A random hopping on a fractal network with dimension slightly above one, \(d = 1 + \epsilon\), is considered as a model of transport for conducting polymers with nonmetallic conductivity. Within the real space renormalization group method of Migdal and Kadanoff, the critical behavior near the percolation threshold is studied. In contrast to a conventional regular expansion in \(\epsilon\), the critical indices of correlation length, \(\nu =\epsilon ^{-1}+O(e^{-1/\epsilon })\), and of conductivity, \(t\simeq \epsilon^{-2} exp (-1-1/\epsilon )\), are found to be nonanalytic functions of \(\epsilon\) as \(\epsilon \to 0\). Distribution for conductivity of the critical cluster is obtained to be gaussian with the relative width \(\sim \exp (-1/\epsilon )\). In case of variable range hopping an ``1-d Mott's law'' \(exp [ -( T_t/T)^{1/2}]\) dependence was found for the DC conductivity. It is shown, that the same type of strong temperature dependence is valid for the dielectric constant and the frequency-dependent conductivity, in agreement with experimental data for poorly conducting polymers.