TalkChess.com

[Larry Kaufman]

Here is how I see the situation. First, Mr. Yonenaga is no longer on a par with Moriuchi or Habu or Watanabe (the top 3), but to compare his shogi level to David Levy in chess is ridiculous. Based on his level when he retired and the amount of time since then, I would expect that he plays about the level of the bottom class of pros, which is to say at the same level as the very best of amateurs. So clearly still "grandmaster" standard, but not Elite grandmaster standard in chess parlance. By the way, Karpov still plays at Grandmaster level, just not elite level. I'm sure I would be no match for Karpov even now, and I earned the GM title 3 years ago and have not declined appreciably yet. Okay, I'm a pretty weak GM, but still...
To define when a computer reaches the level of the top 3 pros, you must specify the time limit. I think it is quite likely that Bonkras is already on a par with the top 3 at 30" per move, the fastest time limit used in any televised "rapid" games. At 3 hours per side (plus 1 minute per move thereafter), a fairly typical "pro" time limit, I think it will happen between 1 and 2 years from now; one mid-level pro told me he thinks Bonkras would already beat him at such a time limit. At 9 hours per side (plus 1 min) over two days, the "meijin" time limit, maybe we need 3 more years.
I propose we define 3 hours plus 1 minute as the time limit for purposes of talking about relative strength of human and computer in shogi.
Finally I'll say that by the end of the decade, pros will need to take a lance handicap from the best engine to have any chance of victory, just as they need pawn handicap in chess to have such a chance now.