Okay, I think I am long overdue for a post in this thread.
- Indeed the questioner was someone who read this thread - in fact, it was me! At the time I wrote that question I hadn't analysed the line with 12... Nc5 too deeply, but I hard turned Stockfish on and found that it was often leading to 0.00 positions where black was returning the pawn and superficially it seemed like black was in return eliminating white's long-term positional trumps (e.g. the dark-squared bishop) and getting free develop. In other words, finally a solution to the Marshall Gambit that would be good OTB without a computer to defend for you! Once I analysed it with Leela it turned out that this was very far being the case, and a lot of those 0.00 positions were very unpleasant defensive tasks that required lots of concrete memorisation. I don't have my files on my right now so I can't give any exact lines, but I think that anyone who has access to Leela (even on a CPU) should be able to find them pretty easily. Gustafsson pretty much concurred with that assessment, which I was happy to see. In line with Szygy's post, my analysis currently indicates that 12... Nh6 is black's best bet. At the moment my mainline is with 19... Ng5, but I can't recall what my file said about 19... Rhe8. I might take a closer look at it and get back to you.
- I do not currently own either Semkov's book or the Ris database. Under normal circumstances, I'm usually able to access books just a few months after they come out at the library near my university (which I believe claims to have the best chess catalogue in the Southern Hemisphere, and based on my experience with it that claim seems pretty plausible to me). Unfortunately, that's not open at the moment for obvious reasons, so if I want to take a look at the Semkov book I might have to pay for it myself... Lee, I wish you'd instead come on to talk about just how bad the book was
! After the discussion in this thread and I've concluded that 5... a6 is indeed more fun for black than 5... c6 in the Catalan, so I'm switching my main recommendation for the course. So I might have to buy that as well!
- Back on the subject of this thread, I have discovered that in line B2 black gets enough counterplay after 18... Rae8 19. Qxa7 Ne6 20. Rd2 fxe5 21. Bxe5 Bg6 22. Qd7 Bf7! and so I am now much more satisfied with black's prospects in the line with 10... e5. According to my current analysis white's most promising try might even be to just relent to a queen exchange with 14. Qxe5, leading to an endgame where white is very slightly better but I don't think black should be worried.
- I am planning to enter into beta testing on Chessable in the next few days and will open a thread on Chessable announcing that and explaining what sort of help I'm hoping for from them. If any readers here are interested in beta testing keep an eye out on the Chessable forums. The preliminary course title is 'The Sharpest Semi-Slav', but I am open to other suggestions.