Happy Bodhi Day Eve!!!
I thought I’d offer something to meditate upon for Bodhi Day tomorrow (Sunday, December 8, 2024).
Scaling Beyond Comprehension
Everything is easy at a small scale. I usually say that a few times during workshops on business intelligence and/or big data. You can solve problems with precision, maintain clarity of purpose, and refine every detail. But scaling up—whether it’s a database, system, a team, or a model—and suddenly, things don’t just get linearly harder; they seemingly shift to another world! Large scale introduces new bottlenecks that demand different mechanisms, often requiring us to sacrifice features we once thought essential. What works beautifully in a small system can collapse under the weight of a larger one.
But there’s not just small and large scale. Beyond what we currently think of as “large scale,” there will always be another level up—a realm where new dynamics emerge, ones we can’t yet imagine. With each leap, we encounter new challenges, but also new opportunities. Features we sacrificed to accommodate larger scale will give way to capabilities we never even thought about. Yet, these new features can remain invisible if we’re too focused on replicating what worked at the previous scale.
Take Poincaré recurrence, for example—a concept I explored back in 2020. It describes how, over unimaginably long timescales, a system could possibly return to a state identical to its starting point. The sheer numbers involved here—vast beyond comprehension—force us to confront the limits of our mental abstractions. These numbers aren’t like imaginary numbers (as in The Eternal Fishnu is the ith manifestation of Vishnu). They’re real, but they stretch so far beyond our intuition that they feel like they belong to another dimension of thought.
The first time I heard of Poincare Recurrence was enlightening in its own way. The probability of the state of the Universe going into some past state is improbable, but non-zero. In the scale of unimaginably big numbers, non-zero means it will happen some time. The Universe is a really big time crystal where all that has been and all that will be is there for eternity. Nothing is ever truly gone—just hard to get back to at worse.
Scaling isn’t linear. It’s not as simple as getting a case of Cheerios from Costco. It’s more like moving from a cup of water, to a tub of water, to a lake, to an ocean—entirely new dynamics emerge, ones that are impossible to predict when thinking only in terms of size.
The Brain as an Abstraction Machine
Our brains are extraordinary, but they are a product of scaled abstraction. Every second, streams of photons hit our eyes, molecules reach our noses, and vibrations flow into our ears. Yet, we don’t consciously process all that raw data. Instead, our brains distill it into meaningful patterns—shapes, smells, sounds. What we perceive as the world is already an abstraction, reduced to its essence to keep us from being overwhelmed.
We often say that we “filter out noise”. But is noise really noise? Or is it a signal whose value we don’t yet understand? Noise isn’t irrelevant—it’s simply information we can’t use in the current moment. Nothing exists in isolation; everything connects to something else. What seems like background static might one day prove critical when viewed through a different lens or at a different scale.
This filtering process serves us well, but it also purposefully limits us. Our senses only perceive certain slices of reality—visible light, sound frequencies between 20 and 20,000 hertz, smells within a narrow range. These aren’t flaws; they’re features. Evolution has fine-tuned our senses to match the necessary skills required for life on Earth. We don’t need the sharp eyes of a hawk or the nose of a bloodhound to thrive in a community of sentient beings.
But this fine-tuned abstraction leaves us vulnerable to manipulation by those who figured out how to see beyond our natural tuning. By focusing on patterns we expect to see, we can miss what’s really happening—whether it’s an optical illusion, a social bias, or an algorithm’s subtle influence. The same abstraction that helps us navigate the world also leaves us open to sleights of hand, both intentional and unintended.
Unimaginable Numbers and New Dimensions
Poincaré recurrence shows us the staggering scales that exist beyond our daily experience. The idea that a system—be it the universe, a weather pattern, or even a human life—might return to a nearly identical state over incomprehensibly long timescales is humbling. It forces us to grapple with the fact that our mental models are woefully inadequate for imagining these scales.
The numbers involved in Poincaré recurrence aren’t abstract in the way imaginary numbers are; they’re grounded in physical reality, yet they stretch our minds beyond a breaking point. It’s as if they belong to another dimension—one where our intuitive grasp of scale no longer applies. These scales remind us that what seems chaotic or irreversible in the short term may simply be part of a larger cycle when viewed across eons.
This concept feels almost alien because it exists at a scale we can’t operate within. But it also invites us to ask: what happens to our systems—our models, our intelligence—when we scale them to unimaginable magnitudes?
Scaling AI: Beyond the S-Curve
Look at large language models (LLM) today. After just a few years, they’ve seemingly reached what feels like the top of their S-curve—the start of diminishing returns where bigger models don’t necessarily mean better outcomes anymore. Instead of scaling single models endlessly, AI is now shifting focus toward systems of AI, where multiple specialized models collaborate. This mirrors the shift in focus I described in my posts on The Five Aggregates of Machine Learning and The Process of Intelligence.
At a certain scale, it’s not about size; it’s about relationships and processes. Scaling beyond today’s AI might involve entirely new dynamics, where intelligence isn’t just a sum of its parts but an emergent property of how those parts interact. This isn’t unlike the brain itself—not vast in the number of like experiences it processes (like a database table of billions of sales transactions) but in the diversity of the experiences it abstracts.
When we think about scaling AI, it’s worth asking: what features might emerge that we can’t see yet? What dynamics will only become visible when we stop thinking of AI as just “bigger models” and start exploring the relationships between them?
Abstraction as a Feature
Ultimately, we live in an abstraction of reality. Each of us is a composition of countless atoms, yet we don’t experience life at the atomic level. Instead, we move through a world shaped by the limits and strengths of our senses, brains, versatile but still limited Earthly bodies, and tools we’ve crafted for ourselves.
Scaling forces us to confront the limits of those abstractions. It challenges us to see not just what’s bigger, but what’s different. As we scale systems—whether physical, biological, or computational—we uncover new dynamics, new features, and new truths about the nature of reality. Be open to what new scales open for us. Don’t be afraid of it. It’s all a part of our Universe—which means it is us.
On Bodhi Day, as we reflect on the nature of enlightenment, it’s worth remembering that what we think of as “noise” today may one day reveal itself as signal. Scaling isn’t just about growth; it’s about transformation. And in every transformation, there’s the possibility of seeing the world anew.
My Bodhi Day 2024 Service
For those who would like to join my Bodhi Day service—at least in spirit—please look at my blog on Bodhi Day Eve 2017. It describes the service, which I’ll start at 5:00am U.S. Mountain Time on December 8, 2024. I hope that someday I might livestream my service, but as of today, I don’t think anyone will show up … hahaha.
At the very least, you must enjoy a chawan of Mrs. Hanamoku’s rice and milk dish.
Bodhi Day 2024 Themes
As it has been for the past few years, artificial intelligence plays a central role in the theme of the year. My interest in AI is pretty much based on its value towards our introspection as individuals and humanity as a whole—as a collective, distributed intelligence. I urge you to ponder my AI-related blogs for the year. These blogs aren’t about AI itself, but how it relates to our knowledge as Zen practitioners:
- Five Aggregates of Machine Learning – My experiment with categorizing different types of AI among the Five Aggregates.
- Cycles -Everything is a process, where each iteration is a cycle.
- The Yin and Yang of A.I.
- The Process of Intelligence
- End of the Maintained Trail – The state of A.I. and other technologies (ex. quantum computing, genetics, robotics) places us at the doorway to a new paradigm for us. That’s just like it was when we figured out a few tools such as knapping, the wheel, smelting, printing, mass manufacturing, etc.
Faith and Patience,
Reverend Dukkha Hanamoku
Bodhi Day 