Most data sets for evaluating crop production practices have too few locations and years to create reliable probabilities from predictive analytical analyses for the success of the practices. Yield monitors on combines have the potential to enable networks of farmers in collaboration with scientists and farm advisors to collect sufficient data for calculation of more reliable guidelines for crop production showing the probabilities that new or existing practices will improve the efficiency of food production. The creation of sufficiently large data sets requires the pooling of data from numerous farmer networks, but such pooling of data is currently not possible because there are no standards for sharing of data across networks. The objectives of this paper are to: 1) provide a rationale for a community for on-farm data sharing; 2) describe the challenges of sharing data from on-farm networks and of sharing research data in general; and 3) identify the benefits of data sharing by reviewing what could be gained if data were shared across existing networks in the Corn Belt of the US. Writing and publishing standards for stewardship of data from farmer networks that would include standards for sharing and confidentiality of the data will encourage the creation of large data bases of results from replicated strip trials. The benefits from large data bases of such results are enormous. The greatest benefit is agronomists would be able to move away from the common practice of analysis that answers only the question whether there was a treatment effect to analyses that provide reliable probabilities of the chances a crop production practice will improve the efficiency of food production, and the magnitude of the treatment effect.