Aired July 9, 2007

Computer scientists say in the future swarms of inexpensive small robots could help humans with tasks such as mining or even exploring other planets. But they say we have a long way to go to understand how swarms work and transfer that knowledge in a way that robots can use. Scientists at John Carroll University are looking for clues on what they call swarm theory by watching people. Their most recent experiment involved 81 people and the popular Japanese number puzzle, Sudoku. ideastream's Mhari Saito signed up for the game.

Sudoku is usually played with a pencil and paper. Puzzlers arrange numbers in a grid typically made up of 81 squares. On this recent Saturday, the grid is laid out with red and black tape, covering a concrete patio at John Carroll University. The puzzle pieces are people wearing color-coded t shirts with a number on it. My yellow shirt has a number 9. Grandmother Peggy Anderson, a sudoku fan wearing a pink shirt with a number 6 on it, explains the rules.

Peggy Anderson: You have to have one of each number 1 through 9 in each of these, 1 through 9 across and 1 through 9 down.

Peggy and I are among 81 people getting ready to collectively solve a giant human sudoku puzzle. There doesn't seem to be a leader. We're just waiting to salk out onto the grid and find our proper space.

Peggy Anderson: I'm used to looking down at the paper not looking around. Its hard to even picture, but it should be a lot of fun. As long as there's no pushing and shoving.

Are you going to push and shove?

Peggy Anderson: Well, I can get away with it at my age... (laughs)

Tape: Okay, the first puzzle there will be no constraints, it's the easiest puzzle in the world... so get ready... go.

Peggy and I head into the crowd. Its sort of like being in a giant mixing bowl: 81 people milling around, trying to find the square where you won't repeat numbers horizontally, vertically or within your own smaller grid.

It takes me a while to find the right square. When I do, someone's in it.

Mhari Saito: I have to be in this nine...this is the only one I can be in.

Person: Ummmmm...

Mhari Saito: He moves. I get my square and look around. We're almost done.

But another part of the grid can't find a solution. Watching them without being able to go and help them is making us nuts. The woman next to me says someone should take over, someone else disagrees.

Person: They're trying to figure out ...
Woman: I know.
Person: ...how humans solve this, then work out a computer algorithm and reverse engineer it so stupid robots could do it.
Mhari Saito: It would be much easier if one person took over.
Another person: That's right!
First person: Yeah but that's not the idea.

Maybe. But after 15 minutes, advancing science is really annoying.

Person: Can I make a suggestion?
Crowd: Yes.
Person: I just made my own Sudoku the other day, I was bored out of my mind. It just went like that (snaps fingers) 1 through 9 right in that row.

And we all fall into place. When a problem is found, a woman does what the rest of us have been DYING to do: whips out her pencil and paper, finds a fix and organizes everyone.

Dan Palmer: You guys got within two swaps of solving the whole thing in about 7 minutes before that guy took over.

John Carroll University's Dan Palmer says we solved the puzzle, but weren't the perfect swarm. Computer scientist Marc Kirschenbaum says we ran into the biggest problem techies have with swarms - they're not efficient at finding the perfect answer.

Marc Kirschenbaum: This is typical behavior we see in programs using swarm theory that its easy to get most of the things done but its really hard to finish.

The scientists will watch videos of the experiment and map out our movements, at least the part before human leadership took over. I'm Mhari Saito, 90.3.