[Imports] [OpenSeaMap-develop] Bad listoflights import

[Paul Norman]

> -----Original Message-----
> From: Malcolm Herring [mailto:malcolm.herring at btinternet.com]
> Subject: Re: [Imports] [OpenSeaMap-develop] Bad listoflights import
> 
> Some observations:
> 
> 1. Our (seamark:) tags are not rendered or interpreted by any the land
> map browsers, applications, etc., so do not present conflicts to the
> wider community. Only those specialist browsers & applications actually
> interested in seamarks will be affected, so we need to have any
> discussions with those people, rather than the community as a whole.

If I understand you correctly, you're arguing that because the import 
isn't rendered on any of the default osm.org layers you don't have to 
consult with the local lists? If that's the argument, I'd have to regard
it as wrong. You're obliged to consult before importing data.
 
> 2. Many land map imports, landsat tracings, and even member surveyed
> edits have similarly large positional errors, so why are we being
> castigated for importing inaccurate data?

I'd take issue with any imports that were 100m or more out. Imports are 
held to a higher standard than traditional data gathering methods, but 
100m seems extreme for even landsat. Don't forget, the import in question
was into area with good aerial imagery, not just landsat.

In any case, I'd take issue with any data with positions of 100m or more
off and I expect so would most of the local community.


> The whole idea of bulk imports
> to get a starting point from which our mapping contributors can survey &
> refine.

There's definitely not a community consensus about that, either with the 
community as a whole or with talk-ca@
 
> 3. We do not use the FIXME tag, we use the seamark:fixme tag.

That's one of the tagging issues that I raised and was not dealt with
before imports were resumed.

> 4. The intended users of the NGA LoL are commercial shipping who
> probably don't care about a 200m positional error. They just need to
> know an approximate position, then look out of the window and/or use
> their radar to actually locate the mark.

That explains why the data source is so inaccurate, but doesn't mean that
it's suitable for importing into OSM.