Read on Twitter
Some years ago now I wrote a review & the editor changed it, in substantive ways, without my knowledge or permission. When I discovered this, I questioned him about it and he attributed the changed text to a technical glitch in the submissions system 1/ https://www.sciencemag.org/news/2020/10/delete-offensive-language-change-recommendations-some-editors-say-its-ok-alter-peer
the explanation I received: "on very rare occasions when the system is undergoing maintenance and several editors are simultaneously processing papers, the links between individual reviewer documents submitted to the journal and forwarded to an editor can be muddled"
could such a thing ever occur?
I replied "It seems to me that a technical glitch of this nature--especially one that results in grammatically perfect and consistent end products!--requires serious and immediate attention... Have the publishers been notified?”
The publishers were notified. By me.
If you're wondering about the headline of the article, let me assure you that this case had nothing to do with offensive language being removed. The changes looked like this 
And deleting qualifying sentences like this "To be clear: I think the actual question asked to participants is an interesting and worthwhile question—I just wouldn’t present it as a question about ‘certainty’."
The review I wrote was positive & the recommendation I chose was 'accept pending minor revisions". Yet the editor wrote to the author: "the reviewers have spoken in nearly a single voice in their recommendation to me that I decline publication of the paper in its current form’
I still don't understand why the editor changed and misrepresented my review. I expect such extreme cases are rare. But the 8% in the bottom line of @danielghamilton's survey gives us reason to think it might not be a completely isolated experience
