I disagree. The problem is that the Aspergers-vs-Autism distinction was largely "looks normal at first glance" vs "doesn't", which is almost entirely unrelated to the many axes of actual symptoms.
An glaringly incorrect division is more harmful than a lack of division. If it had been a "good enough" division it could've been kept.
The best-attested actual axis is "has epileptic seizures?"; partitioning by it gets rid of most of the bimodalities in the data.
My estimate is that there are around 5 individual axes but many of them are continuous and it's not perfectly clear if some are actually correlated to others. And since the others are usually not bimodal it becomes difficult to choose different diagnoses based solely on symptoms, in the absence of knowing and testing actual mechanisms.
An glaringly incorrect division is more harmful than a lack of division. If it had been a "good enough" division it could've been kept.
The best-attested actual axis is "has epileptic seizures?"; partitioning by it gets rid of most of the bimodalities in the data.
My estimate is that there are around 5 individual axes but many of them are continuous and it's not perfectly clear if some are actually correlated to others. And since the others are usually not bimodal it becomes difficult to choose different diagnoses based solely on symptoms, in the absence of knowing and testing actual mechanisms.