Measuring Primitive Accumulation: An Information-Theoretic Approach to Capitalist Enclosure in PIK2, Indonesia

This paper applies information-theoretic and statistical-mechanical tools to eight years of satellite data, quantifying the rapid, planned, and irreversible spatial transformation of Indonesia's PIK2 mega-development as a measurable process of capitalist primitive accumulation.

The Gist

The Big Picture: A High-Speed Land Grab

Imagine a massive, low-lying patch of land just outside Jakarta, Indonesia. For years, this area was a mix of rice paddies, fish ponds, mangroves, and small family farms. It was a "commons"—a shared space where people lived and worked.

Suddenly, a giant developer decided to turn this entire area into a massive, high-end city called PIK2. They built roads, skyscrapers, and luxury housing, effectively "enclosing" the land and turning it into private property to be sold for profit.

This paper is like a forensic investigation into exactly how fast, how organized, and how permanent this transformation was. The researchers didn't just look at photos; they used advanced math (from physics and information theory) to measure the "heartbeat" of this land grab.

---

The Three Main Tools Used (The Detective's Kit)

The authors used three specific mathematical tools to tell the story. Here is what they did, using simple metaphors:

1. The "Speedometer" (Information Geometry)

The Concept: They wanted to know when the land changed the fastest.
The Analogy: Imagine driving a car. Sometimes you cruise at a steady speed; other times, you slam on the gas. The researchers treated the landscape like a car. They measured the "distance" the land traveled from its old state (farms) to its new state (concrete).
The Finding: They found a massive "speed spike" between 2019 and 2020. This was the moment the construction really kicked into high gear. Interestingly, they also found a smaller spike in 2022–2023 where developers were clearing land before they even had the final government permission to build. It was like a driver speeding up before getting a green light.

2. The "One-Way Door" (Absorbing Markov Chains)

The Concept: They wanted to know how long it takes for a piece of land to become a building, and if it can ever go back.
The Analogy: Imagine a game of "Snakes and Ladders," but with a twist. Once you land on the "Built City" square, there is a one-way door. You can never go back to being a farm or a forest. The math calculated the "expected lifespan" of the land before it gets swallowed by the city.
The Finding:

• Rice fields had about 46 years left before they would likely be turned into buildings.
• Trees had about 38 years left.
• Water (the ocean) had the longest life (70+ years) because it's harder to build on, but even that is shrinking.
• The "Leakage" is tiny: Once a piece of land becomes a building, it stays a building 96% of the time. It's almost impossible for a skyscraper to turn back into a rice paddy.

3. The "Connectivity Puzzle" (Percolation Theory)

The Concept: They wanted to know if the city grew randomly (like a virus spreading) or if it was planned (like a grid).
The Analogy: Imagine dropping drops of ink on a sponge. If you drop them randomly, they stay as tiny, separate dots until you drop so many that they suddenly connect into one giant blob. This is called the "tipping point."
The Finding: In a normal, random city, you need to fill about 59% of the land with buildings before they all connect into one giant city.

• But PIK2 is different: Even though only 10–16% of the land was built up, the buildings were already connected into one giant, massive blob.
• Why? Because they built roads first. The roads acted like bridges, connecting the buildings long before the area was full. This proves the city wasn't growing naturally; it was planned and engineered from the start.

---

The Key Takeaways (The "So What?")

1. It Was a "Speed Run," Not a Slow Creep
The data shows this wasn't a slow, organic city growing over decades. It was a rapid, aggressive transformation. The math proves that the biggest changes happened in specific, short bursts, often driven by political decisions or speculative bets.

2. The "One-Way Street" of Capital
The study highlights a harsh reality: once land is turned into "Capital" (real estate), it rarely goes back. The math shows that the system is designed to lock land into development. The farmers and fishers who lived there are effectively pushed out, and their land is permanently converted.

3. The "Ghost" of the Future
The researchers found that developers were clearing land (turning it into "Bare Ground") before the government officially approved the project as a "National Strategic Project." It's like a construction crew starting to dig the foundation before the city council has even signed the permit. This suggests the project was moving faster than the law could keep up with.

4. The Human Cost
While the paper is full of math, the story it tells is about people. The "Agrarian" land (farms) that disappeared was replaced by "Capital" (buildings). This means the local farmers and fishers lost their livelihoods. The study quantifies this loss: nearly 13% of the land that used to be farms is now concrete.

The Bottom Line

This paper uses the language of physics and math to prove a point that sociologists have been making for years: The creation of mega-cities in places like Indonesia is not a natural process.

It is a highly coordinated, fast-moving, and often legally controversial "land grab." The math shows that the city was built with a specific blueprint (the roads connecting everything early on) and that once the land is taken, it is almost impossible to get it back. The "primitive accumulation" (the initial grabbing of land to start capitalism) is happening right now, and these tools allow us to measure exactly how fast and how far it is going.

Technical Summary

1. Problem Statement

The paper addresses the challenge of quantitatively measuring primitive accumulation (PA)—the process by which non-capitalist spaces (commons, agriculture) are enclosed and converted into commodified real estate. While urban growth is often studied via percolation models or fractal geometry, these approaches rarely integrate the structural categories of political economy to explain why specific land conversions occur.

The authors focus on PIK2 (Pantai Indah Kapuk 2), a massive coastal mega-development north of Jakarta, Indonesia. This area represents a site of intense speculative development on reclaimed land, characterized by regulatory volatility (e.g., designation as a National Strategic Project and subsequent legal challenges) and rapid transformation from agrarian/mangrove landscapes to built environments. The problem is to develop a framework that can quantify the velocity, directionality, and topological signatures of this enclosure process with statistical precision.

2. Methodology

The authors propose a quantitative framework combining Information Geometry (IG), Markov Chain Theory, and Percolation Theory applied to eight years (2017–2024) of high-resolution (10m) Sentinel-2 Land Use/Land Cover (LULC) data.

A. Data Preprocessing & Aggregation

• Data Source: Esri Sentinel-2 Annual LULC time series (2017–2024).
• Study Domain: 173.4 km² (1.73 million pixels), excluding the marine boundary (70% of the area).
• Marxian Aggregation: The 7 original LULC classes are aggregated into three political-economic categories on a probability simplex:
  • Commons: Trees, Flooded Vegetation, Rangeland.
  • Agrarian: Crops (smallholder agriculture).
  • Capital: Built Area and Bare Ground (treated as a precursor to capital due to its role in reclamation/construction).

B. Information-Theoretic Measures

• Fisher-Rao (FR) Geodesic Distance: Used to measure the "velocity" of landscape transformation on the probability simplex. Unlike Kullback-Leibler divergence (which is asymmetric), FR provides a symmetric Riemannian metric invariant under sufficient statistics.
• Transformation Pulse: Defined as a transition where the FR distance exceeds the mean plus one standard deviation ($\bar{d}_{FR} + s_{FR}$).
• Entropy Analysis: Shannon and Rényi entropies are calculated to track landscape heterogeneity and diversity changes over time.

C. Absorbing Markov Chains

• Model Formulation: The landscape dynamics are modeled as a discrete-time Markov chain where Built Area is the sole absorbing state (once land is built, it stays built).
• Metrics:
  • Expected Absorption Time ($E[T_i]$): The expected number of years for a pixel in a transient state (e.g., crops, trees) to transition into the built environment.
  • Stationary Distribution: The long-term equilibrium state if current transition rates persist.
  • G-Test: Used to test the null hypothesis of temporal homogeneity (i.e., whether transition rates change significantly over time).

D. Percolation Theory

• Site Percolation: Built pixels are treated as "occupied" sites on a lattice.
• Connectivity: Analyzed using the Order Parameter ($\Omega$), representing the fraction of built pixels contained within the largest connected cluster (giant component).
• Fractal Dimension: The box-counting fractal dimension ($d_f$) of the urban boundary is calculated to assess the irregularity of the expansion frontier.
• Critical Threshold: Compared against the random percolation threshold for a square lattice ($p_c \approx 0.593$).

3. Key Contributions

1. Quantitative Framework for Primitive Accumulation: The paper successfully bridges statistical physics (IG, Markov chains, percolation) with political economy, providing a rigorous mathematical language to describe capitalist spatial enclosure.
2. Identification of Transformation Pulses: The use of FR geodesic distances allows for the precise identification of non-linear "pulses" of development that correlate with specific political and construction events, rather than assuming smooth, linear growth.
3. Evidence of Planned vs. Stochastic Growth: By demonstrating that the built environment forms a "giant connected component" at occupation probabilities far below the random percolation threshold, the study provides quantitative proof that the expansion is driven by planned infrastructure (roads, grids) rather than organic, stochastic settlement.
4. Temporal Horizon Estimation: The absorbing Markov chain model provides exact estimates of how long different land-use types (e.g., cropland) are expected to survive before being absorbed into the built environment.

4. Key Results

A. Landscape Trajectory & Velocity

• Structural Shift: Between 2017 and 2024, the Capital fraction increased from 39.7% to 53.8%, while Agrarian land declined from 42.5% to 29.3%. This represents a nearly one-to-one transfer of agricultural land to capital.
• Transformation Pulse: The highest velocity of change occurred in 2019–2020 ($d_{FR} = 0.405$ rad/yr), coinciding with major construction activity. A secondary pulse occurred in 2022–2023, preceding the 2024 designation of PIK2 as a National Strategic Project (PSN).
• Entropy Paradox: Shannon entropy initially increased (0.844 to 1.089 nats) as capital entered the landscape, creating transient heterogeneity. This challenges the assumption that enclosure always immediately reduces diversity; it first fragments the landscape before homogenizing it.

B. Markov Chain Dynamics

• Absorption Times: Under pooled transition dynamics, the expected time for Crops to be absorbed into the built environment is 46.0 years, and for Trees, it is 38.1 years.
• Non-Stationarity: The G-test decisively rejected the hypothesis of constant transition rates ($p < 10^{-10}$). Transition dynamics are highly volatile, reflecting political shocks and regulatory changes.
• Self-Retention: Built Area has a high self-retention rate (~96.4%), indicating that once land is developed, it is structurally locked into the accumulation circuit with very little "leakage" back to agriculture.

C. Percolation & Topology

• Supercriticality: The occupation probability ($p$) of built land is low (0.096 in 2017 to 0.162 in 2024), far below the random threshold ($p_c \approx 0.593$). However, the order parameter ($\Omega$) is extremely high (0.89–0.95), meaning 89–95% of all built pixels belong to a single giant connected component.
• Implication: This "supercritical" state at low density confirms that the growth is planned (via road networks and utility corridors) rather than stochastic.
• Fractal Frontier: The fractal dimension of the urban boundary increased from 1.316 to 1.397, indicating an increasingly irregular, "tentacle-like" expansion into the agrarian periphery, moving away from compact growth.

5. Significance & Implications

• Political Economy Validation: The quantitative results corroborate qualitative critiques of PIK2. The data reveals that "speculative land clearance" (spikes in Bare Ground) often precedes formal regulatory authorization (PSN designation), suggesting a pattern of "facts on the ground" created before legal approval.
• Legal & Social Context: The study provides a physical baseline for understanding the displacement of smallholder farmers and fishers. The rapid conversion of agrarian land (46-year absorption time) translates directly to the loss of livelihoods for communities in villages like Muara and Kronjo.
• Methodological Advance: The paper demonstrates that information-geometric tools can detect the "kinematic and topological signatures" of capitalist accumulation. This offers a new way to monitor and critique mega-projects globally, distinguishing between organic urbanization and state-backed, speculative enclosure.
• Policy Relevance: The findings support recent legal challenges (e.g., the Indonesian Supreme Court's 2025 decision revoking PIK2's PSN status) by providing empirical evidence that the project's spatial evolution is driven by contested, non-stationary political forces rather than organic market dynamics.

In conclusion, the paper successfully operationalizes the concept of primitive accumulation, transforming it from a theoretical political economy concept into a measurable, quantifiable spatial process using advanced statistical mechanics.