Shut Up Smug Curling Robot

When word broke in 2020 that Korea’s curling robot, dubbed Curly, had defeated human curling teams in three out of four test matches, the internet went bonkers.

“Watch a Robot A.I. Beat World-Class Curling Competitors” bellowed Scientific American.

“Curly the Curling Robot Can Beat the Pros at Their Own Game” screeched The Smithsonian.

Even good old SI chimed in—this was before their partnership with The Curling News—with ”Machines Notch Another Victory in War vs. Humans With Curly the Curling Robot.”

Butt the most irritating story came from Mashable: ”Smug robot consistently defeats humans in one of the most difficult sports—curling.”

This wasn’t necessarily irritating at the time. What has become thoroughly exasperating over the years since is to see the Mashable social media team repost the same story, over and over again, week after week, month after month, year after year.

Then, of course, it gets picked up by other feeds, many of them bots—like “TechButthead.”

Smug Robot. Smug Robot. Smug Robot. 

Well. The Curling News is here to say shut up to the robot ... and to Mashable.

First of all, Curly isn’t even the hottest curling robot around anymore. He’s been usurped by a Chinese creation which made its public debut on the Olympic curling ice in Beijing. Not only does it slide and throw the stone, it was even seen standing up afterwards … which was ultra-creepy.

(We were previously impressed with Beijing’s off-ice robot brigade, but the thing we saw on the ice was impressive.)

Second of all, some are questioning the entire Korean experiment.

“The ‘real games’ played against humans were, in three of the four games, only two ends in length,” reports Doug Suerich, whose online persona CurlingGeek sees him regularly exploring the details most digitally-distracted humans can’t bother to investigate. We should know, as he is an occasional contributor to The Curling News.

“The games were played under mixed doubles rules, and there was no sweeping allowed for humans or robots,” said Suerich. “A key game was against a women’s team, while in the fourth game, the robot’s opponent was Korea’s reserve wheelchair curling team, and that match was scheduled for five ends.”

There are all sorts of problems with this.

First of all, nobody plays two-end curling games. Standard championship-level matches last either eight ends or 10 ends, which puts the players—and robo-players—on the ice for between two and two-and-a-half hours in duration. That’s a lot of time for ice conditions to change—mid-game—which would really challenge the robot’s admittedly impressive ability to react and respond.

Once in a while you’ll play a six-end match or even a four-ender; these tend to happen at charity fundraising events or corporate outings, where fun and frolic is the goal and the quality of play is … well … let’s not bother commenting on that.

Meanwhile, the mixed doubles format is, by design, a shorter curling game. Only 10 stones are used (five per team) as opposed to 16 in the four-player format.

What we are saying, and the CurlingGeek agrees, is that while the robot’s on-ice exploits were quite remarkable, the achievements weren’t exactly as hyped.

“One of the robot’s victories was a 4-0 scoreline,” said Suerich. “Curly scored one in the first end and then stole three in the second. That’s the headline, but it doesn’t reveal what actually happened.

“Curly was lying two in the first end and chose a really dumb shot on his last one. He could have drawn for three but he threw what looked like a very aggressive double attempt for four that jammed, and he scored only one.

“In the second end the Korean humans totally blew their last rock, messing up a fairly simple double-takeout that would’ve given them the win or at least a tie.

“If you mess that up in the eighth or 10th end, okay, you probably deserve to lose. But losing a ‘game’ because you missed a key shot in the second end? Doesn’t really fit the headline of the robot beating the humans … the humans beat themselves.”

Theoretically, the robot’s five-end match against the wheelchair curling team might have been the most accurate gauge of comparative curling skills between man and machine. As fans were recently reminded, there is no sweeping in the Paralympic discipline, and many able-bodied curlers who have tried delivering stones with a throwing stick report that it makes for decent accuracy, at least on takeouts (albeit few have tried throwing from a wheelchair). No wonder throwing sticks are banned in able-bodied championship events.

Another possible red flag: while Suerich believes the women’s team was a strong one—we took a peek and spotted Un-Chi Gim, who is indeed one of Korea’s top female players—the scientific paper doesn’t identify any of the curlers who competed against the robot. The authors called them “expert” and “top-ranked” which an over-enthusiastic media promptly rewrote as “pro” and “world class.” With respect, there are very few true “professionals” in this still-amateur sport, and how many of them would be as psyched for this exhibition as they would for a real tournament?

Give us all the humanoid names, Curly, or we have to question the skill set and motivation of your opponents. Sorry… that’s the way our stones roll.