Speaking Truth to Power:
An Agenda for Change
Presented at Spatial Data Quality Symposium, Quebec City, May
1998
Nicholas Chrisman, Geography
University Of Washington
Abstract
Despite substantial attention by the research community, data quality perspectives
still do not inform the developments of new tools. This paper introduces
a different perspective on "truth" in science based on studies
of science and technology. Then it develops an example of the quality dimensions
hidden in the black box of coordinate transformations. This paper calls
for actions to bring greater capacities (like improved estimation algorithms)
to users.
Outline of Presentation
The Gap
Data Quality (Accuracy Assessment)
Come a long way in 15 years...
[AUTO-CARTO 6 paper: The role of quality information in the long-term
functioning of GIS]
Expanding interest in research community
This third symposium as evidence
Standards efforts abound
Fitness for Use: basic principle
These changes occur in a context of massive social, economic and technical
change.
Yet
Data Quality has not become a driving force on Software Toolkit.
Speaking Truth to Power
A first reading:
Quaker resistance to political power:
Persuasion should defeat prejudice.
Original paper I planned to present...
Underlying assumption: Accuracy assessment connects to "truth"
A second reading:
requires a detour through a more careful understanding of the role of truth
in science:
Studies of Science and Technology
Studies of Science and Technology
Some of my sources:
Latour, Bruno and Woolgar, Steven, 1986, Laboratory Life: The Construction
of Scientific Facts (Princeton Univ. Press), 2nd edition.
Latour, Bruno, 1987, Science in Action (Harvard Univ. Press).
Latour, Bruno, 1988, The Pasteurization of France (Harvard Univ.
Press).
Latour, Bruno, 1993, We Never Were Modern (Harvard Univ. Press).
Latour, Bruno, 1996, ARAMIS or the Love of Technology (Harvard Univ.
Press).
Shapin, Stephen and Schaffer, Simon, 1985, Leviathan and the Air Pump:
Hobbes, Boyle, and the Experimental Life (Princeton Univ. Press).
Fujimura, Joan H., 1992, Crafting Science: Standardized Packages, Boundary
Objects, and "Translation", In Science as Practice and
Culture, edited by A. Pickering (Univ. of Chicago Press), pp. 168-211.
Galison, Peter, 1997, Image and Logic (Univ. of Chicago Press).
Porter, Theodore M., 1995, Trust in Numbers: The Pursuit of Objectivity
in Science and Public Life (Princeton Univ. Press).
A literature reacting to the accepted "paradigm"
Kuhn, T. S., 1962, The Structure of Scientific Revolutions (Chicago
Univ. Press).
Key Concepts
Expansion and Diversification of Science
Cumulative development of theory and knowledge
Positivist interpretation: Observations => Theory
Science creates impacts on society.
Kuhn's Critique
Normal Science vs. Revolutionary Paradigm Shifts
Anti-positivist: Theory pervades observations
Sociology of Scientific Knowledge
"Strong Program" p; Society controls Science
Social Constructionism
Two-way interaction between society and science
· Black boxes · Actor-Network Theory
· Irreversibility · Translation
· Boundary Objects · Trading Zones (pidgin)
Science and Truth
Scientific Method
Since Bacon: based on observations of nature
Truth is there to be discovered.
Observer can create error, but not truth.
(Many sources of human failings)
First Denunciation
Truth not inherent in nature, but projected there by social action.
[Long line of critics: Durkheim to post-modernists]
Society: hard; Nature: soft
Second Denunciation
Objective forces shape society; Nature transcends
Nature: hard; Society soft
Bruno Latour's View of Modernism
"Constitution" of Modern Society requires:
First Paradox
Nature is not our construction; it is transcendent and surpasses us infinitely.
Society is our free construction; it is immanent to our action.
Second Paradox
Nature is our artificial construction in the laboratory; it is immanent.
Society is our construction; it is transcendent and surpasses us infinitely.
Constitution
First Guarantee: even though we construct Nature, Nature is as if we did
not construct it.
Second Guarantee: even though we did not construct Society, Society is as
if we did construct it.
Third Guarantee: Nature and Society must remain absolutely distinct: the
work of purification must remain absolutely distinct from mediation. (We
never were Modern, Fig 2.1 p. 32)
Latour's Response
Modernist division between
Nature and Society
no longer tenable.
(So many hybrid technical creations:
part social, part natural)
Post-modernist critique continues the cycle of denunciations.
Maybe "we never were modern" p; meaning that
the divisions were not as complete...
Why should we care?
Yes, French academic philosophers can always stir up huge clouds of talk...
BUT, this scheme has practical consequences for the design and operation
of GIS.
Theories of Error (in the sense of falsehood)
· Assymetrical Explanation
Truth implicit in Nature;
Falsehood arises from human sources.
· First approach to Symmetry
Truth and Falsehood both have human origins.
· Generalized Principle of Symmetry
Both Nature and Society are to be explained.
Hybrid "quasi-objects" implicated.
Speaking Truth to Power Reinterpreted
Assessment of Data Quality
confronts powerful institutions with observations they might not be so pleased
to discover.
Persuasion
lies at the core of scientific activity.
Great scientists assembled all kinds of allies: social, political, natural,
logical, observed
No Guarantees
of arguments / approaches that prevail automatically
An Agenda for Change
Need to avoid the tyranny of
the way things have always been done.
Research community should voice concerns about the solutions imbedded in
routine technology.
Examples
Prime case:
· Coordinate Transformations used for Registration
Less developed cases:
· Wetlands Mapping
· Low-Level Radioactive Waste Facility Siting
Coordinate Transformations
If GIS integrates diverse sources, geometric registration required.
- digitizer device units => projection of map
- remote sensing image => map space
- air photograph => control points
Despite its prevalence, registration software is rarely discussed or decribed.
- Input: coordinates of "tic points" in source and destination
coordinates
- Solve best fit transformation, using affine (possibly some alternate
geometric model)
A BLACK BOX: encapsulates science, hides details.
Prying Open the Black Box
Coordinate transformations involve decisions:
Each decision has a social component.
Number of Points
Estimation requirements
How the black box is presented:
"Select 4 widely spaced points common to maps A and B to be used as
tics for A."
(ESRI, 1991, Map Projections and Coordinate Management: Concepts and Procedures,
page 5-13)
Image registration extension only permits 4 points. (PlanetOne,
http://www.planetonegis.com)
Corner points preferred despite known problems (sheet corners not ground
surveyed).
Why the affine?
Affine fits a different X and Y scale.
If sources really in same projection, X=Y.
Parameters reported in format that includes rotation and scaling in common
values (user not informed)
Rotation, translation and scaling NOT adequate to convert a cylindric projection
to a conic (eg. UTM to State Plane) [over a sufficiently large extent...]
Possible explanations:
Legacy effect: residual of hand calculator era (?) (though this would argue
for similarity, not affine)
Differences in solutions insignificant (?)
Justified for printed maps whose paper might shrink more in one direction
than another (?)
Software written without professional expertise.
In any event: research community has failed.
Why Least Squares?
Simplicity:
Closed form equation: minimal software effort
Efficiency:
Best Linear Unbiased Estimator (BLUE) uses all information available to
produce estimates
BUT
only has this property if points come from
normal iid (independently and identically distributed)
Depends on Division of Error
- Systematic
- Random
- Blunders (not handled!)
Fundamentally a social division of labor
What is a Blunder?
Random Error:
defined as what Least Squares requires.
Numerical properties extend to procedures
p; basically, random = good error
Blunder Detection:
User's responsibility (reversed digits, wrong object, ...)
Residuals "too big"
but Least Squares gives great weight to outliers, so residuals can be misleading.
With 4 points, blunder ensures wrong fit; more points required to be able
to select.
A blunder is an data point that should not be used in Least Squares estimation...
Alternatives
Old constraints (particularly calculation) no longer applicable.
Robust Statistics:
estimation less dependent on distributions
outlier detection; exploratory procedures...
Least Median Squares: an example
- sample possible combinations of points
- fit regression with least median squares
- select best solution
Can tolerate up to 50% blunders (high breakdown point)
No weighting functions or complex user intervention
Comparison of Residuals
[Shiahn-Wern Shyue, PhD, 1989, University of Washington]
Solution to Oblique Photogrammetry:
27 control points including three blunders
Solved with Least Median Squares (LMS)
and Ordinary Least Squares (OLS)
Note that regular methods of blunder detection with OLS would correctly
select the double (X and Y) outlier, but not immediately detect the other
two.
Redividing the Labor
Robust Statistics can tolerate blunders.
Models such as LMS offer a redefinition of error.
Research community can innovate.
But, unpublished PhD dissertations do not help get the word out...
Innovations should be launched into practice.
GIS users should expect better registration, but don't know to ask.
Software vendors should change.
The "most powerful" techniques are probably measured on the wrong
scale.
Persuasion requires a comprehensive approach to socio-technical networks.
Conclusions
No external "truth";
Social, economic, political, ethical choices appear in many guises.
No clean slate;
Historical path of technology leaves traces;
"legacy" decisions are maintained by being converted into unquestioned
"black boxes"
disciplinary background imperfectly shared
No villains;
Researchers, users, software developers each have responsibilities to communicate
in ways that will reach each other.
Innovative ideas should alter practices.
Version of 15 January 1998