Archive for February, 2007|Monthly archive page

Journalism and GIS

Watched a good program on Frontline last night about what is happening in the news industry.  Particularly interesting is the part about “hyperlocalization” of newspapers.  An old saying goes: dog-bites-man is not news, but man-bites-dog is.  Not for hyperlocal news though.  If a dog in my neighborhood bites someone, I want to know about it – quickly.

A neighbor of mine down the street saw a suspicious looking truck driving slowly around.  She emailed a long list of neighbors speculating what it might be doing.  Replies and replies to replies evolved the speculation towards panic.  Remember the lady in the Simpsons always saying “think of the children!”

So what does this have to do with GIS programming?

Eventually I think there will be a grid of video cameras, much like what has happened in Britain.  As this proliferates there will be a need for GIS tools to interact with a map of icons representing video feed sites.  I can imagine a map based site whereby you could track each step of a car along a path in your neighborhood by clicking on near real time video feed icons on a map.

I expect neighborhood associations will some day require RFIDs in all cars before allowing entry.  With congestion pricing in UK and toll road RFID in US, the infrastructure is already here.  Also expect the car insurance industry to offer discounts for those agreeing when and where to drive, as verified by RFID and/or GPS tracking.   Tools that connect the dots of RFID readings on a map in a sensible manner will be a challenge.  Routing applications will be able to plan trips using conditions from real-time feeds or based on historic time of day stats.

Maybe as news feeds include GeoRSS tags I will be able to write tools to allow support subscription based on distance:  Murders within 10 miles, burglaries within 2 miles, vandalism within 1 mile.

While I do regret seeing good newspapers suffering, their cartography is usually less than impressive.  Cartographic presentations of news summaries could also be presented in many interesting ways.  Even of classified adds, like a colorramped raster of garage sales per square mile.

Names, Places & Patterns

I’ve noticed a disproportionate number of place names that end with the short a (ah) sound:  America, Africa, Austrialia, Alaska, Alabama, … Zambia.  I wonder if this is some sort of linguistic universal for place names.  Kinda reminds me of relatives with odd names  given to them by their first born grandchildren.  These names tend to stick.

Names matter though.  Dave has discussed the virtues of simplicity.  This all sounds fine, but it should be noted that simplicity is typically achieved through a divide and conquer strategy.  The result is often stovepipes.  GIS is sometimes promoted as a means to integrate these stovepipes.  Indeed, a few years ago the mantra was “Geography is Our Common Language“.   Sometimes all the different dialects in this language are overlooked.  It’s almost as if systems integration of these stovepipes via geography requires something akin to a rosetta stone.

For example, in an electric utility each department might have a slightly different idea of what a “pole” is.   Different departments have developed applications that manage assets associated with poles.  Some departments may see a pole as simply a place (like the north pole) where there are assets , while others may see it as the wood (or steel) object existing at a location.  Designing an enterprise data model satisficing each department becomes quite challenging.  ESRI has attempted this with its various data models.  The comments I’ve heard is that these models try so hard to please everyone, they ended up pleasing no one.  I’d be interesting in hearing other views.

It seems like instead of data models, it would be more useful to have examples of patterns implemented on top of a data model.  For example, something like a Data Mapper pattern that creates an editable InMemory featureclass from some enterprise (non-GIS) table.  By the way why can’t I edit a featureclass based on an InMemoryWorkspaceFactory?  It’s not as if there’s a file protection issue … OK, back to work.