The NAS report on sexual harassment is a deeply-researched guide for improving the climate in academia. If you’re in any sort of leadership role in STEM, I highly recommend reading it. I’m not going to try to summarize it – but I’ll highlight some key/controversial passages below
Sexual harassment isn’t always direct (“Person A” says something awful to “Person B”). Like second-hand smoke, sexual harassment can be “ambient”, creating a toxic situation for others in the environment. 
A ton of attention has been paid to high-profile faculty harassers like Geoff Marcy and Thomas Jessell. But supervisors aren’t the most common perpetrators – peer/co-worker harassment is much more likely to occur.
The more male-dominated a group is, the more likely it is that harassment will occur. (As my previous research showed that elite/Nobel-labs are male-dominated, this might suggest that these labs are particularly prone to harassment…)
There are **a lot** of supervisor-supervisee relationships in academia. Even if they’re fully consensual, they can still be a form of harassment, if they create an environment in which there’s favoritism or in which younger women are seen as “prey”
(Somewhat surprisingly to me) the NAS report comes out against “zero tolerance” policies for sexual harassment. It suggests that inflexible policies are a disincentive for victims to report harassment and take away the victim’s agency.
Additionally, the report links “zero tolerance” policies with potentially harmful “sex panics”. Previous sex panics in the workplace led to significant discrimination against queer/HIV+ individuals.
I had no idea – but the NAS report concludes that there isn’t any credible evidence demonstrating a link between sexual harassment training and decreasing the incidence of harassment. Other strategies are needed.
One area in which the NAS report falls short, in my view, is its treatment of harassment against men. In the statistics that the report cites, a significant fraction of men report being victims of harassment (in science, about 1 for every 1.5 women):
The report notes that most harassment against men is also perpetrated by men – but the report largely fails to discuss what ways (if any) it differs from harassment against women, how it contributes to hostile environments, how it affects careers, how it can be prevented, etc.
How do you make an academic climate better? Like so much in science, it comes down to $$. British universities started adopting major new changes when funding decisions were tied to goals set by the Athena-SWAN equity initiative (cc @NIHDirector).
In total, it’s an incredibly powerful document, and I really think that people in leadership roles should read it. You can download the PDF here: nap.edu/catalog/24994/…
Oh, and on page 160, there's a 3-page long subtweet on the role of professional societies that seems to be directed at the NAS leadership and their refusal to expel convicted harassers.
The report also suggests some pretty fundamental changes to academic structures: multiple advisors per advisee, financial support from departments rather than labs, and in general a diffusion of power away from the strict hierarchies that are currently common.
