ThoughtStorms Wiki

Context: OnScience

This page for people who want to improve or move beyond current scientific practice and institutions. And trying to fix ProblemsWithScience



Quora Answer : If science were to transform into something better, what do you think it would be like?

Mar 21, 2019

It would be quite like science is today but

a) academic journals would be non-profits, funded by contributions from, or a tax on, universities. Plus maybe a tax on patents. All serious publications would be "open access" ie. free for anyone to read online. This would NOT involve a kind of "vanity publishing" of dodgy journals with low standards charging people to be published.

b) academia would get its online story sorted. All lectures would be online and freely available to all. All universities would use MOOCs.

c) universities would grok social media. Today, social media is the enemy of serious academic knowledge. And it's starting to corrode it. That's largely academia's fault for not engaging and learning to use social media better. Instead academia tried to stand aloof from social media, and social media is full of nonsense. Furthermore, social media is more successful at selling its nonsense to the general public than academia is at disseminating difficult truths.

d) But in a better world, we'd have online forums where members of the public could debate science with the professionals. But there'd be rigorous editors and high standards for the "publications" that the members of the public could contribute. Academia would have taught its culture of rigour to the public that wanted to engage it. And the public would have embraced that rigour.

e) we'd have much better management of data and resources for storing, manipulating, analysing it. This has been way too sloppy in the past. Today, people in the software industry can set up a Kubernetes pipeline capable of creating virtual machines in the cloud for testing / continuous integration etc. from a simple script. Academia should use the same tools. If I read a paper that depends on a data-set and statistical analysis of that data, then a necessary prequalification for that paper being accepted and getting the "stamp of approval" from the journals / academy is that I can run a single script to grab the entire data-set and recreate the entire analysis framework in containers on a cloud somewhere.

f) we need more tools to help with the statistical analysis, data visualization and explaining what we're actually learning.

g) science is going to get invaded by AIs pretty soon. By which I mean we're going to have AI / machine-learning techniques looking at data, building models and making predictions. And no human will understand the details of or even the rationale behind those models. It will be very easy to dismiss them. And very hard to justify why we should trust them. Except for the fact that their predictive capacity works. Somehow we need to prepare the scientific community and the general public for this this situation.

See also :