I think this procedural aspect is critically important. Patent litigation is incredibly expensive. Defendants generally expect to spend >4 million dollars to defend against allegations of infringement. Post-Alice, judges have been ruling on the issues very early in litigation and cutting of those costs and timelines. As Theaeteus mentions, the problem with this approach is that it is generally done without full development of evidence. As a policy-matter, it remains unclear which approach is better (we don't know how good judges are at making the 101 determination (knowing it when they see it)).
They've been great in the two major 101 cases, Bilski and Alice. Hardly anyone thought those claims were (or should have been) valid, as the methods were old and well known... But that's also like going back to the old obscenity cases and having your "test case" be Two Girls, One Cup. The old saying "bad facts make bad law" is true primarily because everyone agrees with the outcome based on those bad facts... but then it's used as precedent in cases where the facts aren't nearly so bad.
As an aside, patents have a historic tie to property law, hence things like permanent injunctions and willful infringement damages... but maybe that's not such a good thing, as they're primarily economic instruments. It leads naturally to equity-based judgements rather than concrete tests and rules... people would be outraged if judges were routinely invalidating contracts as a matter of law on a "I know it when I see it" justification.
If we did away with the property tie and abandoned permanent injunctions and what are effectively punitive damages, in exchange for compulsory licensing and reasonable royalties, a lot of the issues people have with trolls would go away, too.