Download presentation
Presentation is loading. Please wait.
Published byWilla Jefferson Modified over 9 years ago
1
Epistemology and Engineered Systems (Frey, MIT) Epistemology - the branch of philosophy concerned with the nature of knowledge, justification, evidence and related notions.
2
Knowledge in Engineering different than for Science? Science proposes theory that explain phenomena, but observations may disprove –better theories by evolution (or revolution) Engineering certainly uses science at times But engineering needs normative logics –can’t properly analyze specs without it? –does Simon really say this?
3
Popper’s falsifiability criterion? Any respectable theory of science is falsifiable –“God is love” is not falsifiable –“Turner was born in Philadelphia” is falsifiable –this is not perjorative (ethics, morality, faith) Does Simon propose we propose falsifiable theories about “designs” ? Btw, are software requirements a theory about the released product?
4
Normative concepts? Normative statements fail to be falsifiable –but we might transform them into examples: NormativeFalsifiable Hypothesis The preferred choice is the alternative with highest expected utility Engineering design performed without the axiomatic framework of decision theory will result in an attendant loss which is on average a factor of two or more in profitability Maintain the independence of functional requirements Designs whose functional requirements are coupled have < 1% probability of meeting their functional requirements
5
Can Math be the Epistemological basis of a Science of Design? NSF states a preference for math rigor in its support of a science of design Math concerned with self-consistent sets of propositions based on axioms (and its cool!) Design is a human activity –cognitive limits, psych, group dynamics, social reality
6
Evaluation of Design Theories based on Practice? Evaluation of design theories by observation of its effects on the practice of design and evaluation of outcomes? –medicine: germ theory that proved a benefit by evaluation of practice –statistics: theory of optimal design of experiments resulted in less improvement of systems than would have been achieved with response surface methods (Box) –an existence proof that math devoid of human factors cannot represent design problems well? This must be a probabilisitic exercise –caution regarding just a few “successes” or “failures”
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.