Re: Previous approach
There's been (as you can probably imagine) a fair amount of academic interest in whether contextual or behavioural (personalised) advertising with quite a few studies into the effectiveness of each. In controlled environments, contextual advertising consistently outperforms behavioural (just do a quick search on papers and the trend is pretty obvious).
The problem, probably, is that behavioural advertising is something that a lot of 'industry experts' make a lot of money out of. The systems are already in place to support it and it's an industry that is pretty easy to keep new folks out of. As a result it's pretty robust to industry disruption and a fair bet for making lots money.
Contextual advertising is a simpler affair, requiring a lot less data aggregation and so easier to break into. I can see why existing folks in the behavioural market don't want to see a shift. It opens them up to greater risk of competition. And so it's in their best interest to keep shilling behavioural over contextural.