Innocence is not enough to get a conviction overturned but it also isn’t necessary to get a conviction overturned. Procedural flaws are far and away the most common reason. So the overturned rate, in capital cases or elsewhere, doesn’t tell us much about the true facts of the matter—-not even as a floor.

Edit: I realized writing this response that I forgot to link my citation 3 in the post you replied to. I also wasn't clear what it was measuring. But to clarify that number is exonerations only, which requires innocence. A legal finding of significant procedural flaws might lead to a reduction to life in prison or a new trial and wouldn't be counted as an exoneration.

Getting a verdict overturned based on procedural issues isn't as easy as popular culture makes out. You have to show both that your rights were violated and that no reasonable jury could have possibly found you guilty if your rights weren't violated. Simply having signed statements from your jurors that they wouldn't have found you guilty if they knew what the prosecutors covered up isn't sufficient, for example.

It depends on the flaw. For example, if an appeals court finds that a search was no good and tosses evidence as the fruit of the poisonous tree that’s almost certainly going to lead to a tossed conviction. But it doesn’t mean factual innocence, on the contrary.

But that wouldn't be an exoneration. The exoneration rate for criminal defendants sentenced to death is around 4%, see my citation 3 in the original post.

(And it wouldn't "almost certainly" lead to a tossed conviction. The defendant would have a chance but not a guarantee at getting a new trial. See the comment you replied to for details)

The study you linked defers to The Death Penalty Information Center’s determination of exoneration.

It’s definition is here: https://deathpenaltyinfo.org/stories/criteria-for-inclusion-...

It includes acquittal or dismissal of charges after the original conviction was tossed. But that does not imply actual innocence. For example, if a key piece of evidence is excluded because of an unconstitutional search that could lead to acquittal or dismissal of charges.

I was using Wikipedia and the numbers were slightly different. Maybe they were older.

The innocence thing isn't just death row. There are groups who have helped people who are not on death row, mostly using DNA. The majority of the people they help are not on death row. Sure, it could be higher. The point was that there is already what I would consider to be a high wrongful conviction rate. In my own experience, I think the wrongful conviction rate for non-custody offenses is multiple times higher.

My impression is that it's higher than those numbers too.

What I meant to say is that death penalty eligible defendants get better representation that pretty much everyone else. So their wrongful conviction rate should be lower than average.