[Talk-us-massachusetts] road closure edit timing

[Greg Troxel]

(This note is about norms in MA.  I realize there are other parts of the
world where getting fresh bits is harder.)


The idea of how short a closure should be mapped recently came up in a
note about a road in Berlin.

https://www.openstreetmap.org/note/4631995

In this case, the road is closed and will be for 4 months.  I've been,
when I get to it, marking bridges impassable when it's been that length
of time or more.  I have not been doing it for 2-week closures.

I have always viewed "don't map short closures" as being "don't map
closures that are shorter or the same order of time as the update
interval of typical data users", and thus changing over time.
The wiki text about 6-9 months seems to date from 2009.

Currently, I see OsmAnd having an update interval of 1 month, and
Organic Maps of 1-2 months.  (One can use OsmAnd Live, but that's for
mapper nerds, not normies.)  MA has good networking; even people with
wifi only devices and no cell service can update maps on public wifi
quite easily.  (I don't tend to, but find myself having access to public
wifi at least weekly, without even trying.)

I really don't know of ways to use OSM data for routing that end up with
data less fresh than 1-2 months, unless someone chooses to build for
Garmin and not update.


I think it's more important for closed roads to show closed than open
roads to show open; navigating to a closure is worse than avoiding a
usable road.  This is especially true as roads get sparser.


Philosophically, I think it's better to optimize for data users that are
holding up their half of the bargain, rather than theoretical users that
might not exist.

So personally (today) I would draw the line at <30 days, don't edit, >60
days, edit, and in between it's an interesting discussion.


A question:

  Do you know of situations where people will be using OSM data for
  routing, such that they will be using data older than 1-2 months,
  assuming that they are making some small effort to have fresh data?
  (I don't.)

And opinions:

  What do you see as the expected length of time to decide yes/no on
  changing the map?   I suspect we all agree that 1 year, yes, and 1
  week, no, but where do you draw the line?

Greg