I think it stands to reason that the quality of a game depends on you definition of quality. To some, this could mean optimal play. To others, there is something fascinating - skillful - maybe finesseful (?) about having the skill to play well under time pressure.
Consider the field of AI. Many machine learning algorithms are not deterministic. The use probability to infer likely best decisions given a set of incomplete information. Some such algorithms can do amazing things, from detecting stuff in images, to predicting stock trends, to even playing go under time constraints. If I were to measure the quality and intelligence of such algorithms compared to a brute-force algorithm that they teach in a high school math class, the non-deterministic algorithm is often much more complex, elegant, and impressive to me.
In the same way, I am not impressed if someone can brute force the solution to 5x5 tic-tac-toe in 5 days. I am impressed if someone can show the aptitude to make probabalistic decisions in the face of uncertainty under tight time constraints playing the same game.
I won't claim that crappy fast play is impressive. But near-optimal play under fast time constraints is. It shows intuition, skill, and quick thonking more than slow games often do.
If a computer can brute force go by calculating for 1000 years, that's impressive. But if there's a 5d algorithm that can decide moves in under 5 seconds - well, that's more pressive to me. It shows true skill and aptitude in the underlying algorithm - something more than basic, run-of-the-mill brute force.
Wow, there's an awful lot to challenge there. The very first statement - you pays your money and you takes your choice - may seem unobjectionable, but definitions of quality are also subject to quality tests. You get what you pay for. While I can believe that a tag artist who climbs onto a bridge parapet and daubs his graffiti at high speed before the police helicopter comes over might impress some people more than an artist who first builds a secure scaffold and then paints the Sistine chapel at leisure. But I don't think they are being impressed by the art.
I'm no mathematician, but I still have a hunch that the work of people like Samuels who found the alphabeta algorithm and all its subsequent variations that made brute-force searches a practical reality in AI is at least as impressive, era for era, as the probabilistic work of other academics who seem mostly to be building on that work.
But the main objection is surely to the implication that go experts who play slowly are using brute-force. As far as I can tell they are using exactly the same probabilistic thought processes as if they play fast, with the crucial difference that they have time to do a quality check. I can't see any significant downside to slow play in go quality terms. Fast play just suits the short attention spans of modern tv audiences better. Not much of a recommendation.
