10 years ago, you could get by doing terrible numerical models and make rookie statistical errors in understanding data and only 400 people would read it, while even fewer would understand it. Not any more, everyone can now parse data on the Internet so your clever analysis of fMRI images or convenient massaging of results to match your end-oriented beliefs are going to get tripped up pretty quickly.
And legacy science academia is soon going to be in trouble because scientists are getting good at handling big data. While young researchers are told they should dutifully wait in line until their 40s, when maybe they will get the golden ring of tenure, the corporate world will happily find a place for them right now, if they have the proper skills, and hand over a 6-figure salary along with it.
At Pythonic Perambulations, Jake Vanderplas invokes Eugene Wigner's famous "The Unreasonable Effectiveness of Mathematics in the Natural Sciences", and for good reason. The rest of science is now facing the same issues that have faced physics for a while. And to keep the best talent from abandoning the ivory tower, he has some recommendations because, let's face it, the modern research skills that are prized by the corporate sector are not even recognized by academia as valuable (yet).
Some of them are non-starters. A minimum pay for post-docs, for example, is not going to happen when we are producing 6X as many PhDs each year as there are jobs in academia. Establishing a minimum pay means that 3 post-docs will be employed rather than 4 - these studies are funded by government grants. Just the US government alone is already spending $600 billion more per year than it takes in, so asking for a 30% boost in science funding isn't realistic. We allowed unlimited student loans to balloon the pay of tenure and faculty and to finance a whole lot of new buildings, so those costs are not going down - that means post-doc salaries are not going up.
The others make a lot of sense, if keeping the smartest researchers in college is important.
Give it a read: The Big Data Brain Drain: Why Science is in Trouble by Jake Vandenplas