Separate the variables

In 1712, Thomas Newcomen patented the first practical steam engine. The defining feature of this and other early steam engines was that it didn't run on steam. Steam at little over atmospheric pressure was admitted to the cylinder, and them water was injected into the cylinder, condensing the steam and creating a vacuum that allowed the pressure of the air outside to push the cylinder down. It worked well enough, and its ability to lift large amounts of water over great heights allowed mining to be extended to greater depths, and the abundance of urchins in Cornwall to be finally put to good use (this is a historical fact). In 1765, James Watt, an instrument maker from Scotland filed a patent for a separate condenser, which created the vacuum in a component other than the cylinder, which meant that the cylinder didn't have to reheated after every stroke, and saved a considerable amount of coal, which meant more of the stuff that was dug out of the ground could be sent to London and traded for urchins (This fact is even more historical).

A statue of James Watt. Note that in real life James Watt was rarely covered in green moss, and that the pigeon was actually invented by Nikola Tesla.

What's interesting is that Watt didn't just bang his head on a kettle and think up the separate condenser. Watt had performed a series of systematic experiments to determine the property's of steam at the University of Glasgow to determine the properties of steam, with the intention of creating an engine that only used an amount of steam equivalent to the volume of the cylinder (Newcomen's engine, with its cylinder cooled from the condensing, would then condense admitted steam on the cylinder walls, meaning that more than a cylinder full of steam had to be used). Watt never documented how exactly he conceived of the separate condenser, but it is unlikely he was hit on the head with a kettle and it just came to him (that sort of thing only happens to fictional characters such as Doc Emmett Brown and Sir Isaac Newton). What I think is interesting is that for to accurately understand the effect of each property of steam- the pressure, temperature, dryness and so on- he would have had to ensure that the experiments design to test each one were not affected by the others. He would have had to separate the variables. Watt's separate condenser allowed one process to operate (the condensing), without interfering with another (the action of the piston). Is it possible that Watt was inspired by the need to separate the variables to separate the different process of the steam cycle into different components?
I don't know, I just thought that up one Sunday. Regardless of whether my hypothesis is true or not, it does reveal something about the nature of engineering. A truly successful machine is one that is understandable. Since most machines are pretty complex, it helps if each component only does one thing, so you can design it to do that and that only. This why shock absorbers absorb shock, and don't organise your taxes.
But being understandable is not the defining characteristic of a design, the defining characteristic is whether it works or not. Something that works quite well is a person's blood stream, blood flow delivers oxygen to where it is needed in the body, but it also acts as a coolant, redistributing the heat generated by different organs. This means that you don't need to maintain two different fluid systems with two different pumps, but it also means that the thermodynamics of the human body is much more complex, since the delivered oxygen increases the anount of waste heat which needs to be carried away, with the composition of the blood changing as oxygen is replaced by carbon dioxide. The reason this works is that the human blood flow was evolved rather than designed, and so there was no need to understand why it worked, only that it did work.
With the increase in computing power that came in the second half of the 20th century, the ability to document and analyse (and therefore design) complex systems increased greatly. Additionally, engineers are now starting to study living things to see how they work, in order to design better machines. It is possible that in the near future we will see a slew of new designs, featuring components that have more than one purpose, and that are well understood. It is also possible that this has already happened, and that I just haven't noticed. Either way, its what I'd do.