diff --git a/doc/runtime.md b/doc/runtime.md
new file mode 100644
index 0000000..f920a5e
--- /dev/null
+++ b/doc/runtime.md
@@ -0,0 +1,111 @@
+# Runtime
+
+Any ginger data-structure can be put into a ginger file. The data itself has no executional meaning on its own, but if
+the data is properly formed it can be parsed by the ginger interpreter and a thread of execution can be started. For
+example, if I put the following in a file called `test.gng`:
+```
+(1 2 3)
+```
+
+It doesn't have any meaning, it's just a list. However, if you put the following:
+```
+(add 1 2 3)
+```
+
+and run `ginger test.gng` then a program can be interpreted from the given data. This file describes how given data can
+be formed into a valid program.
+
+## Eval
+
+Ginger evaluation is done the same was as other lisp languages: the first item in a list is the function name, the rest
+of the items in the list are arguments to the function. In the example above, `add` is the function name, and `1`, `2`,
+and `3` are arguments to the function.
+
+Arguments to a function can be functions to be eval'd themselves. An equivalent to the example above would have been:
+```
+(add 1 2 (add 1 2))
+```
+
+## Doing multiple things
+
+It's not very useful to only be able to do one thing. A vector of lists is interpreted into sequentially eval'ing each
+item in the vector. For (a trivial) example:
+```
+[ (add 1 2)
+  (sub 4 (add 1 2))
+  (mul 8 0) ]
+```
+
+## Variables/Scope
+
+The above example does a few things, but it repeats itself in the second part (with the `sub`). If we could save the
+result of the addition to a variable that would be awesomesauce. The `=` function does this! It makes it so that the
+first argument is equivalent to the evaluation of the second argument. Variables' first letters must be upper-case, this
+is how they're differentiated from raw string literals. The above could be re-written as:
+```
+[ (= AdditionResult (add 1 2))
+  (sub 4 AdditionResult)
+  (mul 8 0) ]
+```
+
+In the above example `AdditionResult` is a valid variable inside of the vector that contains its declaration, and any
+vectors contained within that vector. For example:
+```
+[ (= AdditionResult (add 1 2))
+  [ (sub 4 AdditionResult) ;This works
+    (mul 8 0) ] ]
+
+[ (add 4 AdditionResult) ] ;This does not work!
+```
+
+## Literals
+
+We've determined that a list is interpreted as a function/arguments set, and an upper-case string is interpreted as a
+variable name which will be de-referenced. What if we want to actually use these structures without eval'ing them? For
+these cases we have the literal function `l`:
+```
+[ (concat (l (1 2 3)) (l (4 5 6))) ; => (1 2 3 4 5 6)
+  (println (l "I start with a capital letter and I DONT CARE!!!")) ]
+```
+
+## Functions
+
+### Anonymous
+
+Anonymous functions are declared using the `fn` function. The first argument is a vector of argument names (remember,
+all upper-case!) and the second is a list/vector to be eval'd. Examples:
+```
+[ (= Add3 (fn [Num] (add Num 3)))
+  (Add3 4) ; => 7
+
+  (= Add3Sub1
+    (fn [Num] [
+      (= Added (add Num 3))
+      (sub Added 1)
+    ]))
+  (Add3Sub1 4) ] ; => 6
+```
+
+`fn` returns a function, which can be passed around and assigned like any other value. In the above examples the
+functions are assigned to the `Add3` and `Add3Sub1` variables. Functions can also be passed into other functions as
+arguments:
+```
+;DoTwice takes a function Fun and a number Num. it will call Fun twice on Num and return the result.
+[ (= DoTwice
+    (fn [Fun Num]
+      (Fun (Fun Num))))
+
+  (DoTwice (fn [Num] (add Num 1)) 3) ] ; => 5
+```
+
+### Defined
+
+Defined functions attach the function definition to a string literal (lower-case) in the current scope. They are useful
+as they support more features then an anonymous function, such as inline documentation. These extra features will be
+documented elsewhere. To create a defined function use the `dfn` function:
+```
+[ (dfn add-four-sub-three [Num] [
+    (= A (add Num 4))
+    (sub A 3) ])
+  (add-four-sub-three 4) ] ; => 5
+```