It's a bit speculative. They need placebo-controlled trials and so far the results have not been very convincing for many psychiatric outcomes. There is exciting research around gut microbiota, but many of these studies which make rounds in HN and other online forums, seem to be mostly open-label or uncontrolled studies.

It does not have to be misdiagnosis. If kid has both autism and gut issues, the gut issues could make the autistic symptoms worse, by causing distress to the kid, which could make the interactions with caregivers harder for both in a quite formative period. Treating the gut issues could help this way without gut being directly related to autism and without it being a misdiagnosis. It is telling that they report quite high (0.7) correlations between improvement in gut and autistic symptoms.

However they say they also have an adult trial running that seems to show similar effects, so there might be something more into it.

"High-functioning" and "low-functioning" are not meaningful as permanent diagnoses, because how a person functions is heavily-dependent on their environment (social, physical, and societal), including stimulus and chemical signalling from their own body.

An environment in which one person can thrive, labour, and enjoy life could be boring or incomprehensible or unduly stressful to another. I know people who would be diagnosed "low-functioning" if assessed in an everyday environment, but "high-functioning" if assessed in a clinical environment, and I know people who might not be diagnosed at all in an everyday environment but would be diagnosed with several seconds of acronyms in a clinical environment; and I know people who've been able to fight to get themselves an everyday environment that works for them, and I know people with vast potential who have conceded that fight and are rotting in the social care system.

If you eliminate what "should not be a factor" from the diagnostic process, then you eliminate the high-functioning / low-functioning distinction entirely. But all models are wrong, and some models are useful: while I find this particular model distasteful, there are contexts where it is necessary to get people in an environment where they can thrive (which, for children, usually means getting their parents the support they need, and occasionally the education).