Interesting to read the David Crystal post. I hadn't had time to comment on this thread over the last few days; but I'm a native speaker who has no problem with "since... ago": to me, it's perfectly good English. I then checked it with three BE-speaking family members, who all unanimously and immediately said the following sentence was wrong: "I've known him since ten years ago."
One thing I learned on my linguistics degree was that there are often biasing contexts that will make an utterance feel very natural, even if it feels wrong in isolation. 2006 provided a good example of that for "since... ago". The same thing can actually happen with all sorts of utterances a grammar book would call incorrect.
I just want to make a couple quick points from a syntactician's perspective:
1) If you ever do any thorough study into native speaker's perceptions of the grammaticality of certain utterances, you'll find that in many areas, a group of people will be divided in their judgements. Though it doesn't become clear very often, most native speakers actually have slightly different grammars in their heads - and none of them is wrong.
2) Someone wrote that they'd be surprised to see an educated native speaker write "since... ago". Comparing writing and speaking, style is different; grammar is not. Moreover, the English produced by even the most educated native speaker is no more grammatical than that of a completely uneducated individual: both naturally produce completely grammatical utterances, and both sometimes slip up and produce ungrammatical ones.
The grammar native speakers learn in school is not real grammar, though a better education may help you to communicate more effectively and in the appropriate style when writing. Even the grammar we EFL teachers instruct our students to use is only our best imperfect attempt to describe the grammar that exists naturally in our heads.