Maybe an example is appropriate. Monocrop cultures are efficient short term, easy to control ... but also simplistic and brutal. In this system, gmo can help reduce the nefarious intrants (fertilizers, pesticides ...). This sounds like optimization.
Alternative cultural models, multicrops for instance, are saner but implies many technical and economical issues that still have to be dealt with. This is more like an architectural change.
One is more like studying a natural process while the other is the mechanics of that process.
[1] https://www.scientificamerican.com/article/fertilizer-runoff...
I do agree we should invest more heavily in tech like bug resistance to reduce necessary pesticides.