I saw mention of the Art Of The Propagator paper on Lambda The Ultimate and was intrigued. I was expecting a paper that reinforced the discussion of constraint propagators in SICP (which would have been fine by me), but instead was provided with an unexpected topic.
The paper delves into the question of how you can design a system that handles partial information, information from multiple sources and even conflicting information. Combine this with the ability for propagators to back into information given a known result, and you've got a really interesting approach to system design.
I can imagine a system like this would be helpful when you're analyzing business data and making assumptions from it. As more data is provided, conflicts will no doubt arise, and the ability to handle contradictions seems essential.
This does have me wondering, how does a concept like this move from research into the Real World? Regardless, it's definitely cool research to keep an eye on.