diff --git a/README.md b/README.md
index 5e5d66d..fb27474 100644
--- a/README.md
+++ b/README.md
@@ -15,3 +15,40 @@ Explanations will be installed automatically as part of the Scribble documentati
 -
 
 Solutions in the respective annual subdirectories.
+
+
+## MB’s Advent of Code tips
+
+* The problems are often designed around a particular computer-y abstraction. If you notice what the abstraction is, and then find the closest analog in Racket, the solution tends to come together quickly. Otherwise, you can spend a lot of time reinventing the wheel.
+
+* Complex numbers are a nice way of modeling two-dimensional positions.
+
+* Use lists whenever feasible, because there are many useful list functions in the Racket library that don’t have vector equivalents. In particular, [these list functions](https://docs.racket-lang.org/reference/pairs.html?q=racket%2Flist#%28part._.Additional_.List_.Functions_and_.Synonyms%29) are very useful, especially `argmin` and `argmax`.
+
+* Vectors are better than lists in situations where you need random access to members.
+
+* `eq?` is the fastest equality check, but it only works for symbols and fixnums (therefore, use more symbols and fixnums so you can use `eq?`!)
+
+* `match` is fantastic.
+
+* Association lists (= lists of pairs) are underrated. They’re compatible with all the usual list functions, of course, but also dictionary forms (like `dict-ref` and `in-dict`).
+
+* The `graph` library can be helpful for graph-based problems.
+
+* It’s good to know about sets and mutable pairs.
+
+* Also the fancier `for` iterators, like `for/first` and `for/or`.
+
+* `let/ec` is a way of jumping out of a deeply nested computation, akin to how `return` works in other languages. 
+
+
+
+## My solutions
+
+* I try to write solutions that are succinct but not cryptic. 
+
+* I don’t optimize for speed.
+
+* I like doing the Advent of Code problems because it forces me to use parts of Racket that I don’t ordinarily use. So I treat it as a chance to expand my awareness of the Racketverse.
+
+* I’m unlikely to finish every problem. Judging by past years, there is a point where the problems get sufficiently complex that I’d rather put that time into improving my other Racket projects :metal: