The participation of citizens through mobile applications in detecting fires or other events, as well as scenarios where there exists a large number of sensors with different noise characteristics raises the question of which data points to accept for the estimation task. If the underlying state dynamics and noise statistics are known, there are various filter-based approaches in the literature, with the well-known example of the Kalman Filter. In this paper, we tackle the problem of selecting which points should be considered to estimate the state of a system with both sensor characteristics and dynamics unknown. By exploiting the techniques from resilient consensus, we first build the intuition that the choice must follow some scoring function. Thus, resorting to rating and reputation systems, we propose an algorithm that assigns scores to the measurements and maintains a pool of the points considered to have better quality. We prove that the rating procedure returns mean scores that are better for sensors with smaller variance and show through simulations the reduced mean error of the estimator in comparison with the state-of-the-art alternatives.