A previous calculation [P. H. Diamond and T.-S. Hahm, Phys. Plasmas 2, 3640 (1995)] of the renormalized dissipation in the nonconservatively forced one-dimensional Burgers equation, which encountered a catastrophic long-wavelength divergence ∼ k −3 min , is reconsidered. In the absence of velocity shear, analysis of the eddy-damped quasi-normal Markovian closure predicts only a benign logarithmic dependence on kmin. The original divergence is traced to an inconsistent resonancebroadening type of diffusive approximation, which fails in the present problem. Ballistic scaling of renormalized pulses is retained, but such scaling does not, by itself, imply a paradigm of selforganized criticality. An improved scaling formula for a model with velocity shear is also given. PACS: 52.35. Ra, 05.60.+w In a famous calculation of the "large-distance and longtime properties of a randomly stirred fluid," Forster, Nelson, and Stephen 1 (FNS) analyzed the consequences of various forcing scenarios for the Navier-Stokes equation and, to some extent, Burgers equation. They considered both a conservative forcing (Model A) and a nonconservative one (Model B), 2 and predicted nontrivial properties for the lowfrequency, small-wave-number limits of the two-point correlation and response functions. They did not explicitly consider Model B for Burgers equation; however, that was later studied in considerable detail by Hwa and Kardar 3 (HK). Recently Diamond and Hahm 4 (DH) attempted to use the Burgers Model B in support of a paradigm of self-organized criticality 5,6 (SOC) for plasma transport. In the course of their discussion, they performed a calculation of the renormalized dissipation coefficient η k that describes the mean propagation of small-amplitude pulses with Fourier components k. In the absence of macroscopic velocity shear V , they invoked a wave-number scaling for the turbulent dissipation (η k ∼ k 2 ) that disagreed with that predicted by HK (η k ∼ |k|). As a consequence, they encountered a catastrophic long-wavelength divergence ∼ kmin dq/q 4 , where k min is a minimum wave-number cutoff. Their result for V = 0 also exhibited pathologies. In the present work, I reconsider the calculations. For V = 0, I find only a benign logarithmic divergence and a wave-number scaling in agreement with HK.