In this Note a social network model for opinion formation is proposed in which a person connected to q partners pays an attention 1/q to each partner. The mutual attention between two connected persons i and j is taken equal to the geometric mean 1/ √ q i q j . Opinion is represented as usual by an Ising spin s = ±1 and mutual attention is given through a two-spin coupling J ij = JQ/ √ q i q j , Q being the average connectivity in the network. Connectivity diminishes attention and only persons with low connectivity can pay special attention to each other leading to a durable common (or opposing) opinion. The model is solved in "mean-field" approximation and a critical "temperature" T c proportional to JQ is found, which is independent of the number of persons N , for large N .Key words: sociophysics, random networks, opinion formation, Ising modelRecently Aleksiejuk et al.[1] proposed a social network model for opinion formation by introducing ferromagnetically coupled Ising spins on a Barabási-Albert network [2]. The strength of the couplings between linked nodes, J, was taken to be uniform, independent of the number of connections to a node. Simulations of this model indicated the existence of a critical "temperature" below which opinion formation is possible (two-phase coexistence) and above which common opinion is unstable (disordered phase). A peculiar feature of this model is that simulations indicate that the critical temperature depends on the number of nodes, or system size, N in the manner T c ∝ log(N), so that in the thermodynamic limit an initially imposed common opinion always persists (in the absence of an external opposing "field"), no matter how weak the uniform coupling J between each pair of connected persons, or "partners" [1]. Incidentally, this peculiar divergence of T c with N is missed in the simplest mean-field approximation to the model [1], but is captured correctly inEmail address: joseph.indekeu@fys.kuleuven.ac.be (J. O. Indekeu).