On May 12, 2020, at 8:43 PM, <eze...@gmail.com> <eze...@gmail.com> wrote: > Butting in for my *long* two cents:
I appreciate you doing so. > Please correct me if I'm wrong in characterizing boundary=administrative like > this; I'm still not sure that I've nailed it down. I'd say (and for the many-th time, I'm not a political scientist) that when you cross a boundary=administrative, you are in "another" legal jurisdiction from the one you were just in, where rules are different (however slightly) and "it isn't THESE statutes, police, laws, courts, etc. which ascribe, rather THOSE do." I don't think we want to measure where a different bus district network, water service agency or dog catcher provides service, but maybe I'm mistaken. > The question is, then, is this a useful delineation for most use cases of > OSM? That's a question that only the end users of OSM can really answer. I > don't consider myself one, but, like Martin, I would say that if I were to > use counties in OSM, I would want to use it to visualize more granular data > across the US. Except, and I don't mean to split hairs needlessly here, a "county" in 46 states (or 48 if we count county-equivalents in Alaska and Louisiana) isn't the same thing as a county in two (Rhode Island and Connecticut). So, in the above scenario when you describe "using them to visualize..." they are not equivalents. IMO, OSM should delineate this, and years ago we evolved a tagging scheme that does so, in the USA, and we were careful to get the final finishing nails on New England in summer of 2017. I think we left some rough edges on Connecticut, perhaps because it had recently (in 2013-5) undergone its transformation from 1960, dissolving counties, 1980, reformulate a much weaker version of something like counties based on the COG structure of towns (and there were 15 of them), to the "these are all RCOGs now and there are nine." Today, we attempt to both identify that there is a distinct thing in Connecticut which is called "county" (and which seems to have certain "weakly bound to the state" governmental aspects like defining judiciary boundaries, though they are largely geographical besides that) and that there ALSO are RCOGs which are similarly-sized as counties (shouldn't matter). While governmental, again, in a "weak" manner (having jurisdiction over only landuse, allocating federal transportation planning dollars and similar), COGs are quite limited and overlap in "crazy quilt" kinds of ways with other jurisdictions. For example, some COGs also serve as either federal metropolitan planning organizations (MPO), rural planning organizations (RPO), or share staff with one or more MPOs/RPOs within their borders; the Western Connecticut COG supports both the Housatonic Valley MPO and the South Western CT MPO. In short (sorry, a summary is needed!) "using counties in OSM" means "using some THING that is like a county in other states, but isn't one in Connecticut." AND it must contend with the fact that COGs exist there, too. > Counties and county-equivalents are the only sub-state division that cover > the whole of the country and provide much more reasonable localization than > states do. If I were to do something GIS-like with counties, I would see > "admin_level=6 is counties. Let me extract all of those". This causes a > problem, which I could see going down in one of three ways: Not strictly true when you include aboriginal_lands. > 1. In a good case, would notice that certain parts of CT, RI, and MA are > missing completely, realize after going through the arcana table that those > are tagged "boundary=region", and download those counties. Now I'm mildly > annoyed that I lost my flow to consult the documentation and my code will be > harder to maintain because I did essentially the same thing twice. It is not OSM's job to avoid harshing your mellow or interrupting your flow. If a county is a county in 46/48 states and different in one, two, three or four, OSM not only has the capacity to properly tag this (with plastic yet fits-into-an-understandable-scheme worthiness) it SHOULD tag these differences. Even when subtle. Otherwise, we shoehorn something which isn't something into a box that say it IS that something. We shouldn't do that. > 2. In a worse case, I might say "whelp, time to hard-code those counties in" > -- which is an indication that the database has failed. No, your code would accommodate that there are different kinds of boundaries, some administrative, some not. What we are trying to determine is the right answer to this question: if you say "give me all administrative boundaries, including those in New England..." should you see "holes" in parts of New England? (At the county level?) I'd say yes. We might argue over where and what tags, but these show up as holes because of both how you asked the question and because the answer is "they are slightly different beasts." Or, maybe not, and "judiciary only is enough" tips the scales to call them 6 in Connecticut, I don't know. So, let's discuss it. We are. it is exhausting and I keep an open mind but we are discussing it. > 3. In the worst case, I might not notice that CT and RI have failed to appear > at all, leaving me with a broken application. It wouldn't be broken, if the underlying assumptions are correct and the code / query which captures "administrative only" is correct. If your use case includes "wider than administrative, i.e. includes regional counties in New England" well, then do that (in your code or query). The data must be factually "reality" or as best a facsimile as OSM is able to achieve. That's all I'm doing here. > Or, more likely, some other fifteen-year-old might come along, tag it as > admin_level=6, I blindly use admin_level=6, not realizing this is the wrong > tagging for Southern New England, someone reverts it back, and now my > application is broken. We can't 15-year-old proof OSM. Well, not perfectly, anyway. People coming along and messing things up because they don't understand a concept or didn't read documentation isn't the fault of people who hash out a consensus, document it, polish up the map to have reality match what we agreed and documented. It is the fault of the messer-up-er. That can be corrected if and when it happens. > Then the opposite question. What is so bad about tagging counties in Rhode > Island, Connecticut, and all of Massachusetts as admin_level=6? It may be that in the case of Connecticut, this is the way to go (there ARE judiciary boundaries, and that is a form of government boundary). In Rhode Island, I don't think there are admin_level=6 boundaries because there is no boundary=administrative going on: there is no government happening at that level. Counties there are vestigial and essentially geographical, not governmental. And yes, me changing RI from five to three was a simple mistake on my part, I have changed it back to the correct number, five. My apologies. > But there also seems to be an understanding among most people from the area > who have cared to comment on either this mailing list or on the wiki > discussion that counties in southern New England should be admin_level=6. > This is probably a situation similar to what Steve mentioned at the start of > this thread (COGs in Oregon, didn't go over well with locals, &c), and I'm > sorry to have extended this Michael-Bay-explosion-of-a-discussion further. I read this loudly and clearly, and NOT with a grain of salt. Local perspective is vital to OSM. In this case, I honestly think a university professor or legal scholar familiar with the issues and history could benefit us as well. Especially if they "get" OSM's admin_level, which IS a rough tool. > Overall, I think the main issue with this is what Steve mentioned very early > on: > >> If you fancy yourself (or know one!) a political scientist steeped enough in >> US law, history and politics sufficient to discuss subtle, nuanced topics >> like Home Rule and Dillon's law, a Discussion in our wiki could use your >> wisdom and guidance. > > None of us are political scientists, and, despite all of our energetic > keyboard mashing, we have yet to consult a political scientist to resolve the > core issue. In my mind, the fact that we would even need to get a political > scientist to suss this out is a problem. I disagree. I have often benefit from seeking out learned, scholarly people on topics of their expertise. So has society, the news media do this often. It is true that sometimes people heed and sometimes people don't, doing whatever they want and ignoring the sage wisdom and scholarly perspective. When/as they do, "caveat usor." > When an end user is using the data, are they going to have a political > scientist on speed dial to explain why Connecticut and Rhode Island don't > have counties? They DO have counties, they are different than what we mean by "county" in other states. Let OSM capture those semantics as accurately as it might. For both experts and lay persons alike. > If we can't (and we're supposed to collectively understand what the heck > we're doing with the database), how can we reasonably expect a user to know? > And how would you explain to fifteen-year-old me the deep political history > entwining US government and how that leads to me not being able to put a line > over the Hunt River saying that "Kent County ends here! Do not go past this > line unless you want to enter the haunted South County!" on my cool map? Tag as we say we should (and more-or-less do). You might be confusing what you see in a rendering (Carto, your own renderer, perhaps)...) with the data in the map. Simply because a renderer doesn't map boundary=region doesn't mean it isn't there (in the database). It is, it renders differently (or even not at all). These are often hashed out topics (for 15 year olds, for adults) about digital map production. I don't mind talking about them again, though because renderers change, laws change, better/newer tagging schemes sometimes emerge, they almost MUST be talked about every so often. We're simply discussing. I strive mightily to keep my mind open and not seem autocratic and having an attitude of "I'm certain I am right." We all should. But let's also acknowledge that things change and emerge in OSM and we still seem to need to fine-tune middle-level admin_levels in parts of New England. I'm OK with that, but my patience wears thin. It doesn't make me mad at anyone in particular for that, just, um, tired of typing so much. SteveA _______________________________________________ Talk-us mailing list Talk-us@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-us
