a) doing nothing at all for his autism,
b) doing whatever is current standard practice for treating autism,
or
c) doing something that is bizarre and zany for autism.
She sees some behaviors change for the better, behaviors that might have changed for the better on their own. She has no way to sample the tissue of her child's brain to know what is happening to his brain. On the basis of her self-report, she doesn't appear to be doing him active short-term harm, but as a parent I try to validate what I do for my children with the observations of many other observers. She would be able to help her child with more likelihood of success if placebo-controlled, double-blind studies compare all the possible treatment approaches out there.
If there are several main types of strains, then well, I suppose your study needs to include more people -- not that hard. If strains change over time, that seems like another area for more study. This is what studies are for, and I cannot figure out why anyone would be opposed to it, or say it's not feasible.
Actually maybe you are onto something there.
The author isn't proposing giving marijuana to all childhood autistics. She's just saying that it helps her son.
Besides, doing placebo-controlled, double-blind studies on whether marijuana is an effective medicine is like doing placebo-controlled, double-blind studies to figure out which is the best girlfriend. There are dozens of strains to choose from, and there's no reason to believe that a strain which is good for one person would be good for someone else, or even the best strain today would be ideal for tomorrow.