Nowadays, the Internet is mostly used for content generation, sharing and access. Users are no longer interested in connecting their computers and mobiles to an end system located at somewhere in the network border, but at obtaining pieces of content. This paradigm shift in Internet usage has motivated the proposal of Information-Centric Networking (ICN) architectures. ICN architectures replace the current connection oriented architecture for a general data oriented one. Contents are identified by their names and the architecture uses those names for routing and retrieving contents. The routers temporarily store copies of requested contents. Therefore, ICN introduces a largely distributed in-networking caching infrastructure. When content’s copies are highly spread across the network routers, end-to-end routing algorithms between client and servers are no longer the most effective way of retrieving contents. In this work, we extend DIVER, a routing algorithm that explores the network in order to search and retrieve router’s copies that are closer to the clients. DIVER uses probe packets to find routers copies, and the routers, upon receiving the probe packets, answer them by inserting their availability information in a space-efficient data structure. DIVER keeps the most diversified as possible the acquired availability information in the explorations in order to raise the probability of satisfying future requests for the same content. Simulation results show that DIVER is able to retrieve four times more chunks from routers compared to the end-to-end traditional routing.