Findings that hold whoever tests them cannot be reduced to the tester's viewpoint

The argument

A second argument turns on what scientific results do once they leave the people who produced them. Proponents point out that a conclusion reached in one laboratory is routinely confirmed by researchers in other countries, trained in other traditions, holding other political and religious commitments, and often hoping for a different answer. If the content of a finding were set by the standpoint of the person who found it, proponents argue, this convergence would be inexplicable, because the standpoints differ in every respect except the procedure followed. The point is pressed hardest through prediction. A theory is used to say in advance what an instrument will read before anyone has looked, and the reading comes out as stated. Proponents hold that the world, not the community, is what settles such a case, since no amount of shared assumption explains why a figure derived from a theory should match a figure nobody had yet measured. The same applies to the technologies built on these results: devices designed from a theory work for users who know nothing of the theory and would not endorse it if they did. Opponents of this reasoning are taken to be committed to something stronger than they usually admit. If objectivity is impossible, proponents argue, then the agreement of hostile investigators and the success of untested predictions must be either coincidence or convention, and neither account explains why the conventions cannot simply be chosen otherwise. On this view objectivity is not a claim about anyone's state of mind but an observation about results that travel: what survives being tested by people who wanted it to fail is not a reflection of their perspective, since their perspectives disagreed.

Premises

Counter-arguments

Critics argue that agreement and usefulness are weaker evidence than they sound. Investigators who disagree about politics may still be trained into the same methods, instruments and standards of what counts as a reading, so their convergence shows shared formation rather than contact with an unmediated world. Successful prediction, they continue, establishes that a model is reliable within the range it was built for, which is a claim about instruments rather than about truth; models now regarded as wrong also predicted well for a long time. Working technology is said to prove even less, since a device can keep functioning while the account of why it functions is replaced. On that view the position confuses objectivity with practical adequacy.

Rejecting the premises

[Rejecting P1] Investigators who differ in politics or religion are typically trained into the same methods and standards, so their agreement may show common formation rather than independent confirmation. [Rejecting P2] Successful prediction shows a model works within the range it was built for, and models later judged wrong also predicted well, so predictive success does not establish that the account behind it is objectively true. [Rejecting P3] Neither coincidence nor bare convention is required: a community can converge because its shared training makes certain results the only ones it would recognise as results. [Rejecting C] At most the argument shows that scientific findings are robust across investigators, which is compatible with their content still depending on a framework no investigator stepped outside.