter than it is now. What they did was combine 209, 213, and 2042 to make 12 longer and perceived as one of Isan’s main road networks.

To follow French way, we can create clear guidelines for exceptions. But, at the very least, keeping the official classification as the baseline is clearly an advantage rather than a loss, especially since we’re aiming to make the guidelines more “objective.”

I believe it is possible. Anyway, there is another option. I understand that listing them all in the wiki might not assist because the mapper who does the “classification changing” will not carefully read it, as this happens frequently.

See https://forum.openstreetmap.org/viewtopic.php?pid=862397#p862397