One could make an argument, that the set placement of japanese rules is the most efficient way of placing 2-9 handicap stones on an empty board, and if black chooses to place those stones freely anywhere else on the board it is “less optimal”
So in a way, white should be happy if black chooses anything else than starpoints, since it is giving white small advantage over the more traditional handicap placement.
I actually recently had a 9hc game whre black decided to do this…
So yeah…
the calibration with the rating system can be argument. But then we could work on this and try to determine with free placement, how to calibrate with the rating. It could lead to some interesting conclusion for the traditional Japanese system, in my mind the big jump between 6 and 5 stones for ex.
Not even a real game, but rather for teaching, so whatever both sides agree.
In a tournament setting, always star for the sake of simplicity.
Please elaborate on this argument, I’d be very interested in the rationale.
Why are 8 star-points more efficient than - I don’t know - 4 corner enclosures?
I highly doubt that the traditional Japanese handicap placements (on the star points) are optimal. However, I don’t think they were ever meant to be optimal, and their continued use has to do to with adhering to tradition, rather than any belief of optimality.
For 8 handicap stones, I would much prefer @martin3141’s suggestion
or something like this…
I wonder what KataGo thinks is the optimal placement for the X handicap stones…
While 4 enclosures isnt bad either in my opinion, it is leaving more room for white to settle his/her groups right from the start. If white is indeed 8 stones stronger than black, white should be able to keep those b’s enclosures from growing into huge territories.
But yeah, i feel like making shimaris is also really good option for black. Almost as efficient as star points ^^
There is a proverb saying “First corners, then sides, then center.”
@_KoBa would you prefer the traditional placement of 5 Handicap stones
… over one where the center stone is positioned at the side?
tbh i dont see that much difference between the 2, both seem like good positions to make with 5 stones. Lower one is like normal 6hc position but with 1 stone taken out. Can’t be bad, right?
Optimality considerations aside, I feel that it’s also a matter of individual preference. Even if some configuration of handicap stones is objectively worse than the usual placement, the player may feel more comfortable and get better results.
By my experience the difference between 4 and 5 Handicap stones is much less than the difference between 5 and 6 Handicap stones. Why might that be the case? Because even though tengen is a popular opening move due to its symmetrical position, it’s far inferior to moves near the corners and sides.
So from the two positions above I believe that the second one is much more efficient than the first.
Kata seems to think that difference between “normal” position and well but freely placed positions are tiny.
For example with 9 stones, normal fixed placement gives black about 109 point advantage, while 4 shimaris + tengen is worth “only” about 107 points. With less than 9 stones that difference is even lower, assuming that black places those handicaps in a reasonable way.
When comparing the two 8hc positions you posted, kata gives about 99 point lead for black with both of those placements. And for both of those 5hc positions kata gives about 58 point lead for black.
But the thing is, i cant find any set-up what kata considers better than the normal fixed placement, at the very best its worth the same as fixed placement. And there are many possible placement where blacks lead is worse.
So i’d say that free handicap placement can be just as good as the fixed one, but it can also be a lot worse.
ps. If someone is running kata on good hardware and wants to spent the effort on analysing different positions, i would love to know what are the placements which kata considers equal to fixed placement, and if there is indeed something what could be even better.
Kata sees no difference because nobody can give it 5 stone handicap. 
I did play with that a tiny bit, but with only the level 2 ai. If the difference is like 3 or 4 points at say 8 stones I can’t tell if that’s a real difference or still within the error margin/confidence interval for katago at this early stage of the game.
Maybe with more playouts one could be a bit more certain but even still.
“only” about 88 point lead for black, so less optimal than 8hc could be xD
The thing is you don’t need a margin of 100 to win but you need more security to get a winning margin.
If cutting solidly in half the board make it easier to grab more as to put 9 stones in the traditional way and ask to make them work together…
Played around even more with the score estimator and found this gem:
… compared to
… but the difference is certainly negligible.
I think there are perhaps even hundreds of very playable one- to nine-stone handicap opening formations.
The subtlety lies in these words like “optimal”, “playable”, “good” etc.
The difference between them in this context is, to me, not very interesting. I don’t care greatly whether an initial lead is evaluated at 90 pt or 100 or 110. That is not a useful measurement.
