Argumentation as Engineering and Vice Versa David E. Goldberg Illinois Foundry for Innovation in Engineering Education University of Illinois at Urbana-Champaign Urbana, Illinois 61801 USA [email protected]
May 17, 2015
Argumentation as Engineeringand Vice Versa
David E. GoldbergIllinois Foundry for Innovation in Engineering EducationUniversity of Illinois at Urbana-ChampaignUrbana, Illinois 61801 [email protected]
Vice Versa Came First
• Provocative title interesting, but vice versa came first.
• Developed economy of models argument as part of Design of Innovation.
• Found Toulmin’s model useful as theoretical way to connect formal and informal engineering modeling.
• Started teaching this in senior design and it proved helpful.
Roadmap
• What Engineers Know & How They Know It.• Engineering modeling lesson from a life in genetic
algorithms.• A demarcation problem. • An economy of models & a modeling spectrum.• Lessons from senior design & the missing basics.• Qualitative modeling as missing skill.• Toulmin as a way to articulate & unify engineering modeling.• Tales from the trenches: Toulmin & tortillas.• Argumentation as engineer: Homo habilis & Gary Klein.
Engineering vs. Scientific Knowledge• Vincenti distinguishes engineering
knowledge from science with examples from aeronautical engineering history.
• Suggests engineering is not merely applied science.
• Two Vincenti cases:– Control volume models.– Flush riveting.
• Quantitative & qualitative models that are different because of their usage.
• Can we go beyond distinctive historical exemplars?
A Life in Genetic Algorithms
• Met John Holland in 1980 upon return to Michigan for PhD.
• Did dissertation applying GAs to gas pipeline optimization and rule learning.
• Needed better understanding to improve GAs.
• Received criticism for my “engineering style” of modeling.
• Models not “proper” or “rigorous” but they were helping me design faster, more effective GAs.
• Could I make rigorous defense of my method?
The Science-Engineering Demarcation Problem
• Engineers & scientists think in terms of models.• Scientists is in business of model making.• Engineer is in business of artifact making. Model making &
usage is instrumental to that aim.• In era of technoscience, models themselves not necessarily
distinct.• Engineers explicitly, necessarily & systematically use &
develop range of models with different precision-accuracy and costs: an economy of models.
• This economy of models fairly reliable demarcation of engineering modeling practice from science.
An Economy of Modeling
ε, Error
C, Cost of Modeling
Engineer/Inventor
Scientist/Mathematician
Spectrum of Models
Low Cost/High Error
High Cost/Low Error
Unarticulated Wisdom
Articulated QualitativeModel
DimensionalModels
FacetwiseModels
Equations of Motion
Qual-Quant Divide
Modeling Costs and Benefits
• Engineer is economic modeler when marginal costs do not exceed marginal benefits: ΔC ≤ ΔB.
• Benefit to what: To designed artifact.• 3 points:
– Calculation usually not explicit.– But modeling economy taught in pedagogy: e.g. Statics before
Dynamics.– Uneconomic model use common engineering manager’s complaint:
Modeling for modeling’s sake.• Scientist studies at frontiers: New model (error frontier) or better model.• Engineering modeling often cost improvement (lower error at given cost)
or improvement of modeling for practice: e.g. FEM vs. analytical elasticity solutions.
Approach of Design of Innovation
• First part of DoI methodological. • Applied modeling methodology to
selectorecombinative GA design problem.
• Constructed little models, quantitative models of different facets, integrating them to design and tune GAs that scaled to large hard problems.
• Quantitative analysis was prime concern.
• Concern for modeling left (qualitative models) came with engineering education reform efforts.
Lessons from Senior Design
• Coached 20 years of senior design.• Students
– expect clean problems with well-defined data &– are Pavlovian dogs when it comes to Newton’s laws or
Maxwell’s equations. • Real-world problems & data
– are ill-defined;– come in form of narrative;– vary in feasibility & quality
• Students have trouble making sense of problem & data.• Misled by their classroom experience of clean problems, with easy,
single solution, and spend first half of course unlearning.
The Missing Basics of Engineering
• “The basics” = math, science, and engineering science.• Reflections on 20 years in industry-sponsored senior
design.• After 4 years students don’t know how to
– Question: Socrates 101.– Label: Aristotle 101.– Model conceptually: Hume 101 & Aristotle 102.– Decompose: Descartes 101.– Experiment/Measure: Bacon-Locke 101.– Visualize/draw: da Vinci-Monge 101.– Communicate: Newman 101
• Call these the missing basics (MBs).• Using term “soft” accepts MBs as outside engineering.• Fundamental to engineering, organizational & learning
prowess. Socrates (470-399 BC)
12 © David E. Goldberg 2010
How It Works: Key to Engineering
• A key qualitative model in engineering is representation of causal chain of the way things work (or not):– As narrative.– Or diagram.– Or working prototype.
• “This led to this led to this.”• Critical model, but students
think “if no equation, no model.”
• Field example.
The Tortilla Problem
• Interesting example in tortilla factory.
• Company was using too much dusting flour relative to historical recollection.
• Flour cost was rising.• Wanted students to study
process and reduce dusting flour usage.
© David E. Goldberg 2009
• Students heard story.• Too much flour gets in air
flour burns falls on tortilla customer mistakes for mold complaint.
• Causal chain a model.• Students don’t recognize as
model.• How can we help them?
Burnt-Flour-as-Mold Problem15
Help from Argumentation Theory
• 1958 book by philosopher Stephen Toulmin formed basis of argumentation theory.
• How do people really make arguments?
• How do people give reasons for what they think or do?
• Form of reasoning ties together formal and informal engineering reasoning.
Formal Reasoning: Logic
• Modus ponens (modus ponendo ponens: mode that affirms by affirming): – if p then q– p is true– therefore q is true
• Method of mathematical logic & formal reasoning.
• Note: Once premises and rules in place, formal logic derives conclusions mechanistically.
Aristotle (384-322 BCE)
Toulmin: Elements of a Human Argument
• Like modus ponens:– Claim. A single statement advanced for the adherence of
others.– Grounds. A statement about persons, conditions, events, or
things that says support is available to provide a reason for a claim.
– Warrant. A general statement that justifies using the grounds as a basis for the claim
– Backing. Any support (specific instance, statistics, testimony, values, or credibility) that provides more specific data for the grounds or warrant.
– Qualifier. A statement that indicates the force of the argument (words such as certainly, possibly, probably, usually, or somewhat).
• Warrants can be generalizations, cause, sign, analogy, authority.• Backing can be anecdote, stats, testimony, credibility, and values.
Rieke, R. D & Sillars, M. O. (1997). Argumentation and critical decision making. New York: Longman.
€
G → C↑
B
↑
W
↑
Q↑
B
Back to the Tortillas: Burnt Flour Model
Grounds. Dusting flour is spread onto the moving dough on a continuous tortilla line.
Warrant. Excess flour becomes airborne & burn in the oven, deposits (authority).
Claim. Burnt black flour deposits is mistaken for mold, resulting in quality complaints
Qualifier. Sometimes
Backing. Client story & increased flour results in increased spot problem.
Tradeoff: Improve Backing or Solve Problem• In resource limited environment,
often face decision:– Should you improve warrant
and backing?– Or should you work on solving the problem?
• Can be difficult choice.• Key query: If assume correctness of warrant/backing &
wrong, will you fail to solve problem?• Tortilla problem: Students took explanation as true because
it didn’t affect investigation (reducing dusting flour usage reduces this side effect).
Argumentation as Engineering
• Outline of argument:1. Argument is an externalization of human thought processes
believed to be correct. 2. Artifacts are first fully expressive externalization of human
thought processes.3. Research on naturalistic decision making suggest that mental
simulation processes similar to those of argumentation.4. Therefore, first tech artifacts may be thought of as first expressive
evidence of argument-like processes in human ancestors. 5. OK to think oral or written arguments as offspring of first
engineering efforts. • Go back 2.5mya.
Tech Histories Don’t Go Back Far Enough
• Let’s start 2.5mya.• Homo habilis: First tool
maker, 4’-3” tall, 88 pounds, bipedal hominid.
• Lived on open savanna (Lake Turkana)
• Social.• Made and used stone flakes.• Did not speak (de Boer,
2005).de Boer, B. (2005) The Evolution of Speech, in: Brown, K (Ed.) Encyclopedia of Language and Linguistics 2nd edition, Elsevier.
Oldowan Tools
• First discovered by Louis Leakey.
• Used 2.5mya to 0.5 mya.• Know they were used by
scavengers.• Scrape carcass clean of
meat following kill by another animal.
• First known fully expressive externalization of human thought.
Homo Ergaster
• Better tools around 1.6-1.7 mya.
• Hand axes and cleaving tools with sharp edges.
• Butchering of large animals. • Tamed fire.• Still not talking.
Naturalistic Decision Making
• Gary Klein has studied how those under pressure make decisions.
• Naturalistic decision making.• Rational decision making used
infrequently & not under pressure.
• Cannot know Homo habilis mind.• Assumption: Homo habilis mind
was likely similar to our minds under pressure.
The Role of Mental Simulation
• Klein identifies different modes, recognition primed decision & constructive decision, for example.
• Mental simulation is key to all.• Many decisions made with single
simulation that shows adequacy.• Satisficing: First adequate
solution chosen.• Imagined artifact simulated step
by step.
Object Then Made & Used
• Steps:– Artifact imagined in context of
use.– Simulated step by step.– Device created.– Used for simulated purpose.
• Step by step imaging of adequacy a causal chain played out in mind.
• Thus, creation of first tech artifacts are first, external fully expressive evidence of human argument-like reasoning.
• Thus argumentation may be viewed as offspring of engineering.
Bottom Line
Engineering as Argumentation• Engineers are broad-spectrum
modelers.• Qual is part of the canon.• Toulmin’s model provides
unifying framework for math, science, & qualitative modeling.
• Introduction to students helpful in aligning behavior with needs of practice.
• Makes “soft skills” part of engineering not something apart.
Argumentation as Engineering• Notion of argument as external
representation of mental reasoning traces back to engineered artifacts.
• Tech as first shared representation of output of mental simulation.
• Packaging of mental constructs.• Sometimes forget ancient
prehistory of engineered artifacts.
For More Information
• Illinois Foundry for Innovation in Engineering Education (iFoundry): www.ifoundry.illinois.edu
• Philosophical writings on PhilSci archive: http://philsci-archive.pitt.edu/perl/search (author search for Goldberg).
• This and related powerpoints: www.slideshare.net/deg511