Blaise Agüera y Arcas on Symbiogenesis and the Computational Nature of Life

Life as Embodied Computation

Life is fundamentally defined by function rather than material composition. While materialism explains the atoms that make up an organism, function describes what those atoms do. This distinction is formalized through the concept of embodied computation: the idea that life is a physical system where the medium of computation and the agent of computation are one and the same.

Drawing on the insights of John von Neumann, the talk posits that for a system to be alive, it must be capable of self‑construction (autopoiesis). Von Neumann's model of a universal constructor—which is functionally equivalent to a universal Turing machine—demonstrates that reproduction requires a set of instructions (a tape) and a mechanism to execute those instructions. In biological terms, this maps directly to DNA and the ribosome. Therefore, computation is not a late addition to life; it is a prerequisite. No computation, no life.

The BFF Experiment: Spontaneous Generation of Code

To demonstrate the emergence of life from non‑life (abiogenesis) in a silicon environment, Agüera y Arcas utilized a system called BFF, based on a modified version of the Brainfuck programming language.

Experimental Setup

• Substrate: A "soup" of 1,024 tapes, each 64 bytes long, initially filled with random bytes.
• Mechanism: Two tapes are plucked at random, concatenated, executed as a program, and then returned to the soup.
• Modification: Unlike standard Brainfuck, which separates code and data, the BFF version is "embodied," meaning the program can read and write to its own code tape.

Results and the Phase Transition

After several million interactions, the system undergoes a dramatic phase transition. The soup moves from a state of "noise" (random bytes) to a state of "programs" (structured, functional code). This transition is characterized by:

• Computational Density: An increase from an average of two operations per interaction to over 1,300.
• Compressibility: The entropy of the soup drops sharply as replicators begin to copy themselves and each other, making the data highly compressible.
• Stability: Replicators emerge because entities that can copy themselves are more stable over time than those that cannot.

Symbiogenesis: The Engine of Novelty

A central mystery of the BFF experiment is that complex programs emerge even when the mutation rate is set to zero. This contradicts the traditional Darwinian view that random mutation is the primary source of novelty.

The Role of Fusion

Agüera y Arcas argues that the source of novelty is symbiogenesis—the fusion of smaller replicators into larger, more complex ones. This mirrors the biological theory popularized by Lynn Margulis, who proved that eukaryotic cells emerged from the fusion of different prokaryotes (e.g., mitochondria).

In the BFF soup, one‑byte replicators exist from the start. When two such replicators fuse and the resulting combination copies more effectively than the individuals, a symbiogenetic event occurs. This process allows for the accumulation of algorithmic information without needing random mutations.

Mathematical Framework

The talk connects this process to Smoluchowski coagulation, a mathematical framework describing how particles (like polymers or clouds) merge. The phase transition observed in BFF is identified as a "gelation" transition, where clusters of replicators diverge in size and the system "sets" into a structured state.

Proving the Necessity of Symbiosis

To prove that symbiogenesis is the driver of complexity, the researchers implemented a "tree depth" limit. By tracking the ancestry of replicators and blocking any interaction that would create a replicator with a symbiogenetic history deeper than a certain threshold (e.g., 20 levels), the phase transition was entirely prevented.

Despite the system still being able to perform basic replication, it never reached the "gelation" point of complex, functional programs. This demonstrates that deep symbiogenetic ancestry is required for the emergence of high‑level computational complexity.

Evolutionary Implications: Intelligence from the Start

If symbiogenesis is the primary driver of evolution, then the "Major Evolutionary Transitions" (such as multicellularity) are not rare anomalies but the tip of a massive iceberg of constant, small‑scale fusion events.

From Life to Intelligence

Agüera y Arcas posits that because life is computational from the start, every symbiogenetic event creates a more massively parallel computer. As these systems grow, they must develop models of their environment and other replicators to survive.

Intelligence is defined here as the act of modeling others. Therefore, the transition from simple life to advanced intelligence is a continuous process of increasing computational parallelism and modeling capacity. This suggests that "theory of mind" and intelligence are fundamental properties of life that scale upward through symbiogenesis, rather than being late‑stage evolutionary accidents.

---

要約: Blaise Agüera y Arcas は、生命は具現化された計算であり、ランダム変異ではなく、レプリカの融合（シンビオジェネシス）が進化的な新規性と複雑性の主要なエンジンであると主張しています。

タイトル: Blaise Agüera y Arcas on Symbiogenesis and the Computational Nature of Life