No you're absolutely right. This is more a holy crap paper from Nasa, this actually might work.

I stand corrected. They appear to have measured forces in tens of micronewtons while the error is in single digit micronewtons. Still the confidence interval is an order of magnitude smaller than the measured force which is not bad.

The paper is actually reporting measurement of millinewtons with 40/60/80 watts, and estimating that when pushed up to a 1kW it would generate micronewtons.