Cognitive Demands:
Wikipedia's section, "Computers and Go," within the article "Baduk Board Game" gives us a fine explanation of why it took several decades longer for computer programmers to finally defeat the top human players in baduk The fastest computer on earth (as of March 2014, NUDT's "Tianhe-2," which could sustain 33.86 petaflops) needed four hours to analyze the next 8 moves in baduk. This is something trained humans can do at a glance. However, to play well, one must analyze the next 100 moves, and this quantitatively overwhelms the computer. Another point to consider is that a play at the beginning of the game can have ramifications one hundred or more moves later. While humans can handle this complexity and beauty, training computers to "think" like humans required a revolution in AI development. Google and Facebook competed for the honor of defeating the first top level pro, and Google, with Alpha Go team finally won that battle in March of 2016.
Furthermore, given that both hemispheres of the brain are fully engaged, excellence in baduk play depends not only on the quantitative functions of the“left brain,” but also requires rich participation in the cognitive realm of the aesthetic, creative, intuitive and numinous functions associated with the "right brain.” Hence, it was little wonder that computer and baduk experts had such difficulty predicting when, if ever, computers would catch up with humans in baduk -- which was the last game, among cognitive strategy games, for which computers could not beat the top human players, according to "The Mystery of Go: The Ancient Game That Computers Still Can’t Win.” BY ALAN LEVINOVITZ, 12th May 2014 in Wired Magazine.
Here is an engaging article discussing the first great computer go victory over a human pro. The gains in Artificial Intelligence (hereafter "AI") were astounding, as programmers needed to develop programs which could recognize patterns, obviating the need to do exhaustive searches in a game of go.