Entropy-Based Evidence for Bitcoin's Discrete Time Mechanism

Bitcoin derives a verifiable temporal order from probabilistic block discovery and cumulative proof-of-work rather than from a trusted global clock. We show that block arrivals exhibit stable exponential behavior across difficulty epochs, and that the proof-of-work process maintains a high-entropy search state that collapses discretely upon the discovery of a valid block. This entropy-based interpretation provides a mechanistic account of Bitcoin’s non-continuous temporal structure. In a distributed network, however, entropy collapse is not completed instantaneously across all participants. Using empirical observations of temporary forks, we show that collapse completion unfolds over a finite propagation-bounded interval, while remaining rapid in practice.

💡 Research Summary

The paper “Entropy‑Based Evidence for Bitcoin’s Discrete Time Mechanism” offers a comprehensive empirical and theoretical investigation of how Bitcoin’s notion of time emerges from its proof‑of‑work (PoW) consensus rather than from any external clock. The authors begin by framing the problem: decentralized systems need a reliable ordering of events, yet Bitcoin deliberately avoids a trusted global timer. Instead, time advances only when a miner finds a valid block, an event that simultaneously resolves the uncertainty accumulated during the preceding interval.

The theoretical model is built on three pillars. First, PoW is described as a massive collection of independent Bernoulli trials: each hash attempt succeeds with probability θ = 1/(D·2³²), where D is the difficulty. With a total network hash rate H, the product λ = Hθ is the effective block‑arrival rate. In the rare‑event limit (θ → 0, H → ∞ while λ stays finite) the waiting time T to the next block converges to an exponential distribution Exp(λ). Consequently, block arrivals form a Poisson process. The authors also argue that parallel mining by many participants does not break this property because each miner explores a disjoint sub‑space of the 2²⁵⁶ hash domain; the minimum of many independent exponential waiting times is still exponential with aggregate rate λ.

Second, the paper formalizes Bitcoin’s difficulty adjustment as a discrete feedback rule applied every 2016 blocks (one epoch). The next epoch’s difficulty D_{k+1} is set to D_k · (T_epoch / T_obs,k), where T_epoch = 2016 · 600 s is the target epoch length and T_obs,k is the observed length of epoch k. This feedback automatically compensates for changes in H, keeping λ close to the protocol target over long horizons.

Third, the authors introduce an information‑theoretic view of each block interval. The cumulative probability that a block has been found by time t is p(t) = 1 − e^{−λt}. The binary entropy H(t) = −p log₂ p − (1−p) log₂ (1−p) quantifies the uncertainty about whether the interval has already produced a block. H(t) rises from zero, peaks when p = 0.5, and stays high for most of the expected 600‑second interval. When a block is finally discovered, the entropy collapses abruptly to zero: the local node abandons its current search and starts a new interval from the newly accepted tip. This “entropy collapse” is identified as the discrete tick of Bitcoin time.

To validate these ideas, the authors analyze a dataset of roughly 425 000 block arrival times recorded from 2016 to 2024, covering 211 complete difficulty epochs (excluding 22 incomplete ones). Their empirical findings are as follows.

1. Exponential Inter‑Arrival Distribution – The histogram of inter‑arrival times matches an exponential density with estimated mean 585.43 s (≈9.76 min). The sample mean and standard deviation are nearly equal, as expected for an exponential law.

2. Epoch‑Level Stability – When the mean inter‑arrival time is computed per epoch, values cluster tightly around the 600‑second target, with most epochs falling within 600 ± 100 s. A systematic bias toward slightly shorter intervals is observed in the second half of each epoch, reflecting ongoing hash‑rate growth within an epoch while difficulty remains fixed. The paired‑difference test shows a statistically significant average reduction of 9.30 s in the latter half (p = 0.0010), corresponding to a 1–2 % increase in effective hash rate.

3. Serial Independence – Autocorrelation analysis for lags 1 through 20 (after trimming the extreme 1 % of observations) yields values indistinguishable from zero, confirming that inter‑arrival intervals are independent and supporting the Poisson assumption.

4. Entropy Concentration at Block Discovery – Using the global λ̂, the authors compute the binary entropy for each observed interval length. The resulting distribution is sharply peaked near the maximal entropy region, indicating that most blocks are found while the system’s uncertainty remains high. Only a small tail of very short or very long intervals falls into low‑entropy zones.

5. Distributed Completion of Entropy Collapse – Because nodes receive block announcements at different times, the collapse of entropy is not instantaneous network‑wide. The authors quantify the residual uncertainty by measuring fork durations: the time between the first observed block at a given height and the last competing block before the network converges on a single chain. The empirical survival function of fork durations (log–log plot) shows that the majority of forks resolve within 1–2 seconds, but a non‑zero tail extends to tens of seconds, reflecting propagation delays and simultaneous discoveries. This provides a lower bound on the time needed for global entropy collapse.

Overall, the paper makes four substantive contributions. It empirically confirms that Bitcoin’s block‑arrival process remains exponential across many years and varying network conditions, thereby validating the Poisson model. It demonstrates that the difficulty‑adjustment feedback stabilizes the long‑run arrival rate despite short‑term hash‑rate fluctuations. It provides direct evidence that block discovery typically occurs in a high‑entropy regime, supporting the view that each block constitutes an entropy‑collapse event that advances Bitcoin’s internal, non‑continuous time. Finally, it quantifies the finite, propagation‑bounded interval required for this collapse to be completed network‑wide, using fork duration statistics as a proxy.

The authors conclude that Bitcoin’s temporal ordering is an emergent property of its PoW consensus: a self‑regulating stochastic process that generates discrete time ticks through information‑theoretic resolution rather than through any external clock. This perspective deepens our understanding of how decentralized systems can achieve reliable ordering and may inspire new designs for time‑agnostic consensus protocols.