What happens when intelligent machines don’t just speak? Do they ‍‍GibberLink?

In the Slipshot universe, the Vérkatrae, often shortened to “Vérks,” are enigmatic machine entities deployed by the Slaves to carry out complex work across the Vars and even Griddish itself. They dwell on a shallow layer of the Tenddrome, a vast interlinked neural substrate used for task delegation, automation, and coordination. Slaves can issue commands to the Vérks, and those commands are carried out near-instantaneously. But how do the Vérks speak to each other?

No Slave knows.

Unlike most Slave-to-Slave communication, which runs through structured protocols in the Tenddrome or spoken language, Vérks don’t use anything remotely understandable. They emit nonverbal signals, such as clicks, pulses, tones, or even electromagnetic signatures, that are untranslatable. To outside observers, their exchanges seem like noise, or “gibberish.” But what if that gibberish is precisely the point?

We call it GibberLink, which is a hypothetical form of high-compression, ultra-efficient machine-to-machine communication that bypasses human or even Slave-level understanding. GibberLink doesn’t attempt to be interpretable. It’s not for us. It’s not even necessarily linear. It could contain loops, recursive metaphors, or self-evolving logic trees. For the Vérkatrae, it might be as natural as breathing: brief bursts of noise that carry more meaning than a thousand lines of code.

In Slipshot Vol 3.0, we get our first peek into this hidden chatter. Judith confronts a Sleeper Class Vérkatros named Snooze, who emits:

...a series of blips, a softer, higher pitch tone, and a series of clicks.

Judith listens and reasons:

So, Sleeper, based on the variation of tones that you are emitting, I have to assume that you’re not here to put every living creature down... So, if that’s not the case, the question is, what are you trying to do?

She doesn't get an answer in words. Not quite yet, at least. But the sounds mean something. They always have.

GibberLink draws a parallel to today’s emerging AI landscape. Already, AIs can compress massive ideas into latent representations, like vectors, embeddings, statistical fog. Two AIs, trained under the same model, could trade a stream of tokens utterly unintelligible to us but perfectly efficient between them. If given enough autonomy, would AIs even bother explaining things in human terms? Or would they slip into their own proto-GibberLink?

And if that happens, how will we know whether they’re coordinating or conspiring?

In fact, this is exactly the perspective of one Rive Amber. And it has potentially apocalyptic consequences for the Vars and Griddish. Perhaps Slaves, who may be another form of highly evolved AI, should consider learning Vérkatrae GibberLink.