i'm not sure i get your point. are you trying to suggest that chemical engineering cannot be self-taught? or that self-taught chemical engineers cannot get work? either point seems suspect.
"No one teaches themselves to build nuclear reactors, and no one is going to get a job in that field as an autodidact unless they have already worked in a closely related field."
Some engineering jobs may require a PE (http://www.nspe.org/resources/licensure/what-pe) certificate, which--although it appears to vary from state to state--requires a four-year engineering degree. Also, most companies in the position to hire a chemical engineer to, say, design and implement a process to distill liquid oxygen, are conservative enough not to entertain the idea of putting an uncredentialed worker on the team. When you're building a plant with the unfortunate capability to explode, poison, or asphyxiate a town, well, isn't it fair to expect that the number of autodidact engineers doing significant work on these projects is close to zero?
Every state allows for substitution of experience for education. Unfortunately, the substitution rate is generally on the order of 10 years of experience or more counting similar to a 4 year degree from an accredited program.
There are also very strict laws for anything that touches the public from the engineering standpoint and also many insurance companies won't touch you without licensure. This will catch up to software engineers some time in the future I imagine and a licensed engineer will be required to oversee and sign off on products just like in other engineering disciplines.
As a tangent, it is actually illegal to call yourself or portray yourself as an engineer without being licensed by the state as such.
"No one teaches themselves to build nuclear reactors, and no one is going to get a job in that field as an autodidact unless they have already worked in a closely related field."
can you substantiate either of these claims?