In this paper, we present a new low complexity approach to network coding trading traditional random linear network codes against an extension of LT codes. Network coding is an appealing paradigm in the context of content dissemination as it significantly improves throughput, leveraging path diversity in networks, be they logical or physical. In linear network coding, nodes send linear combinations of packets they have received. However, computing the optimal set of linear functions to apply at each node requires a global knowledge of the network's topology. Random linear network codes (RLNC) address this issue and rely only on a local knowledge of the topology. Yet, decoding requires an O(n^3) Gaussian elimination. Following the observation that randomized network codes have been built upon rateless codes (random linear codes), we explore the feasibility of network coding inspired from another class of rateless codes, namely LT codes, which can be decoded in O(n log n) operations. We propose a re-encoding method to extend LT codes into new low complexity network codes (LTNC). In P2P dissemination of information, we observe that LTNC trades advantageously communication optimality of RLNC with decoding cost as it incurs only 38.5% of bandwidth overhead for a gain of almost 99% in decoding cost.