- implement that conclusion can be a formula instead of just a literal
- allow structural equations to be formulated for literals (ie allow "negative" causal relation)
- What if: Create arguments based on CKB alone (without observations) + arguments based on observations: Is this still correct? would create more interesting AFs
    - Or probably better: argument premise must only be consistent with CM, not the observations
    - Could also handle observations like assumptions (?) Probably not sensible
- add explicit encoding of negative atoms in causalInterpretation

- Parser
- Writer