"The best thing to do when you see science you don't like, is pretend it reinfor...

dash2 · on Oct 4, 2019

This totally makes sense. If the paper shows that A correlates with B, and interprets this as A causes B, but doesn't show it; and if you think B causes A; then this is a great explanation for their result.

sbierwagen · on Oct 5, 2019

There is a big big big gap between an individual newsworthy article, and science tested by multiple replications.

Only 62% of articles published in Nature and Science actually replicated: https://www.nature.com/articles/d41586-018-06075-z About a coin flip if you can believe a given headline at all.

A single article is a data point, not ironclad unimpeachable argument-ending fact.

dekhn · on Oct 5, 2019

I would argue that the DNA paper by W&C was an ironclad unimpeachable argument-ending fact. The nice thing is that you could take their structure, run it forward through the scattering algorithm, and see that the simulated data looked identical to the measured data. Further, the simple prediction the paper made (that duplex DNA could form a template for DNA replication) turned out to be so audaciously true...

(I could quibble and say the W&C structure is technically wrong, because it was done in non-physiological conditions, and subsequent studies did find very minor structural details for B-DNA when done correctly...)

sbierwagen · on Oct 6, 2019

Did I say that no authoritative papers have ever been published?