Recognizing "stolen" content autonomously can only pivot on knowing when something was first published or visible to Google, which is a pretty dubious measurement.
But the weird thing is the stolen content, even when it's on a crappier site with no shares etc., knocking the original out of its slot. I.e., something (probably freshness) is causing the stolen content on the crappy site to be higher-ranked than the original content on the strong site. That seems avoidable (and in Google's best interest).