It's the interpretation that was a disaster.  These guys came nowhere near showing that TV was a cause of autism.  They're not medically trained, and not epidemiologists, so they had no idea how to interpret their findings or where the pitfalls were.  They were so excited about the data they found that they jumped straight to overinterpretation.  The journal editors should never have permitted that.

Statistically, though, I believe it was considered a pretty good study (though of a type that is inherently weak).  And while it doesn't show causation, correlation is nothing to sneeze at.  Nearly all causes start as correlations.  So I certainly hope someone with a better understanding of the issue is following up on the observation.