Actually when writing this comment, I was thinking about the LISP paradigm: RNA is both code and data.

except so much messier, it's more like code and data in the sense that self-modifying machine code is.

Not really. The 3D structure of DNA is also important for its function. The difference is that RNA is more reactive (chemically speaking), else they are pretty much equivalent.

A better way to think of it I think is that RNA is both software and hardware.

Yes but RNA is much more destructible, whereas DNA is more resilient. RNAses(enzymes that degrade RNA) are everywhere whereas DNA is double stranded and the conditions required for its degradation are harsher. DNA may have 'evolved' as an aberrant form of RNA and because of its durability, it has an increased lifetime. This renders it more apt to serve as an archive that is copied multiple times. An organism relying only on RNA for information storage would have to deal with the shorter lifetime of RNA and that it is more damage prone. That organism would have to spend more energy just propagating its blueprint faithfully and so its population would not grow as fast. Still DNA is not so durable that its sequence is not malleable, like proteins. If I were designing the most robust organism, I would prefer to encode information in amino acids. I would then design a polymerase that would read the protein and copy it. The problem is that proteins are so robust that they are not easily unfolded to be read.

Perhaps a tenuous analogy can be made to information storage in transistors versus magnetic media. Transistors can serve as storage medium as well as logic gates. Yet power fluctuations can corrupt their function. Magnetic storage however can be read many times and is more durable with respect to environmental insults. Of course the most durable form of storage might be laser etching defects in a diamond disc but at the cost of a slow energy intensive process.