A view can become the settled position of a field before the evidence for it is strong. The mechanism is social rather than dishonest, and it is fairly well understood.
Citation creates the appearance of weight
An early paper makes a tentative claim. Later work cites it as background, and the hedges do not travel with the citation.
After several rounds the claim appears in many places, each pointing back to the same original source. The apparent breadth of support is an artefact of repetition.
Anyone auditing the chain finds one study rather than many, but auditing chains is laborious and rarely rewarded.
Dissenting early is expensive
The first person to challenge an emerging consensus bears the full cost of being wrong, while sharing the credit for being right with everyone who follows afterwards.
That asymmetry means doubts accumulate privately for some time before anyone publishes them. The public record consequently understates how much disagreement exists inside a field at any moment.
It also means reversals arrive suddenly rather than gradually. Once one credible dissent is published, the private doubts become publishable, and a position can shift within a year.
Policy demand outruns research supply
Decisions have deadlines and research does not. When officials need an answer, they take the best available current view, which then acquires official standing.
Once a position is embedded in guidance or funding decisions, revising it carries institutional costs beyond the scientific ones, and inertia grows.
This is why some conclusions persist in policy documents for years after the underlying literature has moved on.
Replication is undersupplied
Confirming an existing finding is less publishable than producing a new one, so the checking work that would catch errors early is systematically undersupplied and underfunded.
Fields that have invested deliberately in replication have found substantial portions of their accepted results harder to reproduce than anyone expected before the exercise began.
The problem is structural rather than a matter of individual integrity. Careers are built on novelty, and no funder is rewarded for confirming what everyone already believes.
What a careful reader can do
Tracing a claim to its earliest source is the single most useful check available, and it often reveals a smaller and more qualified original finding.
Asking whether a result has been reproduced by an independent group separates durable findings from provisional ones more reliably than counting citations.
None of this argues for dismissing expert consensus, which is usually the best available guide. It argues for knowing how firm a particular consensus is.