[OSM-talk] Announcing Daylight Map Distribution

[Florian Lohoff]

On Mon, Mar 09, 2020 at 05:08:10PM -0700, Michal Migurski wrote:
> Hi everyone,
> 
> I’m writing to let you know about a new OpenStreetMap project Facebook
> just released. It’s called Daylight Map Distribution. Daylight is a
> complete, downloadable preview of OpenStreetMap data we plan to start
> using in a number of our public maps:

I think its a humble approach to try to identify (un)intentional bad
edits. A lot of people try to deal with this.

I am doing a lot of QA myself and i look at OSMCha changesets in 
my greater surrounding on a daily basis.

I fail to see a sane technical way of producing consistent map data
out of some intermingled data of which some changesets have been 
flagged/removed.

I heard Mapbox is working on this and divide data spatially not as
a sequence of changeset. My impression would be that this way
you could produce a "nice looking map" but most likely it will
break for routing purposes in most horrible ways where ways suddenly
are not connected anymore as some changeset inbetween has been
withdrawn/rejected.

So i'd guess the way you and IIRC Mapbox try to solve the vandalism/bad
edit issue is a labour and machine learning intensive task which you
cant win. Once you eliminate changesets you fall behind and you
pile up inconsistencies. This is, i guess, the reason for
your "one shot" dump of your current internal state.

So from my perspective the vandalism/bad edit issue will only be
fixable if we have some review process (Not that i would suggest one)
for strictly sequential changesets where review must be in order
and a once rejected/withdrawn changeset can only be requeued not
put in that sequential place again. And even then you'll see 
vandalism sneak by with innocent looking edits or intentional
3rd party validation.

So i'd love to hear more thoughts about long term ideas how to solve
this in a collaborative manner. OSMCha is probably not the final
solution but currently it brings together analysis, be it human
or machine learning in a transparent way, not that it currently
has an impact on the main OSM database.

Flo
-- 
Florian Lohoff                                                 f at zz.de
        UTF-8 Test: The 🐈 ran after a 🐁, but the 🐁 ran away
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20200310/d36544fc/attachment.sig>