Fallacy?

Ad hoc rescue

When a prediction fails, an extra assumption is added whose only job is to save the theory.

Also called: Moving the goalposts · Immunising stratagem

The test

Does the added assumption have support of its own, or was it brought in only to explain away the failure?

Every theory meets anomalies, and adjusting a theory to meet them is normal science. Karl Popper warned that some adjustments make a theory immune to testing; Imre Lakatos drew the useful line. An adjustment is progressive when it predicts something new that is then found. It is degenerating when it only accommodates what has already been seen.

The processes named in the examples are real. Convergence, gene transfer and lineage sorting have all been demonstrated. The question for each case is whether they are tested, or simply invoked.

Seen on both sides

In mainstream arguments
It would be an ad hoc rescue to explain each conflict between gene trees by convergence, gene transfer or incomplete lineage sorting, case by case, without testing that explanation in each case.
In critical arguments
It would be an ad hoc rescue to explain away each conflicting observation by an unspecified effect of the Flood, without independent evidence for that effect.
Not a fallacy when

The added assumption makes new predictions that are then confirmed, as when Neptune was predicted to explain the orbit of Uranus.

Diagnosed on this site

  • Pillar 2 · The Fossil RecordPossible
    In the pillar as a whole

    Gaps, abrupt appearances and stasis are each given an explanation after the fact. Whether those explanations made predictions of their own is the question each card asks.

  • Claim № 3.01 · HomologyPossible
    In the claim as taught

    Similarities that don't fit the expected tree are reclassified as convergence. That is legitimate when convergence is identified by independent criteria; if not, every outcome fits.

  • Claim № 3.02 · The tree of lifePossible
    In the claim as taught

    Conflicts between trees are explained by gene transfer, incomplete lineage sorting or methodological artefacts. These processes are real; the question is whether each is tested independently in each case, or simply invoked.

  • Claim № 2.02 · Cambrian explosionPossible
    In the claim as taught

    Poor preservation is offered to explain the missing ancestors. That is legitimate if it has independent support. Precambrian rocks that preserve embryos and single cells make the explanation harder to sustain, and it is rarely tested formation by formation.

  • Claim № 2.03 · StasisPossible
    In the claim as taught

    Punctuated equilibria were proposed to explain a pattern gradualism did not predict, by placing change where it would rarely be fossilised. Whether this is ad hoc depends on whether the model made new predictions that were confirmed, which its defenders argue it did.

  • Claim № 2.03 · StasisPossible
    In the alternative reading

    “Created kinds with built-in capacity for variation” explains stasis well, but a kind needs a definition that could be checked against the record. Without one, it too could absorb any pattern.

Further reading

  1. Popper KR. Conjectures and Refutations. Routledge (1963).
  2. Lakatos I. “Falsification and the methodology of scientific research programmes.” In Criticism and the Growth of Knowledge, Cambridge University Press (1970).