Fix the intro to be a bit more informal

Author: willGraham01

README.md

@@ -75,7 +75,7 @@ given $\epsilon > 0$.
 The extreme (max and min values) of $d$ are referred to as the "query bounds on the causal response", and serve as the extrema of our quantity $d$.
 Problems of this type are what `causalprog` refers to as "causal problems".
 
-`causalprog` provides utility for setting up causal problems [using DAGs](docs/theory/glossary.md#abbreviations), which can then be solved via your favourite stochastic optimiser and minimisation algorithm (though the package also provides a few solvers itself to help).
+`causalprog` provides utility for setting up causal problems [using DAGs](docs/theory/index.md#abbreviations), which can then be solved via your favourite stochastic optimiser and minimisation algorithm (though the package also provides a few solvers itself to help).
 
 ## Getting Started
 

docs/theory/index.md

@@ -5,7 +5,8 @@ On this page we attempt to give a brief introduction to the mathematical framewo
 ## Causal Models
 
 Causal models are the central building blocks around which `causalprog` operates.
-At the highest level, one can just think of _a causal model_ as some black-box object $G$ that takes some data point $x\in\mathcal{D}_{eval}\subset\mathbb{R}^N$ and predicts an outcome $G(x)\in\mathbb{R}^M$ based on $x$.
+At the highest level; one can just think of _a causal model_ as some black-box object $G$ that is aware of the existence of some variables $X_i$, and is able to make predictions about quantities derived from the $X_i$ subject to how it has been programmed.
+Here "how it has been programmed" is an analogy for "how the model has been tuned / trained".
 
 ### Definition