In most cases, instances when "dialects" are not mutually intelligible someone is artificially designating them as the "same" language for political reasons. In the case of "Arabic" it's all about the unity of disparate polities that have little in common beyond a shared heritage that stopped being "shared" centuries ago. Of course, it cuts both ways, because if you recognize "Arab unity" then you undermine the argument that Palestinians should live in Palestine, because it's hard to say why they couldn't live in, say, Damascus or Baghdad.

Another great example is Chinese. Cantonese sounds radically different from Mandarin and, although they're related, it's more like the way Russian and Polish are related. The only plausible argument is that literary Mandarin and literary Cantonese would look alike (but not if one picks up popular writing - Cantonese uses a bunch of its own characters that are different from the standard Chinese characters).

In response to your point about undermining the fundamentals of the Arabic language, yes, he does a great job of it. He shows how many of the "archaic" and "classic" constructions were the result of a poor understanding of an Aramaic substratum and were only ever really used by people trying to model their language on the Qur'an.

As for the 'Alaq sura, his contention is that the "first sura revealed" is the result of a poor understanding of what it's saying.

Essentially, if you boil it all down, he says "The Arabs weren't literate and the Qur'an is the result of borrowing Aramaic words and phrases from the high literary culture of the region and applying it to Arabic to sound more 'proper', and after two centuries the Arabs had so distorted the original readings as a result of no pointing that they had to mark it all to avoid drastically variant readings. This led to the creation of a grammar that had never existed as they struggled to understand their own holy book."