The whole point of a 15,000-person survey like this is so that we're not all stuck guessing based on our own experiences. Obviously one can argue that what employers were telling employees they would do post-COVID as of October 2020 is not what employers will actually do (and give reasons why!), but it's still likely to be more accurate than any individual's speculation (including mine!).
In particular, the survey addressed 2 topics that a lot of comments seem to not incorporate:
1. At population, employer, and individual level, it's not either/or. The paper title doesn't do a good job of conveying this. All 3 entities seem open to everything on the continuum, based on their employees, work type, and so on.
Statements like "I promise that it won't stick for the broader society" imply a more binary outcome than the survey results predict (and maybe than anything predicts). The survey concludes that, of the roughly 50% of employees who can work from home, they'll average 2 days per week (or 23% of total hours worked). IOW, WFH doesn't need to move everyone all the time in order to have massive effects. Some people changing some of the time is enough.
2. Employer plans vs. employee desires. Lots of commenters state their preferences and then try to generalize from that, but.. the survey captured that too. To just skim those results, scroll to page 60 for employee-desired vs. employer-planned % of working days. Gist: if WFH is roughly the same productivity, employers plan ~22% of days and employees desire ~45% of days to be WFH. In all productivity cases, employees preferred a higher % of days to be WFH than employers.
Page 43 shows answers to the pay raise or cut employees would take in order to work from home 2-3 days per week.
It also breaks down the change in perception about WFH by demographic cohort on page 55. For example, those with no children and those with children at home both had significant improvements in their perception of WFH. Those with children had more improvement in perception than those without children. As the study wrote: "This 23 percent figure [post-COVID WFH as a % of full time days] is almost five times larger than in the pre-pandemic time use data, but still half as large as what workers want in a post-pandemic world."
Of course, there's a place for speculation and anecdotes. This paper - and hopefully soon, more studies like it - should let us evolve from speculating based on only our personal experience to thinking about its methodology and results.