"The few criminologists who have thus far investigated the phenomenon estimate that the overall rate for convicted felons as a whole is between 2 percent and 8 percent."
I'm sorry but you are quoting an article that doesn't give a proper citation for this at all, so please don't rely on these probably woefully underestimated numbers. I'd be happy to have it pointed out to me if I missed their source though, but for now;
What few criminologists? How do they know there are only a few? Did they do a survey or what papers are they basing this on?
This lack of sourcing undermines the rest of your comment, not to mention you imply that the jury trial standard is "90% or lower" when you don't even specify that what you link is related to tax crimes only...
"Moreover, it's unclear to me that you can do much better than 90-95% accuracy."
Again, you basically smushed these numbers in to fit your narrative, and they are not reliable at all, so your entire argument falls flat.
" What would be your rate of false negatives (i.e. where you let a guilty person go)? And how would society react to that?"
In a society that values the principles of law and the rule of law, it is better to let a guilty man go than an innocent one, given that they should be innocent until proven guilty. Unfortunately our society seems less and less inclined to care about the rule of law, so I'm guessing "society" wouldn't give a fuck about a few more guilty people being let free, just like they don't really seem to give a fuck about innocent people being jailed.
Sorry if I'm being to negative, I just really don't understand where you were trying to go with your comment and it's bad statistics. Feel free to elaborate and correct me.