Kevin T. Kelly
Department of Philosophy
Carnegie Mellon University
Static science: Textbooks teach and defend a completed theoretical system.
Cumulative history: Bad old history supports the textbook image by asking how historical science contributed to the modern system in a cumulative way.
New history: New history shows that science is not cumulative. Rejected science is called "prejudice and supersition".
Historical symmetry: New history treats old scientific developments as systems in their own right, rather than as partial contributions to the modern scientific system.
Insights of new history:
Historical contingency: Method does not uniquely determine new beliefs [accidents in the lab, prior experience, individual psychology].
Paradigms: Shared, unquestioned background commitments lead to group efficiency [ontology, shared education, methodological approach].
Anomalies: Difficulties arise that resist solution according to the shared background.
Revolution: Sometimes new shared commitments are adopted to meet the anomalies, including new scientific rules and methods.
Theory-Ladenness: Facts are partly constituted by theoretical concepts, so the world changes when the background commitments change.
Incommensurability: The new tradition has to creatively rework the old concepts in new terms, so the changes are not cumulative.
Revolutionarity of discovery: New discoveries induce revolutions, and hence are not entirely factual.
Who's on first?: The historian, rather than the scientist, has the right perspective on science.
No place to stand: The normataive/descriptive distinction distinguishes "is" from "ought". History is the former; methodology is the latter. But the new history sees this distinction as a constitutive part of old history, rather than as an objective logical rule that all must obey. The value of the distinction depends on which history is a better "elucidation" of the historical data.
Normal science = research based on a shared paradigm.
Paradigm = a past scientific achievement with two properties:
Unprecedented enough to sustain a following.
Incomplete enough to suggest new research questions
Scientific tradition = research group that shares a paradigm. [e.g., Aristotelian dynamics, Ptolemaic astronomy, corpuscular optics].
Preparadigm science: constant debate over fundamentals, no clear winner, no detailed confrontation with data, ignore phenomena handled by other theories [e.g., ancient optics, 18th c. electricity].
Postparadigm science: sequence of paradigms separated by revolutions characterized by penetrating experimental investigations made possible by the shared labor of paradigm members [corpuscular optics, wave optics, quantum optics].
Arduous road: Difficult to forge first paradigm.
Data irrelevance: no agreement about which facts are relevant.
Blind lists of facts: Baconian science an herbals are proto-science.
Static data pool: The only uncontroversial data arise from shallow observation or technological problems.
False phenomena: without paradigm guidance, some false phenomena slip into the data pool.
Advent of first paradigm
Focuses research: Renders some experiments worthless and others very interesting. Concentrates research on narrow experiments that do not arise in ordinary life.
Censorship: Adherents of other traditions are ignored.
Disciplinary identity: Paradigm adoption was the test that separated each science from Philosophy.
Foundations backgrounded: No need for paradigm members to debate basic assumptions, so studies appear more focused and applied.
Rise of textbooks: Shared assumptions now form the curriculum of the discipline. Work on principles is reduced to a mopping up or pedagogical exercise.
Movement from books to articles: Scientific books are textbooks or popularizations. They don't count as real research, which is confined to narrow journal articles. Book authors are liable to be called "philosophers".
Technicality: the articles become unintelligible to laymen as they specialize.
Short time of crystallization: in electricity, the fluid paradigm solidified between 1740 and 1780, when it accounted for the Leyden jar.
Paradigm Incompleteness:
Limited in scope
Limited in precision
Promise of improvement
Paradigm articulation = working out examples the paradigm already provides (novelties not sought).
Normal Science = paradigm articulation
Function: normal science forces scientists to study nature at unprecedented depth generating phenomena that never would have been seen otherwise.
Normal activities:
Fact gathering:
Observations that determine signifcant facts: observations needed by the paradigm to describe reality [e.g., planetary positions in astronomy].
Obsrevations that test the theory: otherwise curious and uninteresting phenomena that provide direct tests of the paradigm's assumptions. [Michelson-Morly interferometer, precession of Mercury, Eoetvos' test].
Observations that articulate the paradigm:
Fundamenetal constant measurement [e.g., Cavendish balance].
Quantitative laws of subsidiary phemonema of relevance [e.g., Boyle's law, Coulomb's law]
Extension of paradigm into new types of phenomena [e.g., explaining ideal gas law using caloric theory]
Theoretical work:
Significant applications [e.g., radio, calendar]. Considered hackwork.
Increase precision of tests [e.g., add friction, exchange point pendulum for extended pendulum]
Theoretical articulation [alternate formulations of mechanics]
Three kinds of normal problems:
determination of significant fact
matching facts with theory
articulation
Extraordinary problems come only out of normal science. Trying to find them just makes you a quack.
Novelty is not a goal: narrow range of expected outcomes.
Goal is increased precision.
Nontriviality: must obtain new illustrations of the paradigm. Failure to obtain agreement counts against the scientist, not the paradigm.
Puzzles:
Outcome not important
Solvability guaranteed in advance.
Test skill of solver.
Failure to obtain a solution counts against solver [like physics lab classes].
Rules for legitimate and illegitimate solution [measurements must be related to the paradigm by theoretical accounts of the apparatus; accounting for lunar motion by altering the gravitational force law didn't count].
Normal science = puzzle solving
General goals of science
useful
exploration of new territory
tsearch for order
challenge to established views
Individual goals of science: puzzle solving. General goals bring you to normal science. Then it's puzzle solving.
Kinds of rules for valid solutions:
Concepts and laws of paradigm may not be altered.
Acceptable instrumentation technologies [e.g., role of fire in chemical analysis].
General heuristic commitments [e.g., Cartesian mechanism].
Metaphysical: what kinds of entities count as explanatory.
Methodological: laws may only involve explanatory entities.
General (extra-paradigmatic) rules of science [e.g., to reduce disorder to order]
Indeterminacy of rules: paradigms do not uniquely determine how to proceed to solve a puzzle.