Categories
Philosophy

[02] Recursive Harmonics: The Mathematical Structure of Communicative Organisation

02-DD

Disinformation is one form of information. Its persistence belongs to the wider problem of how information circulates and changes in densely connected communication. A reply can recast a challenge as evidence of concealment, and later exchanges can reproduce that relation through different claims and participants. The relational framework follows the dependencies that make this possible. Recursive harmonics concerns their timing and how recurrent activity changes the conditions of its own recurrence.

Recurrence as an Observable

Begin with recorded replies, references and their times. Let Xn count replies in a selected channel during interval n, using equal intervals of duration δt. Retain which earlier item each reply addresses and when the two events occurred. The count gives a signal for recurrence through time. The links show where those recurrences occur and which exchanges they connect. Initial detection can proceed at this level, before assigning claims to categories of meaning, accuracy or intent. The choice of records, links and intervals still shapes what becomes observable.

Autocovariance measures how departures from the mean vary together across a time lag. For a real series with finite variance and constant mean μ, write C(k) for its autocovariance when it depends on the lag alone.

C(k) = E[(Xn − μ)(Xn+k − μ)]

E denotes expectation under the model, and the elapsed lag is kδt. The mean and covariance are invariant under a common shift of the sample indices. This is second-order stationarity. In an evolving communicative process, assess that assumption over the period used for estimation.

From Lag to Frequency

With second-order stationarity and a finite sum of |C(k)| over all lags, the Wiener–Khinchin theorem gives the power spectral density of the fluctuations about the mean (Wiener, 1930, and Khintchine, 1934).

SX(f) = δt ∑k∈ℤ C(k) e−2π i f k δt

The sum runs over all integer lags. Frequency f is measured in cycles per unit time, and i is the imaginary unit. The factor δt gives density per unit frequency. Integrating over one sampling band, from −1/(2δt) to 1/(2δt), gives C(0), the variance of the count. The spectrum describes how that variance is distributed over frequency. See recurrence and frequency.

A peak away from zero identifies a candidate recurrence period. The same period can arise from recurring responses, ordinary patterns of participation or events that affect many participants at once. Compare the spectrum with overall activity and return to the recorded links. If activity changes across the record, compare specified observation windows and assess the statistical assumptions in each.

Timing Across the Network

Let Yn be the corresponding series from another channel on the same sampling clock, with mean μY. For jointly second-order stationary series, cross-covariance compares X at one time with Y at a lag of k intervals.

CXY(k) = E[(Xn − μ)(Yn+k − μY)]

Applying the same Fourier formula to CXY gives the cross-spectrum SXY when the sum converges absolutely. Suppose Y is a copy of X delayed by d intervals, so τ = dδt. The cross-covariance becomes C(k − d), giving the following relation.

SXY(f) = e−2π i f τ SX(f)

The delay preserves the two individual power spectra. Where SX(f) is positive, their cross-spectrum records a phase difference of −2πfτ modulo a full turn. At a positive frequency, a delay of one quarter of a period gives −π/2. Recorded lags help resolve the ambiguity from phase repeating each turn.

Across several channels, the cross-spectra record joint variation by frequency and by pair. Compare them with the links through which people encounter and respond to communication. The topology is the pattern of those routes, including how they change through use. A shared rhythm can arise from a common input. Evidence about exposure and response is needed to identify which paths carry influence.

At the scale of the ensemble, distributions of delays and collective modes become objects of inquiry. Statistical descriptions also make gases and plasmas intelligible through their collective relations. How far these descriptions express a common dynamics remains open.

Meaning and Recurrence

Meaning is a function of the frequency structure of relations. When and where an expression returns, how it is taken up and how often its responses recur all shape its significance. Established meanings then influence what recurs and through which routes. The handbook uses harmonic structure for this distribution of timing and coupling across scales. Each measured series gives one view of it. See semantics follows frequency.

Reciprocal Coupling

Consider two processes represented by cyclic phases θ1 and θ2 in radians. Their uncoupled rates of phase change are ω1 and ω2. Take a two-oscillator form of sine coupling with equal positive strength K in each direction and zero propagation delay (Kuramoto, 1975).

dθ1/dt = ω1 + K sin(θ2 − θ1)
dθ2/dt = ω2 + K sin(θ1 − θ2)

Each rate depends on the phase of the other process relative to its own. Exchanging the phases reverses the sign of the sine term. Define Δ = θ2 − θ1 and Δω = ω2 − ω1. The difference between the two rates governs how Δ changes.

dΔ/dt = Δω − 2K sin Δ

A fixed phase difference Δ* satisfies Δω = 2K sin Δ*. For |Δω| < 2K, the solution with cos Δ* > 0 is locally asymptotically stable. Small changes in the phase difference decay because the derivative of its rate at Δ* is −2K cos Δ*, which is negative. At that solution the phases share a rate while retaining a nonzero difference when their natural frequencies differ. This is phase locking. See How Things Persist.

With equal natural frequencies and zero coupling, the phase difference is constant from the start. With coupling, a phase difference can become stable through the dynamics. An empirical application needs evidence for the cycles, the influences between them and the conditions under which the proposed law applies.

To represent a delay τij from process j to process i, use θj(t − τij) in the influence on θi(t). This requires phase histories over the largest delay. Locking and stability then have to be assessed for the delayed dynamics.

Collective Phase Alignment

For N defined phases, represent each by a unit vector eiθj. The vector points in the direction of that phase. The Kuramoto order parameter R is the magnitude of their average.

z = (1/N) ∑j=1N eiθj
R = |z|

R lies between zero and one. It equals one when all phases coincide and zero when the vectors cancel. Two equally sized, internally aligned groups a half-cycle apart give zero. Examine the phase differences and their evolution alongside the population average.

Recursive Change and Constraint

The oscillator example specifies fixed parameters. In a recursive communicative model, realised exchanges change the conditions of later exchanges. A burst of replies can renew attention to an earlier claim and change which responses recur on a longer timescale. The transition rule introduced in [01] represents this dependence on history. To test an update rule, specify how a realised transition changes later probabilities and compare its predictions with those of a model that holds the rule fixed.

Shannon entropy describes uncertainty over specified outcomes of a transition (Shannon, 1948). Compare it across conditions using the same outcome definitions and time horizon. Assess persistence by following the dependencies that make the next transition possible. Assess physical entropy production through the irreversible processes and exchanges involved in realising communication. These calculations require their own variables and units.

Intervention in the Frequency Domain

One experiment changes when a communication is made available while holding its wording fixed. Compare the timing, paths and recurrence of subsequent replies with comparable exchanges. A second comparison changes the wording while preserving the timing of its release. These comparisons test which temporal relations sustain the pattern and which changes alter its meaning or persistence. The proposal applies to information more generally, including disinformation interdiction.

Control is always provisional. Changes in timing or access become part of the dynamics and can change the relationships a model was fitted to describe. The next chapter takes up measurement and diagnostics for these changes.

References

  • Khintchine, A. (1934) ‘Korrelationstheorie der stationären stochastischen Prozesse’, Mathematische Annalen, 109, pp. 604–615. doi: 10.1007/BF01449156.
  • Kuramoto, Y. (1975) ‘Self-entrainment of a population of coupled non-linear oscillators’, in Araki, H. (ed.) International Symposium on Mathematical Problems in Theoretical Physics. Lecture Notes in Physics, 39. Berlin: Springer, pp. 420–422. doi: 10.1007/BFb0013365.
  • Shannon, C.E. (1948) ‘A mathematical theory of communication’, Bell System Technical Journal, 27(3), pp. 379–423. doi: 10.1002/j.1538-7305.1948.tb01338.x.
  • Wiener, N. (1930) ‘Generalized harmonic analysis’, Acta Mathematica, 55, pp. 117–258. doi: 10.1007/BF02546511.

Comments Closed

Your email address will not be published. Required fields are marked *