Working through a slightly more promising line. Game 1 allowed Black far too much counterplay on the kingside.
As always, feel free to provide any potential refutations.
Novelty discovered in the Berlin Defense
Sort:
Probably the most straightforward path to equality for both sides, unless something better comes along.
Play_e5 wrote:
It's not a novelty. It's the engine's best move on that position
Second best actually. 7. d4 is the preferred line.
You say it isn't a novelty, can you show which previous opening books/players discovered it before? If not, then it is uncharted territory.
Not really. 7. d4 is strong enough objectively to override any temptations from 7. b4, which Stockfish agrees is probably handing Black a small advantage. I sat SF10 on move 7 to a search depth of 47, and d4 is the clear winner of the two.
If anything, 7. b4 occasionally wasn't even second-best in SF's opinion, as 7. Nc3 was also a contender.
I know, and I don't care. My interest is in exploring with what I have, and evolving lines where I'm able.
You're welcome to play it if that's what you believe
Feel free to use the analysis I've posted as inspiration. xD
Let me know what happens if Black plays 7... Nxb4, that's still a bit of a blind spot, though it generally looks like Black wouldn't have adequate compensation for accepting the gambit
A few other unknown contributors on LiveBook have suggested 7... a5?!, 7... Nf7, 7 ... Be7 and 7... Nf5 as possible alternatives for Black's initial response. I have no darn clue which engine thought up a5 and Nf7, but I can confirm Nf5 is Deep Fritz 13 at depth 18 (not likely to be very high quality) and Be7 is from Komodo 11.2, depth 19 (quality engine, but relatively shallow)
It's not even really a pawn sacrifice after
Black doesn't get left with a great position here.
Looking past 7... Nxb4 at the most critical line, Black may actually be able to get away with capturing on b4. After a series of forced moves, Black's position is somewhat closer to equal in development, but still has a ways to go.
Interesting. We may need to check for improvements in this line, because if there aren't any then this whole gambit may just be a fancy draw offer from White. White needs to find some better way to punish Black for accepting the pawn sac, and/or Black needs to find a better way to entrench.
As it stands after 7..Nxb4:
If Black can comfortably draw after White sacs a whole pawn, then this opening is pointless. Can't have that! 
I haven't forgotten!
I've switched computer processing to other lines for now, as our current "mainlines" and results suggested that this novelty wasn't very promising (to date, SF10 in self-play defeated White once with 7... a6 and drew with 7... Nxb4). So current data suggests there's better theory to work through.
I had thought we had seen a brand new, 5th-move Sicilian invented during the CCC, Stockfish vs Komodo. Play progressed as follows--
But apparently it was just the server's fault for missing 5. Qxd4 and moving the entire PGN file awry by one move.
That is indeed the Berlin mainline
en español no le entiendo inglis
Minor note, this project will be revisited soon! I'm getting a massive hardware upgrade, and I am curious to see what LC0 will do to this line, Stay tuned for more inspired nonsense!
So far, looking at previous moves, 36. Rh1 and 26. Qc2 look to be the defining mistakes of that line. However, in both cases White couldn't claim to have an advantage even by that point, so it looks like I'll have to roll it back earlier.