Cancer research produced many paradoxical results in recent years. The reductionist approach now shows its limits. Considering the origin of the disease at the tissue level and increased stochastic gene expression (SGE) as a driving force, while admitting a role for genetic alterations in cancer progression, might solve these contradictions. Undifferentiated cells are characterized by open and accessible chromatin generating global and highly SGE (high expression noise) which is a hallmark of pluripotency, while differentiation is associated with progressive chromatin closing and decreased noise. Cell-cell interactions stabilize phenotypes and homogenize expression patterns from cell-to-cell during development and differentiation, while disruption of these interactions is responsible for increased expression noise that might be the causal event in cancer by producing phenotypic plasticity. It would produce cancer stem cells defined as cells exhibiting increased SGE that are no more controlled by the microenvironment. Following tissue disruption, differentiation and/or quiescence would no longer be maintained because of SGE. Genetic and epigenetic instabilities would necessary appear, increasing the risk of malignant transformation. The classical perspective is reversed: disruption of the tissue equilibrium is the initiator event, and genetic alterations are tumor "promoters." The major role of genetic modifications in cancer progression is not denied, but microenvironmental and epigenetic alterations would precede the emergence of cancer. If mutagenic exposure, cancer predisposition or spontaneous mutations have already produced genetic alterations, precancerous cells would become more aggressive more rapidly, increasing the probability that a tumor forms, but only if the correct microenvironment is not maintained.