[OSM-talk] checking before massive deletion

[Joseph Eisenberg]

Here in Indonesia there were many place nodes imported from official
sources, but the government database had very poor quality in large
parts of the country, especially my province.

At first I tried to move the nodes to the right place, and was
hesitant to delete them, but later I realized that I would waste many
hours trying to fix the damage.

I recommend deleting the nodes if you have found that they are all in
the wrong place by more than several kilometers. (Our place=village
nodes were off by 10, 20 or 50 kilometers, which isn't helpful, and
many were just administrative names)

If they are just 1 or 2 kilometers off, then it would be worth keeping
them with a fixme=* on each one for local mappers to correct the
location later.

But that doesn't sound like it is the case in this situation.

- Joseph Eisenberg

On 3/10/20, Mario Frasca <mario at anche.no> wrote:
> hello here,
>
> I wish to get some sort of support in a decision regarding data which I
> find hard to verify, and actually also hard to believe.
>
> years ago ArielRod completed this changeset:
> https://www.openstreetmap.org/changeset/17323627
>
> I have asked about the source, and got no reply.
>
> the above is not an isolated changeset, for there are at the moment 995
> place=* on ArielRod's name in the "Comarca Ngäbe Buglé", and 342 in
> Chiriquí.  these are the points added by this user, and not edited by
> others.  but there are more, where people have edited them downgrading
> place to isolated_dwelling, just in doubt instead of deletion, I suppose.
>
> the alleged source is the census for the year 2010, but the
> corresponding document, as far as I managed to search
> (https://inec.gob.pa/archivos/P3551P3551cuadro3.xls) does not contain
> coordinates.
>
> I have zoomed into quite a few of these points marked as place=hamlet,
> and in most cases I do not see anything that could justify the presence
> of a `isolated_dwelling`.  in some cases, there is something, but is it
> sheer chance?  could well be.
>
> one example for all, Node: Bulite (2416914927), marked as hamlet,
> supposedly 5 houses according to the census.
>
> https://www.openstreetmap.org/node/2416914927#map=17/8.54709/-81.76901&layers=D
>
> around this node, at less than 1.0km, there's 18 more place=hamlet, by
> the same user.
>
> in the same area, I do see a few buildings (I haven't mapped any), but
> not near the place=* nodes.
>
> of these 19 nodes, only 8 are actually mentioned, verbatim, in the 2010
> census.  but still, whether they have been thrown at a random location,
> or placed here for any other reason, that's obscure to me.
>
> so far the situation.
>
> I would like to do some cleaning up, and am considering deleting all
> such nodes added by this user, either edited by others, or not.  I have
> commented so much in the Panama Telegram Group, where a few local
> editors are active, asking for their opinion.  none has been expressed
> so far.  I would not want to act without some support.
>
> I could also leave it as it is.  after all, why should I care?
>
> comments will be very welcome,
>
> best regards, Mario Frasca
>
>
> _______________________________________________
> talk mailing list
> talk at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk
>