[Talk-GB] OSM Analysis New Data and bot

[SK53 on OSM]

Generally, I am still opposed to a bot. There is a substantial body of 
evidence that automated imports damage the ability to recruit and nuture 
new mappers. Recent posts about Latvia, Austria and The Netherlands on 
talk all substantiate this: in many cases the people recognising the 
issue were those who either carried out the import or agreed to it.

I think a completion bot is a distraction from a much more important issue.

In order to get  a better level of completeness in the UK what we need 
are more mappers. There are several ways to recruit mappers: they 
require a decent amount of hard work, and probably a broader range of 
skills than writing a bot. We need a more organised way of generating 
publicity on a regular basis both for national and local media. We need 
a better press kit. We need to move the emphasis of mapping from getting 
GPS tracks: dont get me wrong this is still valuable, but a local mapper 
without a GPS can do a fine job with Bing, OS OpenData, Walking Papers, 
a camera, and ground surveys. We need more outreach techniques: not just 
mapping parties, or pub meets or mini-mapping, but workshops for people 
interested in consuming data, workshops to review the data from 
particular usage perspectives (cyclists, walkers, sustainable living, 
wheelchair users, etc.). We could do with more supporting materials for 
such things: slideshows, posters,  how to organise .... I'm finding this 
ain't that easy, but at least I'm trying.

We also need to recognise that the more detailed each area becomes the 
harder it becomes for a new mapper to feel that they can contribute, not 
forgetting the "I might break something". If we are to devote effort to 
code its better directed at tools which can make the life of new mappers 
easier: this obviously includes contributing to existing editors, but it 
may mean creating new ones. It almost certainly means working to get a 
much more sophisticated OpenStreetBugs integrated into the rails port: 
many new mappers will initially be happy to point out bugs (see recent 
examples on OSM Help where the first thing someone wants to fix is a 
turn restriction).

I strongly dislike the meme "OS data is always more accurate than OSM", 
because it implies there's no point in doing surveys anyway. Yes, errors 
occur, although mainly in transcription rather than in surveying as can 
be seen by errors in using OSSV & OSL, but tools like ITO OSM Analysis 
and OSL Musical Chairs really help to pick up these errors: I've been 
able to go back to pictures and audio recordings and indeed verify that 
I'd not changed Street to Road when I copied the tag over from another 
way. There is also the spurious accuracy problem: people filling in a 
road name from OS Locator when there is *NO *evidence on the ground that 
the road has that name (pace RichardF in W Oxon): see my blog post on 
Kenyon Road 
<http://sk53-osm.blogspot.com/2011/02/mysterious-case-of-kenyon-road.html>. 
Many of the unnamed roads in the immediate vicinity of where I'm writing 
this are of that type: sometimes dogged persistence can nail down that 
the road is still called that, for instance from address information.

Take a look at Corby <http://osm.org/go/eu7EEN9>: its OSL road complete: 
a small part on the N edge was surveyed, the rest is largely from OSSV. 
There is a huge amount of information missing: footways, paths in parks, 
information about Places of Worship, other POIs. Corby is the classic 
sort of place which is less likely to receive attention from OSMers 
according to Muki's studies: its out of the way, it lacks a strong 
middle-class demographic. There are plenty of people living in places 
like this who are using Skobbler's apps, but we're never going to reach 
out to them if we do the easy bits from our armchairs and leave the 
harder less rewarding mapping activities for others.

Why not build a separate database & render which merges the missing 
names (& roads) from OSSV/OSL and OSM data, but is external to the OSM 
planet database. This could use many of the same techniques as a bot.

A bot is putting short-term gain ahead of our long-term interests.

Regards,

Jerry

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-gb/attachments/20110609/93c3d893/attachment-0001.html>