[OSM-talk] Roadmap for deprecation of name tags in OSM

stevea steveaOSM at softworkers.com
Thu Aug 20 19:45:11 UTC 2020

On Aug 20, 2020, at 11:58 AM, pangoSE <pangose at riseup.net> wrote:
> ...I would go so far as to say that
>> ignoring this problem with missing references makes the names in the
>> whole database worthless to use and contribute to because it could just
>> be random joe next door sitting on a late night and adding/changing a
>> lot of crap names wi h a handful of new accounts to the database
>> objects and no-one regularly does QA-checks to see if it has any link
>> to reality.

I don't like it when people say to me about (my country, for example) "love it or leave it" because that precludes being a good citizen and suggesting or making IMPROVEMENTS to whatever it is that offends.  However, because of the pure falsity of the first part of the above statement ("names in the whole database (are) worthless"), I am tempted to say exactly this to pangoSE:  simply don't use OSM if you find our data so worthless.  Your problem is thusly solved.

Or, (much better) if you are going to use them and complain (as you are and do), propose a way to fix them.  You have done so, but your proposal has elements that are so very over-engineered and reliant upon principles of semantic binding and tuple-/triple-stores (as I mentioned before) that are new, untried/untested, are in the earlier stages of their development and are largely unfamiliar to most (besides academics and first-/early-adopters) that they need several additional "heavy lifts" associated with them to even be considered at a beginning phase by this community.  I listed some of the more important ones in my earlier post and I could list more, but I'll refrain for now.

What pangoSE did was not address my (I believe thoughtful and considered) post, which looks to accommodate pangoSE's proposal over a much-more reasonable longer-term, absolutely required given the complexity and massive kind of changes such a proposal would entail.  Rather, pangoSE "piled on" and "dug in," now hurling insults at the data in OSM which already exist.  This behavior (both kinds I just outlined) does not endear such proposals to the OSM community and dulls pangoSE's effectiveness in presenting them (again, while ignoring constructive reply).

As such, I find his proposal (and concomitant seeming lack of willingness to listen to the feedback he solicited) to be disingenuous, especially as a result of his petulant "sour grapes" attitude towards our data (and project, really).

>> Additionally the tags and their changes over time is really hard to
>> follow on openstreetmap.org (it is much better in JOSM though). Thats
>> bad because it means less eyeballs to make sure we have correct
>> information. A wiki-like interface for all our metadata would solve
>> both these examples and for good reasons wikidata is not the right
>> place for this data, but a community run wikibase could probably work
>> just fine.

In a community (USA-based) "Mappy Hour" I participated in last night about doing imports well, one important question asked about imports which either are or tend towards actual vandalism.  The (correctly offered) answers are (at least) two-fold:  secondarily, there is complexity to doing imports (especially) well that makes for a "high bar" to diminish "script kiddies" and the unsophisticated from largely ruining our map data (though this can and does happen) and primarily and much more importantly, there are quite a fair number of both human eyeballs and 'bot-watching data sentinels which pay attention to bad data entering the map.  Sufficient (at least for now) enough that this project has a handle on it.  Not without somewhat pricey vigilance and sometimes friction, but certainly much more good data enters than bad, so OSM is "winning that race."

>> WDYT?

I told Original Poster pangoSE what I think in terms of a medium- to longer-term approach to these sort of "Web 3.0-style" introductions to OSM in a four-point outline in my last post to this (or a related?) thread.  It is (and remains MY turn to ask YOU, pangoSE):  What do YOU think?" (about my longer-term approach and four-point post).  Can we get YOUR feedback to THAT reply?

Let's not talk "past" each other, let's talk "to" each other.


More information about the talk mailing list