Dawk is presenting this as a logical syllogism (in response to criticism; it wasn't how he first presented it). It isn't actually a classical syllogism, which is the 'all men are mortal' type of deductive reasoning. Really, he's just refuting an implication that some people inferred from a statement of comparison. It's unfortunate that the newspapers that get their content by summarising twitter feeds (I'm looking at you, Independent) couldn't be bothered to find this out, but no matter. While syllogisms and other logical arguments are relevant to linguistics (and I have fun teaching them in my semantics module), more important here is the fact that Dawkins is just banging on and on about logical truth, never seeing that the logical truth of his statement really isn't the point. Nobody seems to have put this in a way that he can understand, so it's not entirely his fault, but still: rape, paedophilia and the like are highly emotive topics and there is currently a lot of discussion about 'rape culture'. Given this, it's not surprising that many people would be less concerned with logical truth and more with the rhetorical effect of such public statements. Many people are extremely worried that some instances of rape are trivialised or simply discounted because the victim was drunk/married to the rapist/didn't say no etc. Saying that date rape is less bad than violent knifepoint rape may be logically true (or it may not be - Dawkins is agnostic on the matter), but humans do a lot more than just compute logical truths. We actually have to work quite hard to see purely logical truth (hence the zillions of logical fallacies that we can make), and we set a lot of store in the inferences we make. Even though saying that X is not as bad as Y doesn't condone X, it still appears to make excuses for those people who dismiss the 'date rape' cases as 'not really rape'.
So, on to 'so'. In the tweet pictured above, someone passive-aggressively tweets about Dawkins, not to him, while @ing him so that he sees it. This is the height of bad twitter manners. Dawkins, in his response, takes exception to Sequester Zone's use of 'so', asking why they used it. I assume that Dawkins is making bizarre linguistic assumptions again, and would hazard a guess that he considers 'so' to be incorrect when it is used as an introductory particle, similarly to 'and' or 'but'. I've seen some other peeving about this lately, with people claiming that 'everyone is starting their sentences with so these days'. As it happens, I agree with much of what it says in the article linked via the response above, but 'so' doesn't always indicate a rehearsed pitch or dumbing-down. It's long been used as a turn-beginning marker, or as a way to indicate that you're returning to a previous topic, or many other things. The OED's got an example from 1602 in their sense 5c, where it's a kind of 'hey I'm talking' marker:
So, let me see, my apron.And in sense 10b(a) from 1710 (Swift, no less), where there is no preceding statement but one is implied:
So you have got into Presto's lodgings; very fine, truly!And 10b(b), which it attributes to 'reflecting Yiddish idioms', where there is no preceding statement, or where there is adversative force (probably the use in the tweet) from the 1950s (this example is 1960):
‘I warn you..I ain't got no wine.’ ‘So who wants wine?’Note, though, that it serves another function within twitter. In a tweet, if you begin with an @-name, it will only be seen by people that follow both parties. The convention, therefore, if your tweet begins with a name but is a mention rather than directed at that person, is to begin with a full stop so that all your followers will see it. Alternatively, you could make sure that the @-name isn't at the beginning of the tweet by using an introductory particle such as 'so'.