This means that an analytical solution would either need to be an optimization of brute force, or exploit some additional structure that results from starting with an empty go board.
As an added wrinkle, such a solution may involve assuming that White [0] plays optimally. Ironically, this means that it may still be possible to beat someone who has fully analyzed and "solved" go, by making incorrect moves that puts the board in a position where their analysis only says that a winning move exists.
[0] I am assuming that Black has the winning strategy.
A full analysis of go will have to include some "if N is less than X" clause for an X at least equal to 20, making it less beautiful, but I don't see that as a big hindrance (and X need not be optimal in a proof for a 19x19 board)
In particular, the proof I know of (http://www.cs.bu.edu/faculty/gacs/courses/cs535/papers/Licht...) uses building blocks so large that you can, maybe, fit 2 on a 19x19 board.
Also, I think it is clear white doesn't have a winning strategy, as black can pass at the start of the game.
"In 2009, the first such programs appeared which could reach and hold low dan-level ranks on the KGS Go Server also on the 19x19 board."
I know virtually nothing about what it takes to become a low-level dan ranked player, but I would think it would take more than "just a few months" to "beat the pants off" them.
Back to the subject at hand: I think we will solve mathematical go before we solve chess (where 'solve' is used in the mathematical sense, so that, for example, we can prove "chess is a win for white, in 53 moves", and mathematical go is as described in https://math.berkeley.edu/~berlek/cgt/gobook.html; its difference with regular go variants is the way half stones are counted).
Reason is that both games, even with extensive pruning, are too complex for an full search of their game tree, and go has a simpler structure, making it easier to reason about it without doing that exhaustive search.
[I also doubt I'll live to see either happen]