TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Robot scientist finds less than one third of scientific results are reproducible

39 pointsby takiwatangaabout 3 years ago

7 comments

readthenotes1about 3 years ago
Something doesn&#x27;t add up:<p>&quot; ...attempted to reproduce the 74 results. Statistically significant evidence for repeatability was found for 43 papers, meaning that the results were replicable under identical conditions; and significant evidence for reproducibility or robustness was found in 22 papers, meaning the results were replicable by different scientists under similar conditions. In two cases, the automation made serendipitous discoveries.<p>While only 22 out of 74 papers were found to be reproducible in this experiment,... &quot;<p>Shouldn&#x27;t that be 9 were NOT reproducible since 65 we&#x27;re reproducible?<p>Or does &quot;results were replicable&quot; mean something other than reproducible?
评论 #30949641 未加载
spicybrightabout 3 years ago
Their sample size was only 74 papers from one field of study.<p>It&#x27;s easy to get onto the &quot;science isn&#x27;t working&quot; bandwagon, but how about some more convincing evidence to prove it&#x27;s happening as widespread as the claims they&#x27;re making are.<p>You know... maybe treating it a bit more scientifically to prove your point.
评论 #30954489 未加载
评论 #30954358 未加载
poulpy123about 3 years ago
But is this paper reproducible and verifiable ?
评论 #30949327 未加载
meremortalsabout 3 years ago
Balaji Srinivasan recently discussed the importance of scientific reproducibility on The Knowledge Project: <a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=NlY8HICFiRs" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=NlY8HICFiRs</a>
longivitateabout 3 years ago
It is long past time when the majority of paper results could be written in machine verifiable language.
评论 #30947678 未加载
smegsicleabout 3 years ago
&quot;trust the academia&quot;
评论 #30948472 未加载
评论 #30947996 未加载
评论 #30947692 未加载
评论 #30950133 未加载
johndunneabout 3 years ago
They should have used a smarter robot.