Integrating different technology is more like design of experiment to me than algorithm. Partly because you can't always do "integration" the analytic way you can with algorithm. Rather "integration" involves planned trial-and-error from observation and testing, sometimes due to the shear complexity of the sum of everything or reasons that are out of your control.

"Integration" is like doing science or social science; algorithm, math.

A totally irrelevant wandering thought: I think simplex algorithm is another algorithm of the century, besides FFT.