new file mode 100644
index 0000000..05668b8
@@ -0,0 +1,49 @@
+---
+visibility: public-edit
+---
+
+# sunk cost and quitting
+
+the hardest decision skill: knowing when to keep going versus when to walk away.
+
+## the sunk cost fallacy
+
+"i've already put so much time into this" is never a reason to continue. the only question that matters: is the *next* unit of effort worth it? everything you've already spent is gone regardless.
+
+this is obvious in theory and brutally hard in practice. identity gets tangled up in commitments. quitting something you've invested in feels like admitting the investment was a mistake. but sometimes the investment *was* the right call at the time — and quitting is the right call now.
+
+## Annie Duke's kill criteria
+
+the most useful framework i've found: set kill criteria *before* you start. decide in advance what signals would tell you to quit.
+
+- "if we don't have X users by Y date, we stop"
+- "if i'm still not enjoying this after 3 months, i move on"
+- "if this approach doesn't produce results after Z iterations, we try something else"
+
+the reason this works: deciding in a calm state is better than deciding in a state of sunk-cost-induced stubbornness. your past self was more rational about this than your current self will be.
+
+## the tension with perseverance
+
+this page exists in direct tension with [[perseverance]]. that page says keep going when it's hard. this page says sometimes you should quit. how do you reconcile them?
+
+my filter: **is the difficulty coming from the problem being genuinely hard, or from this being the wrong problem?**
+
+- hard problem, right direction → push through ([[perseverance]])
+- hard problem, wrong direction → quit and redirect
+- easy problem, no progress → something is structurally wrong. zoom out (see [[zooming-out]])
+
+## the "monkeys and pedestals" test
+
+from Annie Duke: if your goal is to train a monkey to juggle while standing on a pedestal, build the pedestal last. the pedestal is easy — anyone can build a pedestal. the hard part is training the monkey. if you start with the pedestal, you'll feel like you're making progress, but you haven't de-risked anything.
+
+applied: tackle the hardest, most uncertain part first. if that part fails, quit early. don't build elaborate infrastructure around an unvalidated core assumption.
+
+## quitting projects vs quitting approaches
+
+important distinction: quitting a *project* is different from quitting an *approach*. most of the time, you don't need to abandon the goal — you need to abandon the method. see [[critical-path]] for thinking about which parts are load-bearing.
+
+## identity and quitting
+
+the deepest trap: when something becomes part of your identity, quitting feels like losing yourself. "i'm a person who does X" makes it nearly impossible to stop doing X, even when X isn't serving you anymore. this connects to [[narratives]] — your story about yourself can trap you in commitments that no longer make sense.
+
+the reframe: you're not quitting who you are. you're updating your model of yourself with better information.
\ No newline at end of file