Ya, an increased risk of 5-20% on an already very low risk.

That means mothers who don't take Tylenol have baseline 3% chance their child will be diagnosed with autism. And mothers who took Tylenol (at the levels of the study) may have a 3.15% to 3.6% chance (assuming causation, which has not been proven).

It seems unlikely we "cracked the code" here.

The best justification for the high increase we're seeing in the data is still just that the data itself has changed in how it's measured and tallied and so on.