Early integration of expectations into macroeconomics offered us a Platonic Ideal. But as we all know, there are no Platonic Ideals. They're nice stories we make up in our head, of course, but they're not real.
So what do we do? Storm off in a heterodox huff and start macro from scratch? No, of course not. Things like Ricardian equivalence made it into the journals many decades ago for a reason. The reason is that while Ricardian equivalence is not real, expectations certainly are real. The "New Keynesian"* approach of adding various frictions and imperfections is often ridiculed as adding epicycles, but I think that's the wrong way to look at it. The right way is to say that Ricardian equivalence (and things like it) gave us the stark case of what perfect, crisp, clean expectations did to our models. We all obviously took that first approximation seriously - we had no choice, it was a critical result! But we also all knew it wasn't real. The frictions and the imperfect expectations or the heuristic use of expectations and all these modifications is more an example of taking a first approximation insight and applying it to the real world, rather than adding epicycles to an unworkable model.
* We really should get out of the habit of thinking about this strictly as a New Keynesian thing. As Old Keynesian sympathizers like to point out, it's a very Classical way of doing things, and it's something Real Business Cycle theorists and others have adopted as well.
The violinist analogy improved
9 hours ago