> > On 25.10.2017 08:22, Tomas Straupis wrote: > > Yuri later tried to change the whole theme from "osm-wikidata-sql > > tool" to "new general qa tool" in the same thread. > > And now into "is local knowledge really always necessary". I'm sure > before too long Yuri will be starting to discuss whether what we see is > really there, or perhaps just an imagination, and explain to us that we > can't prove either way. >
Frederik, I think you once again try to do a personal stab instead of discussing the issue. Fighting imaginary targets is always better than address real concerns, especially considering that Tomas statement above is incorrect. > > The answer to the "local knowledge" issue is this: Local knowledge comes > in shades. Someone who hasn't been to a particular street but lives in > the same city can certainly make a better guess about things than > someone who lives in the same country but in another city; and that > person's assessment about something will still be better than that of a > third person who doesn't speak the language and lives a continent away. > No argument here - I have been saying exactly the same thing - the further it is from you, the less you may know about it. > > Local knowledge nearly always beats "common sense" applied by someone > from far away. The person from far away *might* change "Thomas Eddison > Street" to "Thomas Edison Street" based on general knowledge, but then > if the local person says "no, it really says Eddison on the street sign" > then that's what it is. > > Bad analogy. Fixing street names is clearly the most well known to the local person, and noone would argue with that. It would be a bad idea to fix street names across the globe, because yes, you are likely to make a mistake. On the other hand, seeing that a Wikipedia link in Czech Republic starts with cz: is an error. It can be easily fixed by someone who is an expert in Wikipedia tags, but might be overlooked by someone with the local knowledge. But even in the absence of a particular objection from someone local, > many in OSM take a dim view of "gardening" in areas where you don't even > know the language or culture. It is much too easy to make mistakes. > I think this is a very broad statement. Knowing language and culture obviously matters with tags like names, but may have very little to do with others tags like the above cz: instead of cs: wikipedia tags. Moreover, I would argue that for *some* tags it is better to let global tagging expert guide the process, instead of a novice with the local knowledge. In short, do not pile together all types of tags - some are highly local, some benefit from the global expertise. All for a simple reason - every data consumer needs to make sense of our data, so if we don't do grooming, every single data consumer has to support every kind of obsolete tagging scheme that was ever used. There are over 100 such autofixes in JOSM alone - do you want every data consumer to be an expert in each one of them? I think that goes against the main idea of OSM - free, easy to use data for map projects. Imagine someone who wants to do disaster aid app all of a sudden has to dig through hundreds of not-yet-cleaned-up obsolete tagging rules, simply because noone in that area happened to work on it yet.
_______________________________________________ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
