My point was that everything they were they were trying to patent was an obvious trick of the trade. They were trying to patent thing along the lines of: generate->filter->rank. It's how basically any recommendation engine works and has been described in numerous existing patents.

But I tried not to be a jerk at all. Every time they asked me about a different aspect of the system, I could just say something like, "We used standard software engineering techniques using an algorithm built in to [common language] to sort things here. It's a fairly common approach to this sort of problem." After a statement like that, the patent lawyers had very little to pull on. They'd move on to another aspect and I'd give a similar answer.