A point process on the real line (time) or in space in which counts in disjoint intervals are independent and the number of events in any interval of length t follows a Poisson distribution with mean λt for constant rate λ (homogeneous case).

Principle

Independence of disjoint increments plus stationary increments (homogeneity) imply that event counts are Poisson-distributed with mean proportional to interval length; equivalently, interarrival times are independent exponential(λ) in the homogeneous temporal case.

Demonstration

Situation: Calls arrive at a call center at an average rate λ per hour, assumed constant over observation. Recognition: Intervals without overlap are independent; the process is homogeneous. Action: Model arrivals as a Poisson process with rate λ and use exponential interarrival times to simulate or compute wait-time distributions. Consequence: The probability of k arrivals in t hours equals e^{-λt}(λt)^k/k!, and interarrival times are memoryless exponential with parameter λ.

Misapplication

Applying the homogeneous Poisson process to data with time-varying rate (nonstationary) or dependent arrivals. The semantic error is assuming constant-rate stationarity and independence; doing so under time-varying intensity misestimates variances and event probabilities.

Consequence

Correct application yields closed-form probabilities for counts and waiting times, enabling tractable performance analysis and simulation. Incorrect application (ignoring nonstationarity or dependence) produces biased estimates of congestion, under- or overestimation of rare-event probabilities, and faulty capacity decisions.

Reversal

If the event rate λ(t) varies deterministically or stochastically over time (nonhomogeneous Poisson process) or events cluster (overdispersion), independence and exponential interarrival assumptions fail; one must use nonhomogeneous Poisson, Cox, renewal, or cluster process models.

Boundary

Clearly within: Photon counts detected from a constant-intensity source over short intervals with independent counts. Boundary case: Arrival data with slowly varying diurnal rate — over short windows a homogeneous Poisson may be acceptable, over long windows it is not. Clearly outside: Events exhibiting strong aftershock clustering (e.g., earthquakes with Omori-type clustering) which violate independence.

Semantic Tension

Simplicity (tractable analytic form) ↔ Fidelity (ability to represent temporal heterogeneity or dependence): Poisson is analytically convenient but can misrepresent structured variability in real data.

Synthesis

The Poisson process ties memoryless interarrival behavior to independent, stationary counts; recognizing when rate homogeneity or independence fails is critical because these assumptions determine closed-form probabilities and the appropriateness of exponential waiting-time reasoning.