Stephen J Gould once famously asked: "Should the tape of life be replayed, would it produce similar living beings?". This question may never get fully resolved, but recent technological developments allowed us to ask a similar question transposed to the context of cancer: "If we initiate tumor evolution in multiple similar cells, will the same phenotype emerge?". To tackle this problem, we made use of organoid cell culture models that recapitulates native tumors with increasing fidelity, and expressed cellular barcodes that make it possible to study evolutionary processes with high resolution, to establish an experimental ex vivo platform to study tumorigenesis. Specifically, TP53-deficiency was engineered into primary human gastric organoids from healthy donors and several clonally derived replicate cultures were evolved for over two years. The cultures were extensively interrogated by DNA and single cell RNA sequencing, and cell barcoding was introduced to examine subclonal dynamics. We observed that several features of gastric tumor evolution could be replicated by prolonged ex vivo culturing, that selection was similar across replicates from the same culture and that cultures from different donors evolved similar malignant phenotypes. This suggests that tumor evolution is partly intrinsic to the tumor cells, without the influence of a complex microenvironment, and moreover that this process to some extent is predictable. By "restarting the tape of cancer" under different conditions, we may find a path forward that thwart malignant transformation.