Lojban is extremely interesting, especially the predicate logic part of it. However, it's still a language that's meant to be written and spoken by humans, which comes with limitations and drawbacks (i.e., binary is extremely flexible and powerful, but not practical to speak and write by a human). It offers no real innovation as far as semantics goes.
What I'm researching is knowledge representation, independently from any physical representation or implementation. Mainly, how to represent facts and intention. The semantic web (linked-data) is very close to what I have in mind as far as facts representation goes. However, it seems to completely ignore the intentional aspect of communication (x wants y because z).
It is not difficult to represent "This pizza costs $10" or "I want this pizza" in RDF (semantic web language). However, that's a naive and simplistic way to think about the situation. It's should be more like a semantic contract:
Person A broadcasts this promise (offer):
- Person B owns Pizza X
- Person A owns $10
Person B broadcasts this promise (demand):
- Person B owns Pizza X
Matchmaking algorithm finds a match (2 people that share similar promises). In this case, both share the "Person B owns Pizza X" statement. However, since all promise statements must be true, Person B has to agree to add "Person A owns $10" to its promise. Once both have identical promises, the contract is sealed. Now, both users are responsible for making these promises (future truths) become reality. The contract is complete when reality matches these promises, and the trust score of both individuals is increased (trust score, like a credit score, determines how reliable people are at respecting promises aka predicting the future). This "trust score" also happens to become the future currency.
We must realize that all communication is about describing either actual truths or future truths. Contracts are simply endorsement of a set of future truths. Whether the truths you endorse become reality or not affects your credibility, and therefore the value of your endorsements. All we need is a framework to communicate these things, and nothing more. That should eliminate 80% of apps we use today, and reduce communication noise significantly.
Of course that's a gross simplification that probably doesn't answer your question.