Glad to find someone who actually met Mark Boon. And yes, indeed, this was a game between AIs instead of “players” (if you pay attention, you would notice I never used the word players in my post or replies to refer to either side). The black side is Goliath, the white side is Go Intellect, the second place of the 1991 Ing’s International Computer Go Congress) for the Computer vs Computer category. Below is the score sheet (they actually met in the first round)
And here is the full-size picture of Mark Boom (I cropped out the computer for obvious reason)
Here is the picture of the creator of Go Intellect - 陳克訓
Here is the picture of the creator of Many faces of Go - David Fotland, who also keeps records of the International Computer Go Competitions. He won the best UI design (at the time called best screen design 最佳螢幕設計獎)
And there are some “inconsistencies” for the record in 1991 Ing’s Computer Cup, here are the pages of the first hand report
It says Goliath beats 3 dan rank players 林至涵 陳威宇 and 張凱馨, who were 4d, 4d, and 1d at the time (林至涵 and 張凱馨 became pros later, currently 9p and 6p) with 17 handicaps, but lost to all 3 of them with 15 handicaps, and there were no records of them playing 16 handicaps, we even have pictures of how the 17 handicap stones were arranged
The 15 handicap games were played after the main event (after dinner), and I highly doubt there was any unofficial “16 handicap event” against “pros”,
The “ranking system” at the time in 1991 in Taiwan was very different from that of today. The Ing’s rules/system has its own ranking method, and technically not a “professional” system. It has three “tiers” 品(pin), 段(dan), 級(kyu). The pin tier is akin to pros, and instead of counting up, it counts down with strength, with 9 pin the lowest, and 1 pin the highest rank. Each rank in pin gives the lower rank 2 points in Ing’s rules (with even game 8 points komi and black wins in draw). Below pin, there is an overlap of the pin and dan rank, 5 pin is 9 dan, and 9 pin is 7 dan. The discrepancy came from dan rank gap is 4 points instead of 2 points, and dan rank players can attend tournaments as well (all tournaments were held with handicap stones or points, so in a way dan players were semi-pro). Since “pros” in pin teir are way stronger than those in competitions, and the dan players in this competiton were all still very young kid at the time, the likelihood of Goliath beating a pin teir mature pro at 16 handicap at the time is very slim.
We can also “measure” Goliath using the Ing’s ranking. Below dan rank, the gap changes over time. It was originally measured with 6 or 8 points per rank (in the 1970s and 1980s), however, later increased to “1 hand” (16 points equivalent in the 1990s). And they are considered real amateurs, and didn’t compete in the same tournaments. 1 dan player at the time needed to give the lowest rank 18 kyu tier players 17 handicap stones. And Ing’s system wasn’t the only “kyu ranking”. There were also the 甲乙丙丁戊己庚辛 groups and each group would give the next at least 2 hands, and this “big gap kyu” was likely the source of why they skip from 17 handicaps to 15 handicaps (and we can see from the records of Ing’s Computer Cup, that the rewards were also awarded based on 17, 15, 13, 11, 9, etc. handicap gaps).
This 1991 competition has historical significance, because this was the first time a computer program Goliath, finally broke into the “kyu” rank of 17 handicaps, and got the reward for the first time ever. So technically, the rank of Goliath at the time would be awarded in the Ing’s system as 18k (or 17k and likely why it was thought as 16 handicaps), but failed to reach 16k. While Go Intellect, failed to reach kyu tier in the Ing’s ranking.
Having said that, when I first saw this record between Goliath and Go Intellect, I was very surprised as how “sensical” Goliath played, and I personally wouldn’t rank it as 18k or 17k, and as most in the thread, would likely put it at least high ddk, or even sdk if I just saw the first 50 moves (thus why I start the post as a guess the rank, instead of a normal historical thread). There was a review by Nick Wedd in 1997, which as 1k can beat Goliath with 9 handicaps. The strength of Goliath is in its “sensible shapes”, but not quite as good at mid-game fighting, hence likely around ddk to nearly sdk range in the 1990s ranks.
If I were to guess, for 1991 they would probably be considered around 9k for Black and probably 12k for White?
