[Design] SotM11: BoF meeting

[Michal Migurski]

On Sep 13, 2011, at 8:32 AM, Tom Hughes wrote:

> On 12/09/11 00:42, Gregory wrote:
> 
>> At SotM 11 we had a BoF informal meet-up outside. I took some rough
>> notes. It would be good if someone could clean them up, then they can be
>> sent to the talk list as there weren't many people there.
> 
> Who exactly took part in this little side channel chinwag?

Myself, Matt Amos, Mikel Maron, Gregory, Aaron Huslage, SteveC (near the end), and I'm blanking on the other names.


>> Need to know use cases
> 
> Agreed - we just need to be careful to make it the use cases we want to support rather than all the use cases anybody on the talk list would like us to support.

More broadly, we talked about the idea that it made little sense to move forward on potential designs before identifying the audiences who needed to be served by these designs. It can be quite contentious to move directly toward proposed solutions unless we successfully de-personalize the issues through the use of tools like personas and use-cases.

Examples of personas might include experienced editors, hopeful newbies, downstream data consumers, upstream bulk data producers, etc.


>> Get the community very involved in Round 0: creating a design brief.
>> Ask a survey. What do you want on osm.org <http://osm.org> % of how much
>> you want it
> 
> If you go down this route somebody has to be ready to stand up to the community and say no to bad ideas, no matter how much support they get.
> 
> I can guarantee you, based on past experience, that you will get a lot of ideas, some of which will have lots of vocal support, but which will still be bad ideas.

The phrasing of questions here is critical. I believe it would be counter productive to ask people what they want the OSM site to look like. Instead, we should ask experiential questions that get closer to people's needs. Examples: how recently did you use the website, why did you go there, how long did it take you to find what you were looking for, etc. There's also a risk here of missing out on participants who might not even be aware of the OSM project in the first place.


> Is that stomach in the people that do the cartography? or the people that were present for the discussion? was there any overlap?
> 
> I'm not saying we shouldn't change it, just interested in the background to this statement.

I suggested that the current Mapnik cartography does an excellent job of showing the depth and breadth of the data in OSM, but may look overwhelming as a usable map. Confusion exists, along the lines of "I don't like OSM because the highways are blue." I don't believe it makes sense to discuss changing the cartography until we understand who our audience is for the site, but once we're there the visible cartography may come up as a point of difficulty for new users.


>> Can we get stats from Grant/Matt
> 
> The stats that we have are here:
> 
> http://stats.openstreetmap.org/cgi-bin/awstats.pl?config=www.openstreetmap.org
> 
> Not sure that they help much though.


That's great!

The thought here was that if we had site usage stats at the app level along the lines of what http://mixpanel.com provides, then it would be possible to judge the long-term effectiveness of a new site design based on real, meaningful metrics such as retention, repeat visits, signup rate, and others. The choice of what those metrics actually are is clearly up for discussion and should happen early.

My mental image of a design process has three stages:
	1. Develop personas and choose metrics, with community participation, data-gathering, etc. The messy part.
	2. Appoint a directly-responsible group or individual to act on #1, produce and launch a new site. The fast part.
	3. Observe metrics and tweak the hell out of #2 slowly, over time in response to changes. The long part.

-mike.

----------------------------------------------------------------
michal migurski- mike at stamen.com
                 415.558.1610