[OSM-talk] Moderating / Quality checking OSM contributions -- was: Re: OSmosa.net run now.., contribution model

[Kate Chapman]

Hi Jaakko,

I wonder if custom presets could help with the road classification.
We've had the same problem in parts of Indonesia, but that is because
the road classifications are simply translated.  Translating the word
primary doesn't mean much for the type of road it is.  Changing the
word used in the preset to the specific known government
classification could help.

I agree with you when a large amount of new people start the quality
can go down.  My theory is there needs to be a certain ratio of senior
mappers to new people to keep the quality high.  However I think as
people learn how to fix their mistakes the quality will go up again
and they will in turn be able to help new people avoid those mistakes.

Best,

-Kate

On Wed, Dec 7, 2011 at 12:18 AM, Jaakko Helleranta.com
<jaakko at helleranta.com> wrote:
> Yes. It's definitely
>
> --
> jaakko at helleranta.com * Skype: jhelleranta * Mobile: +509-37-269154  *
>  http://go.hel.cc/MyProfile
>
>
>
> On Tue, Dec 6, 2011 at 10:15 AM, Serge Wroclawski <emacsen at gmail.com> wrote:
>>
>> On Tue, Dec 6, 2011 at 9:37 AM, Jaakko Helleranta.com
>> <jaakko at helleranta.com> wrote:
>>
>> > That is: as the contributors' tech-savvyness decreases we see a clear
>> > hike in various sorts of data problems.
>>
>> This is an interesting conclusion.
>>
>> Do you have evidence for what you're saying other than anecdotal evidence?
>
>
> I don't have a study/paper to show but my strong experience is that for
> example when people who haven't been exposed to the concepts of different
> types of roads in their daily lives start to map it takes a long time for
> them to grasp the concepts that are much needed for the base map to make
> sense. In Haiti we saw a strong and w/ time increasing case of "Yellow
> screen of death" (in Mapnik) around the country between last fall and this
> spring/summer when the mappers got an idea of marking nearly all roads that
> they edited tertiary. There's been an infestation of POIs in the middle of
> the road, pretty big amounts of data with cApitaLIzATion all over the place,
> people fixing the data per the mis-aligned imagery, etc. This has taken a
> great amount effort from the few people (practically all foreigners at this
> point) in fixing the most blatant problems -- and that's away from making
> the project fly on other fronts.
>
>>
>> I've seen reports that say that as the total number of users
>> increases, the map quality goes up. You're saying the opposite, so I'd
>> like to see some kind of metric.
>
>
> I'm actually _not_ saying the opposite (in general). I'm saying that:
>
>>
>> > Bundle that with not-so-large active user base (of people who do quality
>> > fixing) and you have a problematic situation at hand.
>
>
> So, when (and hence, if) the number of less technically capable users (and
> I'm not talking GIS pros but generic ICT skills + generic geo understanding,
> such as the ability to read maps) goes up _and_ there isn't a strong
> community of more advanced users to not only guide them but also to help fix
> / point out their edit issues so they can fix them and learn from savvy that
> _then_ there's a good chance that there's a spike in quality problems.
>
>>
>> This is why I'm asking for your approach in coming to the conclusion.
>> I see multiple factors in play.
>
>
> Me too, as my clarification hopefully articulates better.
> We're still talking about so small numbers that any data we have on how
> things have worked may well (if not most probably) _not_ tell how things
> will work out in the future. Time will tell.
>
>>
>> > What I'd love to see and what the contributors here would also want to
>> > have is a system where contributors can flag their edits with something that
>> > would que the changesets in question to an (open) review stack.
>> > This kind of review que wouldn't actually moderate anything but would
>> > simply flag the changesets for review ("quality ensurance", if u may).
>
>
>>
>> We don't have a moderation system.
>
>
> I know, and that's why I shouldn't have even put it in the Subject. So, I'm
> talking about flagging changesets (that are already committed into the live
> db) for review.
>
>>
>> If you build one, then maybe
>> someone will think it's interesting/usefu and take a look.
>
>
> Which is why I asked "what would it take" (for someone to do it)?
>
>>
>> now we don't have one, and many people don't want one.
>
>
> +1 from me. I just want to add that we have to be realistic, too (on top of
> being idealistic). I'm hoping there are ways to combine best of the both
> worlds.
>
>> > how much it
>> > would/could cost to create something that would?
>>
>> I'd estimate about 5-10k USD for a good solid working prototype.
>
>
> Thanks. That's a good ball park figure to know/have an idea of.
>
> -Jaakko
>
> _______________________________________________
> talk mailing list
> talk at openstreetmap.org
> http://lists.openstreetmap.org/listinfo/talk
>