GIS for Politics Andrew U. Frank Geoinfo TU Vienna [email protected].

GIS for Politics

Andrew U. FrankGeoinfo TU [email protected]

Sept 10, 1998

GIS PlaNET 1998 2

Key Experience:

Comparing AM/FM with GIS in mid ‘80s:success reports for AM/FMhopeful projects with GIS

Why?

Sept 10, 1998

GIS PlaNET 1998 3

Why is a Technology Successful?Technologist push what is technologically

feasible.Success stories: Fax, wwwGIS did not grow as quickly,

despite potential for use.

Consider: Social demand!Do we respond to demands?

Sept 10, 1998

GIS PlaNET 1998 4

My background

dipl.ing. Surveying from ETH Zurichdoctorate in database modeling for GISNCGIA - U Maineresearch focus at TU Vienna:

modeling with cognitive science perspective

economics and marketing for GIS

Sept 10, 1998

GIS PlaNET 1998 5

My approach to GIS

initially:concern for the limited natural resources. I hoped GIS would help to a better world.

Folk theorem:More information leads to better decisions.

I still believe in rationality, even in politics.

Sept 10, 1998

GIS PlaNET 1998 6

GIS Today:

- successful in day to day management of administrative data

-support for planning

Rapid growth, but not commensurate with potential.

Sept 10, 1998

GIS PlaNET 1998 7

GI used in Political DecisionsSubstantial potential,

but seldom used.

Exceptions: -Political campaigns, - redistricting

Why not more use for decision making.

Sept 10, 1998

GIS PlaNET 1998 8

Reasons for the lack of use of GIS in political decision makingA technologist answer:Politicians are not interested.

A practical answer:GIS are too complex to be used.Improvements of user interfaces are

necessary!

Neither of the two are sufficient answers.

Sept 10, 1998

GIS PlaNET 1998 9

Does the GIS provide the Information Politicians need?Politicians focus on

- changes- decide on actions to change situation

to improve- concentrate on situation affecting

the populationSimilar question can be asked for other

potential users.

Sept 10, 1998

GIS PlaNET 1998 10

What Information Produces

a GISDetailed information about current state.StaticInformation often related to land cover

(data from remote sensing).Limited information on social parameters.

Sept 10, 1998

GIS PlaNET 1998 11

Case: European Agricultural PoliticsKnown: current situationRequired: a new policy to

- reduce cost to EU- improve environment- balance social situation

Possible actions: change rules for subsidiesEminently a “GIS” type problem.

Sept 10, 1998

GIS PlaNET 1998 12

What is Required?

Data about change.Integration of data from

environment, agriculture and social system.

Help to predict effects of rule changes.Models for processes.Methods to evaluate alternatives.

Sept 10, 1998

GIS PlaNET 1998 13

Data Quality Required

Politically fiable Generalized dataRational chain of arguments to link

observations to actionsExamples from past for justification

Sept 10, 1998

GIS PlaNET 1998 14

Integrate Physical and Human Geography DataThe issues are at the interaction of

humans with the environment.The GIS must integrate the data from

environmental sensors with the data from the administrative system.

Technology:Seamless integration of

raster and vector technologyInteroperability of GIS of different agencies

Sept 10, 1998

GIS PlaNET 1998 15

Technology required for spatio-temporal data

Change is crucial, To describe change

a temporal database is required.

Watchword: No GIS without time!

(the requirement is similar for most administrative applications of GIS)

Sept 10, 1998

GIS PlaNET 1998 16

Technology required to Model Process

Politicians decide about rules, not about states.(the same for urban planners).

Process models connect the rules with the state and result in a model of the future state.

This future state can be evaluated to assess the desirability of the new rule.

Sept 10, 1998

GIS PlaNET 1998 17

Limitation to Static Data is in the Logical FoundationFirst order predicate calculus is de rigor

for CS.Base for Relational Data Model.Complex ‘temporal logics.

Sept 10, 1998

GIS PlaNET 1998 18

Higher Order Languages

In first order languages functions, which are the model of process and change,are not generalizable

f (a) is first order, but not

for all f where f (a) = b is f (b) = a

Sept 10, 1998

GIS PlaNET 1998 19

Potential Solution

Use second order calculus!

For Technologists:- Functional programming language based on higher order languages

merged with- Object-Oriented Technology

Sept 10, 1998

GIS PlaNET 1998 20

Haskell

- a functional language with performance within a factor of 3 ... 7 to C++

- logically clean- spatial temporal databases using a

temporal ER model (Chorochronos project)

Sept 10, 1998

GIS PlaNET 1998 21

Conclusion 1: Geographic Information Product - GIP1. Start with potential user and the

decision which must be taken.2. Consider the information required for

the decision.3. Create an information product to

satisfy this need.4. Find technical solution to produce the

GIP.

Sept 10, 1998

GIS PlaNET 1998 22

Conclusion 2: Technology required to satisfy the need of Politicians:

- spatio-temporal database- integration of physical and social science

data- construction of spatial process models- interoperability

Sept 10, 1998

GIS PlaNET 1998 23

Conclusion 3: Research IssuesSemantic integration of data from

different sources using models of common processes

Assess Data Qualityfrom data through process to decision