Bookends still strongly has my vote, it has been actively developed since 1984!!! and yet still receives regular updates from its dedicated developer, I think it is not going away anytime soon.
I remember having looked at Bookends earlier but at the time it couldn’t search Web of Science, Scopus or Science Direct. I checked now and it seems that Web of Science has been added but not the others. Maybe then it could be a reasonable alternative to Papers 3, which I am still very happily using.
The things I like about Papers 3 is the search function, annotations which can easily be ”exported” by copying them together with the reference and abstract to (in my case) Scapple and Scrivener (into Research), grouping of every item into collections, either manually or automatically based on specific phrases or words, and there is an iOS app that is really good which I use most of the time. Especially for reading.
Maybe Bookends can now do all this?
What I really like(d) about Sente are/were its notetaking abilities. Those included the ability to (1) take notes “quoting” either text or rectangular selections (e.g. figures) and (2) add my own notes to each quote. As far as I’ve been able to tell, Papers3 can’t do that. In Papers3, I think it’s also not possible to correct text taken from a PDF when the PDF is of poor quality and there are missing letters or ligatures have been ignored or transposed incorrectly.
Those abilities in Sente meant that I could include all the information I was interested in from a paper in my notes, and in principle never need to refer to the PDF again. It seems that in Papers3 the notes only really work in conjunction with the PDF and/or another program containing the corrected and annotated notes.
The dedication of the Bookends developer is truly impressive. When he retires, though, is there any realistic prospect of the software continuing?
rickdude: agree, but Keith could get run over by a bus (again) and us scrivener users would be in a mess too. What I consider artisanal software like Scrivener and Bookends (and yes Sente was in the same class), is always susceptible in a way larger organisation software isn’t. But I still prefer the close relationship artisanal software engenders. Bookends can export all its data, all my PDF note streams (which are separate to the PDF like in Sente, though I don’t think it does ligature correction!), easily transfer to OPML so you can import it to Scrivener etc. And to be fair to Bookends, there are two Mac developers Jon (the founder) and a newer developer Carlos, but I’d be speculating on what Carlos would do when Jon retires…
lunk: Scopus API is closed and costs the developer to support it, but for example I made a Scopus search for Bookends as part of my Alfred workflow; select a reference and you get “cited-by” and main scopus page links in the Alfred results list (and copied to the reference notes). What does science direct (also closed API I assume) do that WoS or Scopus doesn’t? Bookends does “smart folder” and manual collections. And it has an iOS version, though I don’t really use it (iPad never made its way into my workflow). Each reference manager has its own set of tools, and different ideas about what is important for reference management (more options=better).
But the great diversity we had is slowly eroding, with Sente disappearing and Papers in transition to full-on corporate entity, and as rickdude says, bookends vulnerable to acts of God, perhaps Endnote will be all that is left!!! 
Nontroppo:
Point taken on acts of god. Also, I didn’t know about Carlos…when I frequented the Bookends forums, Jon handled everything.
About ligature correction, I should clarify that Sente never did that. But text quotations from PDFs were just plain text, so it was possible to edit them in Sente to make sure everything was accurate…whereas in Papers, whatever Papers read the selection as being seems to be sacrosanct.
Your mention of the transition of Papers to full-on corporate entity is interesting. Since the old Mek and Tosj days, I guess the company has seemed less personal. Is there anything specific recently that has confirmed that impression? Regarding diversity, it seems to me that the advent of Papers was the key event that triggered the current emerging crisis. The aesthetics of the software and the clever publicity seemed to seduce academics who had never used anything but EndNote, and I remember listening to academic guests on general Mac-centric podcasts singing its praises, especially the software’s “translucent” notes function. That always seemed a weird thing to focus on since, while technically translucent, the notes don’t allow you to see through to the text below and thus are functionally the same as opaque ones. Then Mendeley became really big, and ReadCube came out, and I guess the market became too crowded for Sente: it wasn’t as cool as Papers, nor as long-lived, reliable, and well-supported as Bookends, nor as cheap as Mendeley and Zotero.
Well, you mean apart from the fact it got gobbled up by the pretty superfluous fluff app and corporate ambition readcube? That they have turned papers into a subscription service, and who knows what mess they will make of the feature set?
Reference software is largely a mystery to me, Readcube and Mendeley seem to spend energy forcing themselves between publishers and readers, they want to be “services”, and lipservice to users is flashy UI, search subscriptions and poor reference management. I want a database of my references which I can manage properly (things like SQL and regex control), open PDFs which I can easily access, and a flexible system for citation and bibliography. Readcube seems to force Word 2016 down your throat (sure some strategic partnership). Their “enhanced” PDFs are just a fishhook to trap users into a perpetual subscription, to take an open standard and create yet another paywall for academic content. Bah!
The cynic in me notes that it’s probably less “strategic partnership” (you might be surprised by how little Microsoft cares about partnerships like this, you might not) and more “we rely on features in Word to do the heavy lifting so we don’t have to figure out how to code them ourselves, we just call them via integration code in the Office SDK.”
That, and targeting institutions and enterprises that have already switched to the O365 subscription model for Office so won’t have an issue with a little extra each month for the users that need Readcube licenses combined with only needing to support whatever version is in the Office Pro Plus subscription. Reduce costs for engineering AND support, funnel that money to marketing!
Ha Devin, it sounds like you know what you’re talking about!!! 
I’m sure you’re right, the asymmetry of power is stacked so far in Microsoft’s direction, readcube is the small fry following whatever scraps it can get off the behemoth’s table…
My day job has been in the Microsoft ecosystem for years, so yeah, I’ve seen how their relationships with smaller partners tend to go. That’s not to say that there are not individuals in Microsoft who care very deeply, but that doesn’t always translate to the corporation caring. I can’t count the number of partners I’ve seen who’ve basically been knocked out of the market because their little niche product became just popular enough that Microsoft rolled the functionality (usually the “good enough” 80% subset) into the base product.
I’m glad I asked. I was unaware of all of this. Wow, Bookends gets more attractive every day.