🌿  back

21.29 YRS

infosphere megastructures

I’m excited to share that I’ll give an online talk on building customizable and collaborative AI assistants at Joel Chan’s HCI lab, as part of their weekly speaker series. I’ll touch on a few recent projects, especially dual and oneironomicon, framing them as tool AI versus agent AI. It’s going to take place next Thursday (Feb. 24th) at 12:30 ET, and you can sign up here.


Around a year into the pandemic, I read about China’s relatively isolationist policy boasting majority support from the general public. Having virtually no qualification whatsoever to make moral judgements on the foreign policy itself, I was merely struck by how unfamiliar and widespread the invoked worldview was. It was as if I came across one of those documentaries on exotic creatures lurking in the ocean. I can’t tell whether this odd fish is the cruelest or the kindest one in its habitat, but I can definitely appreciate its strange nature. Oh, and there are an estimated gazillion of them out there? Woah! That op-ed led me to a similar sense of awe, a reminder of how diverse the meme pool is outside our day-to-day environment, should we momentarily suspend judgement. The present article describes a few possible ways of relating to this biodiversity of ideas. Amateur commentary on a controversial topic, so tread carefully.

If we invoke the same ambitions of wildlife preservation and habitat conservation in the realm of memetics, rather than animal or plant life, we might suddenly become reluctant to fully buy into a global infosphere. Throughout history, connecting distant lands by new trade routes led to a sudden uptick in disease, with indigenous animal and plant species being vulnerable to pests carried on board, due to their lack of adaptation. Today, we thoroughly disinfect our spacecraft so as not to bring harms to newly discovered worlds. In a more ordinary setting, hygiene has long been standard practice for helping a variety of plants and animals thrive.

Why then, are we not putting more careful thought in information hygiene? If we recognize the value of diversity in ensuring the resilience of populations, why aren’t we reflecting more on preserving indigenous populations of ideas when casually connecting half of humanity through a handful of digital plarforms? The talking points of team sparse revolve around digital colonialism and analogies to monocultures versus tropical forests. Herbert Schiller’s Communication and Cultural Domination is a representative stance.

The rhetoric at the other end of the spectrum goes something along the lines of… Why encourage closed-off bubbles when we’ve seen that naive interventionism breeds fragile individuals? Let ideas face one another directly, expose their flaws and shortcomings instead of idealizing their uniqueness and wrapping them in a protective blanket. The most useful, truthful, and insightful takes might very well prevail. Haven’t you read Nassim Taleb? However, casting survival-of-the-fittest in such an aggressive and competitive light has led to terrible, horrendous pages in the history book of the previous century.

While good ideas come when ideas have sex, there might be value in incubating nascent thoughts before letting them out into the world. People who are into digital gardens speak of growing conceptual seedlings into coherent evergreen ideas (filter by growth stages on link), yet peer pressure hinders the embryonic development of diverging thoughts, even if they’d later cause a paradigm shift. When scaled to entire communities, the entire collective consciousness is often incentivized into the conformity of dogma. In both cases, while priors provide a useful reference for updating beliefs in light of new evidence, social drives might increase this ideological inertia beyond justification, undermining some of the advertised features of a dense infosphere. The influence of user-misaligned algorithms in shaping this landscape for better or worse deserves its own commentary.

However, there’s a middle way. In this case, it takes on the name of holobionts: islands of densely-connected ecologies which themselves are sparsely-connected with each other. As a blueprint for the infosphere, it provides a stronger connection between communities than nations speaking different languages in the past, yet a weaker connection than the present superhighways of information with certain creators having audiences larger than a dozen nations. Those communities arguably exist already on social media, but are eclipsed both extremely dense and sparse structures. The fediverse is even closer to this architecture, yet lags behind in terms of userbase.

Kauffman graphed this effect as a hill. The top of the hill was optimal flexibility to change. One low side of the hill was a sparsely connected system: flat-footed and stagnant. The other low side was an overly connected system: a frozen grid-lock of a thousand mutual pulls. So many conflicting influences came to bear on one node that whole sections of the system sank into rigid paralysis. Kauffman called this second extreme a “complexity catastrophe.” Much to everyone’s surprise, you could have too much connectivity. In the long run, an overly linked system was as debilitating as a mob of uncoordinated loners.

Somewhere in the middle was a peak of just-right connectivity that gave the network its maximal nimbleness. Kauffman found this measurable “Goldilocks"’ point in his model networks. His colleagues had trouble believing his maximal value at first because it seemed counterintuitive at the time. The optimal connectivity for the distilled systems Kauffman studied was very low, “somewhere in the single digits.” Large networks with thousands of members adapted best with less than ten connections per member. Some nets peaked at less than two connections on average per node! A massively parallel system did not need to be heavily connected in order to adapt. Minimal average connection, done widely, was enough.