Skip to content
TheChatArchiveThe Chat Archive
gpt-4.1-2025-04-14· OpenAIHugging Face LMArena Human Preference 140K

given a given scientific truth, what are the odds that it will be considered truth,…

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with gpt-4o-2024-11-20’s answer.
  1. User

    given a given scientific truth, what are the odds that it will be considered truth, compared to competing theories considered pseudoscience today, even 100 years from now ?

    Make the best assumption based on historical perspective (how many scientific truths of the past have survived the test of time, and how many have instead been overridden by theories considered pseudoscience at the time ?

  2. Assistant

    This is a fascinating question that touches on the philosophy and history of science. Let’s break it down:

    1. Historical Perspective: Survival of Scientific Truths

    • Most scientific truths survive, but are refined: The majority of well-established scientific theories (e.g., Newtonian mechanics, germ theory, heliocentrism) have survived, though often in a refined or expanded form. For example, Newtonian physics was not "overthrown" by Einstein, but rather shown to be a special case of a broader theory (relativity).
    • Some are overturned: Some theories once considered "truth" have been overturned (e.g., phlogiston theory, luminiferous aether, geocentrism). However, these were often replaced by theories that were initially considered fringe or even pseudoscientific.
    • Pseudoscience rarely becomes science: Theories widely considered pseudoscience (e.g., astrology, homeopathy, perpetual motion machines) have almost never become accepted science, though occasionally, ideas on the fringe (e.g., continental drift) have moved into the mainstream.

    2. Odds Based on Historical Data

    • Survival Rate: Thomas Kuhn, in The Structure of Scientific Revolutions, argued that "paradigm shifts" do occur, but they are rare. Most scientific knowledge accumulates incrementally.
    • Estimates: Some historians of science estimate that about 80-90% of well-established scientific knowledge survives over centuries, though it may be reinterpreted or subsumed into broader frameworks.
    • Pseudoscience to Science: The odds that a theory currently considered pseudoscience will become mainstream science are extremely low—probably less than 1%, based on historical precedent.

    3. Why?

    • Scientific method is self-correcting: Science is designed to weed out falsehoods over time.
    • Pseudoscience lacks evidence: Theories considered pseudoscience today generally lack empirical support or are unfalsifiable.

    4. Examples

    • Survived: Germ theory, atomic theory, evolution by natural selection, heliocentrism.
    • Refined: Newtonian mechanics (refined by relativity and quantum mechanics), classical thermodynamics (refined by statistical mechanics).
    • Overturned: Phlogiston, aether, miasma theory.
    • Pseudoscience that became science: Continental drift (initially dismissed, later accepted as plate tectonics). This is a rare case.

    5. Summary Table

    Theory TypeOdds of Survival/Acceptance in 100 Years
    Well-established science80-90% (will survive, possibly refined)
    Current pseudoscience<1% (will become accepted science)

    6. Caveats

    • Revolutionary science is rare: Most changes are incremental.
    • Some "pseudoscience" is just untested: Occasionally, a fringe idea is vindicated, but this is the exception, not the rule.

    In summary:
    The odds that a current scientific truth will still be considered true (perhaps in a refined form) in 100 years are very high—likely 80-90%. The odds that a theory currently considered pseudoscience will be accepted as scientific truth in 100 years are extremely low—probably less than 1%. History overwhelmingly favors the survival and refinement of established science over the vindication of pseudoscience.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.