Hacker News new | ask | show | jobs
by biophysboy 381 days ago
Why do you think reproducibility will improve if funding is cut significantly? That is less money for labor, instruments, and consumables, which means less money for experiments, particularly for follow up experiments that have no deliverable.
2 comments

I agree that cutting it won't solve the issue, especially without that long term plan, but at the current sky-high level of fraud, this just isn't science anymore. At what level of fraud do you suggest pulling the plug if not right now? Funding research until 90% is fake? 95%? 99%? Taxpayer money can be spent on many other life-critical areas with tangible results.
Irreproducibility is not the same as fraud. In any case, I would support high-risk research - stuff that may not be repeatable - if it has a high direct ROI, economic spillover (USA has a big biotech/pharma ecosystem), and leads to cost savings (e.g. no more dialysis machines).
Yes, exactly. Defund the junk science, the circles that are put around the results and called targets, and all the questionable research activity until merit has returned to scientific research and the vast majority of empirical studies can be replicated without p-hacking it.
Ehh.. I think you're overestimating how easy it will be to make science repeatable. Even if we perfectly execute experimental/statistical methods, biological experiments are not always going to be reproducible.

Edit: as a follow up (because I feel like I strawmanned you), I am just trying to say that p-hacking is not always malice or incompetence. Sometimes limited methods/theory miss confounding variables, sometimes labs lack resources, etc.

I just don't even buy this, medicine is definitely advancing. And, like, very rapidly. We have highly effective and safe drugs available today that were unthinkable 10 years ago.
> Taxpayer money can be spent on many other life-critical areas with tangible results.

... but it won't be.

Competition is higher people are hungrier, fraud will go up not down
Yes, exactly