A 8-year effort finds that much cancer research can’t be replicated | Hacker News Reader

A 8-year effort finds that much cancer research can’t be replicated

sciencenews.org

33 pointspessimizer6 comments

I am probably going to get downvoted for being cynical, but why there is so many hacks among scientific community and they are not being excluded? How do you know when you donate money to a cancer charity whether it is a scam or they are actually doing sensible research? I am grateful that there are people who do such check ups.

It's not hacks. It's the stupid academic tenureship system and monetary incentives.

Graduate students need to publish papers to advance in their careers. Postdoctoral students need to publish papers to be eligible for a tenure track position. And tenured faculty need to publish to obtain grants and keep their jobs. It's called publish or perish.

Replicable research needs more time and money to conduct. But what researcher wants to publish once every few years?

Since there is no centralized authority for scientific research to vet papers, it's up to private, for profit publishers to publish the research they deem appropriate.

The publishing industry should be abolished completely. It's a relic from a bygone era. But I'm not sure what should replace it.

Maybe a centralized library of science under the US government. Replace papers by software inspired open source projects where all scientists can contribute, given proper quality controls. Papers and journals are discrete units of knowledge. No one reads journals from the 1940s, not even scientists. There is plenty of knowledge in obscure tomes that may be revolutionary but no one knows exists. So, replace discrete units of knowledge by continuous projects.

I think there are too many scientists, honestly.

IMO there are (almost) never too many scientists. Yes there will be junk science. In fact most findings are false or offer nothing new. The few that are true form the basis of the progress that mankind has made.

That aside, I think two more reason for the reproducibility crisis are

(1) the restriction on papers lengths by journals. I remember having to cut so much stuff out of my fist papers just to get it down to the 12 or so pages that I was limited to. When I finally finished I thought to myself "I pity the fool that tries to implement this technique based on such paltry information"

(2) Some researchers will not share all material and methods because they wish to stay ahead of the competition. I know this from experience.

Re 1) this is very odd. Why wouldn't they have short 12 page print and full version if someone is more interested as a download?

I don't really know. Perhaps old habits die hard. Or they want to deal with two versions. Online journals should consider remove this pointless limit and instead advise authors not to make papers longer than necessary.

TLDR:

"Researchers with the Reproducibility Project: Cancer Biology aimed to replicate 193 experiments from 53 top cancer papers published from 2010 to 2012. But only a quarter of those experiments were able to be reproduced....

"What’s more, of the 50 experiments from 23 papers that were reproduced, effect sizes were, on average, 85 percent lower than those reported.... "

Whatever you call this, it's not 'top' science, it's not even science - no matter how much it cost.

Reply on news.ycombinator.com