On Mon, May 19, 2014 at 4:22 PM, David Bannon <[email protected]>wrote:
> No, no Steve, I worded my last letter really badly and totally apologise > if I unintentionally offended anyone. My comment related specifically to > your line - > Ok, no worries :) > > > Yeah. I'm still deciding what to do about places where Vicmap shows a > > track in the bush that can't be seen on any imagery - probably because > > the vegetation is too dense. > > I meant leave the 'grey' areas to the survey people. There are many > roads (and particularly tracks) that cannot been seen clearly on the > imagery, and many more where some parts cannot be seen. I'd rather the > people working with imagery or other non (recent) survey data such as > Vic Maps did not make "educated guesses" but go and have a look, or ask > some else to go and have a look. > Yeah, it's a real issue. Like I said, I'm still trying to work out how best to proceed. My general approach is to be a bit more liberal with roads that don't go through, and more conservative with ones that do. I'm also trying to use fixme=* to express doubt: fixme=unverified from vicmap (ie, I can't see through the vegetation, so I'm taking vicmap's word for it) fixme=verify access (I can see a track, vicmap has a track, but I'm still a little skeptical that it's public access) Sometimes I also use highway=path rather than highway=track if I'm dubious that the public can drive a vehicle down it. My intention in all this is to minimise the chance that someone gets routed down a road that is not publicly accessible, or otherwise impassable. Personally, I think it's ok to show dead-end 4WD tracks that happen to not be driveable, because I think the people that use those kinds of maps expect that. But definitely willing to discuss this point, and open to all opinions... > > I have had a road (into a new estate) removed, apparently because it did > not show up on Bing. Very annoying to a new owner there who was > directing tradies via OSM ! But that in no way means I don't value the > armchair mappers contribution. I'd suggest adding notes on to the road in question, like "note=This road was built in early 2014 and is surveyed." > I'd just like them to double check their > data, one way or another before committing. > To be honest, I go so much faster if I'm not doing any checking - you might be surprised how fast I map :) It doesn't really make sense to spend 10 minutes verifying a road that I created in 10 seconds - I just take the chance that I'm introducing a couple of errors. But mostly I'm doing stuff out in the bush. I'm usually pretty cautious about deleting anything like you describe. I have come across a couple where I just couldn't fathom why someone had drawn a road somewhere, but I'll usually cross-check against a couple of other sources. Example: http://www.openstreetmap.org/way/172612474 > Maybe what we need is some sort of register ? The people studying > imagery are good at picking up anomalies, differences between image and > map. They could log it and have some local go and check ? Better than > just jumping in. > Are there enough of us to make it worthwhile? Anyway, a better mechanism would probably be through fixme=*, so you can go and look for fixme's in your area at your leisure. > > You may be amused to know that some years ago, I was shocked to discover > I had apparently built my house in the middle of the Bendigo Region > National Park. I was waiting to get a letter telling me to move it when > I realised someone had just followed the tree line, assuming all was > national park. They had swept up the Park it self, the Welsford State > Forest, Sugarloaf Conservation Park and a large number of private > properties. A very quick check would have prevented that error. > Yeah, that seems pretty silly. Although IMHO we need a better approach to maintaining administrative boundaries - it doesn't really make sense for anyone to be able to move them at will, since there is a genuine authority for each. > > I am pretty sure all we want is for the database to have accurate, > relevant data. > > You left out "comprehensive" and "useful". I think I have a higher tolerance for error because I want OSM to be useful and complete-ish *now*. I use it on a regular basis for planning trips, and I can't wait a few years for all the checking. I'd rather a pretty complete map with a few errors which will be corrected over time. But that's all it is - different priorities. Maybe I think 98% accuracy is enough, whereas you want 99.5% - and someone else might want 99.95%... Steve
_______________________________________________ Talk-au mailing list [email protected] https://lists.openstreetmap.org/listinfo/talk-au
