[Freegis-list] Re: GIS grant to help map cities worldwide

[Adrian Custer]

Mr. Dietze,

On Mon, 2003-02-10 at 06:26, Michael C Dietze wrote: 
> I have been continually suprised by the amount
> of energy spent defending the _concept_ of free GIS, 

This list has done many things, from informing people about new programs
to advocacy for political action. In this case it seemed to be a member
of the CGIAR wondering why he should struggle with free geographic
software rather than buy something off the shelf that seems to do what
he wants to do. This is a valid question. Why struggle when something
easy can be purchased? The only reason would be if *Free* gives you
something useful that non-free does not. This is how the conversation
slips from discussion into advocacy. 

The CGIAR, by the way, is a fairly large group. It will be involved in
geographic research extensively over the next few decades. Therefore one
of the question the CGIAR faces is how to leverage the substantial
upcoming investments of money, time and effort which they will make to
obtain the best systems to gather, analyze and present geographic
information. The CGIAR is also linked with other very large
inter-governmental organizations including the UN (FA0,UNEP) so that
some of the costs and benefits might possibly be widely shared which
changes the issue when compared to NGOs. The Brazilian government, for
instance, developed its own software package since it could defray the
costs across all the agencies which would share the system. The CGIAR is
going to spend a lot of money on hardware, software and salaries to
support their GIS activities. Since it's ultimately our money they are
spending, we may also ask ourselves what we want out of that money. 

I'm developing a separate response that is going to attempt to lay out
the decisions facing organizations wishing to implement geographic
capabilities and may be of interest to you. It will take a few days of
work though.

> Telling them to hire a consultant isn't an answer that resonates with me.

Honestly, here it seems that you might start to think differently about
what a GIS really is and how organizations can best develop such a
thing. You've probably heard and developed definitions of Geographic
Information \emph{Systems}. All the definitions developed recently try
explicitly to de-emphasize the software component. G.I.Systems are
really a series of methods to obtain, manipulate and present geographic
data, methods which are carried out by people with a high degree of
training. The people involved in the system are the most critical, in
terms of costs, in terms of training, in terms of probability of success
and in terms of perpetuation of the system within an organization.

When viewed in terms of people, the use of consultants for all GIS
tasks, quite apart from the *freedom* of the software used, starts to
make more sense. Organizations of all kinds contract out lots of
technical work, like chemical analyses of soil samples. GIS seems
similar to me. Find really qualified people and they can probably do
your project cheaper, quicker and with much more sophistication than you
could do your project in house. Why does the idea of buying software and
hiring a newly trained student seem more reasonable than farming out the
work to people who do nothing else for a living? Do we have a good sense
that one is cheaper than the other? Do we know that one is more
effective? 

GIS requires expertise, requires finding data, checking it, manipulating
it, checking it, combining it.... This is the standard expertise of
people who have been trained in GIS. Others have a much deeper
understanding of GIS and software code and robustness issues and
analytic issues. The difference between my muddling through an analysis
and a team of real experts performing the same analysis is significant.


> My vision is to see a Free GIS that is:

Your vision is widely shared. However, it will take several years to
become a reality. What is possible today, however, is for a group of
programmers to develop an effective solution for a particular
organization using all free software. In a few years, such a solution
will be available for download or with other distributions of free
software such as the operating system distributions. 

> p.s. Adrian, I think it's more than a little strong to say that "even the
> best spatial ecologists don't understand <long list of stuff>". 

Fine, but I disagree with you. Not one ecologist on campus doing
geographic ecological research has heard of the "Modifiable Areal Unit
Problem" and I see repeated presentations of garbage produced by very
good people who simply have not grasped the difficulty of the spatial
dimension. Remember that the analysis of dynamics, that is adding the
temporal dimension to ecological analysis, only took off post-war. We
don't yet really understand how to deal with dynamic analysis in ecology
and that involves only a single, uni-directional dimension. Spatial
analysis involves three (at non-planck scales) with actions possible in
any direction. To think that we will understand space and how to analyze
organisms in space without a real struggle seems improbable. 

There are many ecologists with good training doing GIS but honestly the
field is still really young and a whole lot is not understood. I finally
found last month a book with the title "Geographic Information Analysis"
which seemed to be the first book providing a real overview of the
current techniques in spatial analysis. The book doesn't, on a brief
perusal, seem to do what I've been struggling with which is to assess
why we have these particular techniques and how do they fit in the realm
of all possible techniques of spatial analysis. I've spend four or five
years thinking about what spatial analysis really is and spent last
spring trying to bring my thoughts together but have not yet gotten a
real grip on how to think about it, let alone how to carry it out. I
personally suspect that our mathematical understanding is not yet deep
enough to actually undertake true spatial analysis. The hallmark of
scientific measurement and analysis is the coherency, repeatability,
comparability and robustness of reported results. Current spatial
analysis in simply not there yet. Landscape ecology is all about case
studies and we have little ability to compare one analysis to another.
We have a vast area to cover before we can actually say simple things
about the spatiality of organisms. So yes, my statement is strong but it
comes from a long period of reflection and a real desire to change the
current state of affairs. Right now we simply do not understand how
little we understand.

> You might want to take a look at our lab's paper in the new
> (Jan '03) issue of Ecology.

Sure, I'll wander over to the library and take a look. 


And please do not take my statements as a reproach of current efforts or
some kind of assumption that everyone is doing useless work. We are all
struggling with these issues and doing the best we can. There have been
some wonderful and important ideas developed over the past two decades
but these remain hints, suggestions of the importance of certain
factors, illustrations of the great difficulties ahead.

cheers,
adrian