With the rise of big data technology, it is generally believed that big data is more objective than small data, so people put forward the idea of "Letting the Data Speak for Themselves." The objectivity of big data mainly comes from several aspects: the data scale is more massive and comprehensive; the data format is more diverse and representative; data exists before mining without theoretical presupposition; data are mainly generated automatically by intelligent terminals, thus avoiding human pollution. However, due to the artificial design of data generation, the penetration theory of data mining, the bias of data algorithm, and the black box of data decision-making, the problem of human subjectivity penetration still exists in big data. Therefore, the observation penetration theory in the era of small data becomes the data mining penetration theory in the era of big data.