It is known that Earth's short-term temperature anomalies share the same complexity index as solar flares. We show that this property is not accidental and is a consequence of the phenomenon of information transfer based on the crucial role of non-Poisson renewal events in complex networks. DOI: 10.1103/PhysRevLett.100.088501 PACS numbers: 92.60.Ry, 05.20.每y, 05.40.Fb A cornerstone of statistical physics is the fluctuationdissipation theorem (FDT) of the first kind and the linear response theory (LRT) of Kubo [1] on which it rests. Recently, with the increasing importance of phenomena whose statistics do not satisfy the central limit theorem, a special version of the FDT was developed to study the response of renewal non-Poisson networks to external perturbations. This theory is a kind of generalization of the well-known theory of stochastic resonance (SR) proposed in 1981 to explain the periodic recurrence of ice ages [2]. The current generalization is called the event dominated fluctuation-dissipation theorem (EDFDT) of the first kind [3,4] and leads to the remarkable result that under certain conditions a complex network being perturbed by a second complex network takes on the statistical properties of the perturbation. This is the complexity matching (CM) phenomenon recently proposed by the authors of Ref. [5] involving the transfer of information, rather than energy, and can be realized with a perturbative linking of extremely weak intensity [4]. CM is expected to apply to all processes dominated by crucial renewal events, as has been shown in the case of earthquakes [6], blinking quantum dots [3], and brain dynamics [7]. Herein we apply this theory to the linking of Earth's climate to total solar irradiance (TSI). The average global temperature is a consequence of the TSI being absorbed and redistributed by Earth's atmosphere and oceans by means of nonlinear hydrothermal dynamic processes [8]. In the past two decades this phenomenon has been framed as a large-scale numerical simulation incorporating all identified physical or chemical mechanisms in an attempt to recreate and understand the variability in Earth's temperature time series [9]. What is not addressed in these simulations, over and above the temperature increase of the past 30 years, are the statistics of the average global temperature and the reasons why the statistics of the temperature anomalies are not Poisson, Gaussian, or any of the other elementary statistical distributions [10,11].Earth's surface temperature responds to variations in the TSI, which can be partitioned into low-frequency quasicycles and the more high-frequency random fluctuations. The low amplitude, approximately 11-year solar cycle influence on climate is well studied, leading to, for example, the decadal oscillation in the stratosphere and troposphere, as well as the resonant excitation of La Nina [12]. On the other hand, on the basis of large-scale computer simulations (see, for example, [13]), the highfrequency random fluctuations in solar activity have been found ...