Has Harmless Error Become Too Broad in Social Security Cases?

I've been sitting with a recent Seventh Circuit decision for a while now, and I can't shake it.

The court spends page after page explaining why the ALJ got it wrong. Not just one mistake—several. The opinion walks through how the ALJ reversed the regulatory framework for evaluating medical opinions, failed to properly address supportability and consistency, applied inconsistent reasoning across the decision, failed to build a logical bridge between the evidence and the conclusion, and selectively interpreted the record. The court repeatedly calls the analysis erroneous.

Then it affirms anyway.

Why? Harmless error.

That result has me asking a question I don't think has an easy answer: has harmless error become too broad in Social Security cases?

The Doctrine Serves a Real Purpose

Harmless error isn't the villain here, and I want to be clear about that up front. Not every mistake in an ALJ decision justifies a remand. Courts shouldn't force agencies to repeat proceedings when the outcome genuinely could not change regardless of the error. That's not leniency—it's efficiency serving claimants and the system alike. Nobody benefits from a remand that exists only to reach the same result through a cleaner path.

But there's a point where the doctrine starts raising a harder question, and this case sits right on that line.

Whose Reasoning Are We Actually Reviewing?

If an appellate court identifies multiple legal errors, reconstructs what the ALJ should have done instead, and then predicts with "great confidence" that the same result would follow anyway—are we still reviewing the ALJ's decision? Or are we substituting judicial certainty for administrative reasoning?

This isn't a semantic distinction. Social Security law has long emphasized that reviewing courts evaluate the reasoning the ALJ actually provided, not post hoc rationalizations supplied later by the court or the agency's counsel. We remind ALJs every day that they must build a logical bridge from the evidence to their conclusions. That requirement isn't decorative. It's the mechanism that lets a reviewing court trace how an ALJ got from the record to the RFC, and it's what makes meaningful review possible in the first place.

Opinions like this one leave me wondering whether that bridge requirement is slowly giving way to something else entirely: yes, the bridge collapsed, but we're confident you would have reached the other side anyway.

That's a different standard than the one we tell ALJs—and claimants—governs these cases.

Where's the Line?

I don't think the answer is "harmless error should never apply when there are multiple errors." Sometimes several small, genuinely inconsequential errors really do add up to nothing. But when the errors go to the core of the supportability and consistency analysis—the framework that's supposed to determine how much weight a medical opinion gets—the reconstruction starts to look less like harmless error review and more like the court doing the ALJ's job for them, after the fact, with the benefit of hindsight the ALJ never had.