Exchange Layer

~10 million years ago, as the Self's swarms begin exploring higher in the ocean

This is a strange report. A feeding swarm returning from a hunt claims to have encountered another swarm that wasn't there. It heard a greeting, but none of its parts observed another swarm in any direction. This unseen swarm then initiated a slightly garbled status exchange, identifying itself as an exploring swarm although there should have been none in that area.
We would suspect a desynced part is responsible for the confusion, except that the swarm appears fully coordinated. We'll continue to observe its behavior for other signs of concern.

We're no longer concerned about the swarm, but are interested in this new mystery. An exploring swarm brought a report of encountering a feeding swarm that it believes had strayed dangerously far from our wall. It couldn't visually identify this swarm's position, but heard it passing and requested its status to determine its sync.
These two reports in combination lead to the conclusion that their respective or relative locations were responsible for their strange encounter. We'll send a pair of swarms to the same areas with instructions to attempt locating each other in the manner of the previous swarms.

The swarms confirmed the phenomenon. After much changing of position they reached respective areas where, although they were out of mutual sight, they could hear each other clearly enough to converse. There was a slight amount of distortion, but not enough to obscure meaning.
It appears to be a particular layer of the atmosphere where this occurs, since their exchanges were audible at great separation from each other only as long as they were both the same approximate distance from the back of the world. As soon as one retreated a short distance, it stopped hearing the other swarm.

Swarms have been seeking out this layer of the atmosphere. They're taking advantage of the illusion of closeness to relay reports without the need to advance to a sighted position. This reduction in travel time improves the efficiency of each swarm.
We admit to the advantages of this shortcut, but it comes at the cost of not knowing exactly where the reporting swarm is. For now we're willing to accept this uncertainty, but we must discover a means of determining absolute location to keep our information accurate.

After hearing many report exchanges, we've discovered that the length of the time gap between sides of a report exchange is a measure of the distance between the two swarms. By elongating its shape, a swarm can also identify the direction of the other swarm by comparing the same sounds received at both of its ends.
With this, our last concerns about location error are resolved. Any swarm may visit the exchange layer without reservation. By sending feeding swarms to receive reports, exploring swarms are freed from the necessity of returning to the wall. They'll only have to go as far as the exchange layer to re-form between journeys, which will greatly extend the time they can spend traveling.

It's strange, this sense of loss. The new protocol is a success--by using the exchange layer in this way we are learning even more. But now it appears that the inner swarms and outer swarms may never directly meet again. Even though the reports are still shared through the feeding swarms...it's as if we have lost some level of connection with ourself. We didn't expect to feel such sadness.

The ocean's Deep Sound Channel is a layer of water about a thousand meters deep where the speed of sound is at its minimum. Sounds traveling up or down out of this layer refract back into it, and low frequencies may cross entire ocean basins before dissipating.

Time frame: ~10 million years ago