To find out, the team plans to use a variety of technologies. The deep web approach, for example, attempts to outline a grammar system. It then checks whether the conversation ‘units’ meet these rules. If not, adjustments are made and try again. “Computers optimize and verify rules very quickly,” Andreas says, “replicating them thousands or millions of times to create a set of rules that explain the data well.”
Of course, progress depends on researchers gathering enough data. Machine learning requires massive amounts of information, but Gero registrations only run into the thousands. Finding patterns in whale language would likely take tens of millions of codes, possibly more.
And as Jiro suspected with Drop and Doublebend, scientists think they have to compare communication with behavior. Is there a specific coda that appears before the hunt, or a sequence that is produced when whales get together to mate?
“It’s the cocktail party problem,” says Gruber. If you hand out a few microphones at a party, you can record snippets of the conversation. But when you see people – arm-touching, looking in the room for better company – “the whole scene starts to make a lot more sense,” says Gruber.
A revolution in animal communication research
On Monday, the team presented important steps to achieve this goal. CETI leaders partnered with Dominica to spread more whale monitoring technology in the country’s waters. I became CETI, too The Daring TED Project Eight major charitable donors have been appointed for the project. The team also received funding from the National Geographic Society.
CETI researchers have already spent a year developing a massive array of advanced, high-fidelity underwater sensors that will record sounds 24 hours a day in most of the Jiro’s whale study areas. Three of these monitoring systems, each attached to a float on the surface, are lowered directly to the bottom and equipped with hydrophons every few hundred meters.