What is the difference between Elsevier's 'waterfall system' and 'cascading peer review'?
I managed to get some more information on this from William Gunn (@mrgunn), Director of Scholarly Communications at Elsevier. So not an official legal statement from a patent lawyer, as he notes himself, but a plain-English explanation from someone familiar with publishing and who has had a chance to ask the people involved, which is pretty much everything I could have wanted. Short Twitter-transcript follows (stripping out Twitter tags; otherwise as written); I'll probably try to come up with a better way to format this later.
@MrGunn: "The Elsevier legal team says this just covers our implementation & is defensive." (https://twitter.com/mrgunn/status/777989863748448256)
@MrGunn: "IOW [in other words], not to be used to sue anyone else, just to prevent us from getting ripped off" (https://twitter.com/mrgunn/status/777990251667087360)
Me: "Thanks, that's helpful. So what makes system distinct? How do you tell if ripped off?" (https://twitter.com/arboviral/status/778146459044745217)
@MrGunn: "(making room) It's common for software companies to patent their implementations of a more general process. That's all this is." (https://twitter.com/mrgunn/status/778269748236685312)
Me: "thanks - that helps. So it would only be enforced if someone literally copied the Elsevier code and tried to pass it off as theirs?" (https://twitter.com/arboviral/status/778497951232618496)
Me: "Well, not necessarily line-for-line, but if you had reasonable suspicion that they'd used the Elsevier system as a starting point?" (https://twitter.com/arboviral/status/778498336156418048)
@MrGunn: "Enforceability of software patents is not a subject I know very much about, but I would guess so." (https://twitter.com/mrgunn/status/778606526160175104)
So it's not intended to cover the general process of cascade reviewing, but only the specific implementation used by Elsevier.