Building large systems is devilishly hard and all evidence points to us still being bad at it. Since the dawn of electronic computation, the relative effort required to maintain legacy systems has continued to increase, which suggests that we are not solving the problem on our own.
One potential solution could be propagators; a mesh of independent systems who individually listen to neighboring propagators and produce increasingly refined outputs. A system such as this would be more flexible than a traditional system, suggesting that change requests and maintenance could be less expensive.
This talk makes the case that our current development methodologies are unsustainable, introduces the concept of propagators, and shows what a propagator network looks like when written in JavaScript. Hold on because there's code, math, bold claims, and wild-ass-guesses ahead.