We have parallel problems in science and in software.<p>My faith in science was never in the moral character of scientists and their organizations - individuals and organizations are always vulnerable to corruption. My faith was in the principle of replication. If anyone can repeat an experiment, we can all see for ourselves what is true, and a community dedicated to that (and individuals with a healthy fear of the process) is reliable.<p>Only, we don't replicate experiments. We got so busy and excited building on what had gone before that we've built some huge houses of cards on questionable foundations, because who wants to spend time and money doing replication? Distracted by the free riches, we neglected what had always been the source of our strength, and here we are - arguing over who funded studies and fuming over the replication crisis.<p>Where are the critics who say, "I can't trust that paper - it's impossible to replicate!" Where are our Poppers who insist on falsifiability? An entire community that frowned on complexity and opaqueness and walled gardens of data, a community that trusted things insofar as they had been replicated and re-examined from many angles and proven sound, would force us towards a level of simplicity, honesty, and reliability that science should have. Instead, a general agreement to pursue individual and institutional glory at the expense of upholding foundational principles has rotted the foundation of the endeavor.<p>Put simply, I trust science because you can replicate it. But for whatever reason, (and I can't propose a specific solution, but), to the degree our community is not devoted to replication, it loses its trustworthiness.<p>Software has a parallel problem.<p>I don't trust open source software because I trust the character of developers or institutions. I trust it because it can be examined and fixed. Because of reproducible builds. Because anyone can examine it, anyone can build it, no trust of individuals or organizations is needed. A community that insists on such features and abhors offerings that offend these principles will steer us towards a level of simplicity, comprehensibility, reproducibility that open source software should have.<p>But we are all so excited to build things on top of other things that we spend much more time multiplying dependencies and layering on complexity than worrying about foundational principles. We are now seeing the rotting foundations.<p>There are people who complain about whether code can be examined, or factors that make it difficult. It is becoming increasingly important to listen to them! A community that celebrated open source software, not only for what it can functionally do, but for how <i>open</i> it is, is what is needed to maintain those foundations. A community that has trust issues with unexaminable long dependency chains, that is sensitive to the difference between software that has been around the block and examined for a long time, and software that some guy just put out last night.<p>Put simply, I trust open source because you can examine it. But for whatever reason, (and I can't propose a specific solution, but), to the degree our community is not devoted to examination, it loses trustworthiness.<p>Reserve your trust for communities that take seriously the principles that trust is built on.