diff --git a/benchmark/fib.lua b/benchmark/fib.lua
new file mode 100644
index 00000000..576024a7
--- /dev/null
+++ b/benchmark/fib.lua
@@ -0,0 +1,6 @@
+function fib(n)
+  if n < 2 then return n end
+  return fib(n - 2) + fib(n - 1)
+end
+
+io.write(fib(35) .. "\n")
diff --git a/benchmark/fib.py b/benchmark/fib.py
new file mode 100644
index 00000000..1ce57cc5
--- /dev/null
+++ b/benchmark/fib.py
@@ -0,0 +1,5 @@
+def fib(n):
+  if n < 2: return n
+  return fib(n - 1) + fib(n - 2)
+
+print fib(35)
diff --git a/benchmark/fib.wren b/benchmark/fib.wren
new file mode 100644
index 00000000..76973e56
--- /dev/null
+++ b/benchmark/fib.wren
@@ -0,0 +1,9 @@
+var fib = fn(n) {
+  if (n < 2) {
+    n
+  } else {
+    fib.call(n - 1) + fib.call(n - 2)
+  }
+}
+
+io.write(fib.call(35))
diff --git a/doc/implicit fields.txt b/doc/implicit fields.txt
new file mode 100644
index 00000000..bf394eb6
--- /dev/null
+++ b/doc/implicit fields.txt	
@@ -0,0 +1,29 @@
+Q: Can fields be implicitly declared?
+
+The idea is that just using a name starting with "_" somewhere in a class
+automatically defines a field with that name. Implicit fields are particularly
+nice because it means they don't have to be defined all before methods. (Since
+we have a single-pass compiler, we would have to otherwise a method could
+only refer to previously defined fields.)
+
+One potential problem is with nested classes. This is more important if we
+consider a module effectively a class. Consider:
+
+  class Outer {
+    foo {
+      _blah = "value"
+    }
+
+    class Inner {
+      io.write(_blah) // Does this declare field in Inner, or access Outer?
+    }
+  }
+
+Looking at this, though, I think there's already a question how referring to an
+outer field would work. Having an instance of Inner doesn't imply you also have
+an instance of Outer. We definitely don't want to recapitulate inner classes
+in Java.
+
+Q: What about static fields?
+
+A: Different naming convention? __foo?