Notes / 2024
Expected Value, Tilt, and Thinking in Bets
I've been playing tournaments since 2008. Long enough to have watched the game change twice over, long enough to have lost more buy ins than I want to count, and long enough to know that the actual skill poker teaches has nothing to do with cards.
The first thing poker beats out of you is the need to be right immediately. You can make the correct decision, the mathematically sound one, and still lose the hand. That used to feel unfair. Now it just feels like information. A good player isn't the one who wins the most pots, it's the one who makes the same correct call whether it worked last time or not, because the decision was never about that one hand. It was about the thousand hands like it. Engineering rewired the same way once I let it. You don't judge a deploy process by whether last Tuesday's release went smoothly. You judge it by whether the process holds up across a hundred releases, and one clean run doesn't prove anything either way.
Tilt is the part nobody warns you about until it's already cost them money. It's not just anger, it's the quiet decision to deviate from a plan you know is correct because the last ten minutes felt unfair. I've seen it wreck sessions and I've seen the exact same pattern wreck production incidents. Someone gets one alert that turns out to be a false positive, gets annoyed, starts skipping the checklist, and the second alert that actually matters gets the sloppy response instead of the practiced one. Tilt doesn't care if you're at a felt table or a terminal. It's the same failure, wearing a different shirt.
Bankroll management taught me more about risk than any postmortem ever has. The rule in poker is brutally simple: never put yourself in a position where one bad outcome ends your ability to keep playing. Move that rule into engineering and it stops sounding like poker advice and starts sounding like basic system design. Don't build the thing where one failure takes down everything downstream with it. Keep enough reserve, enough redundancy, enough margin, that a bad night doesn't become a bad month. Every over leveraged player I've watched go broke made the same mistake every fragile system makes. They sized for the average case and got hit by the one that wasn't.
What poker really trains, more than math or patience, is deciding well with information you'll never fully have. You don't get to see the other player's cards. You don't get to see the exact load a service will take in production six months from now. You size your bet, or your architecture, based on probabilities and get comfortable being wrong sometimes in service of being right on average. These days poker is a couple of sessions a week around a day job that doesn't leave much more room than that, but the sessions I do play still feel less like a hobby and more like maintenance on a way of thinking I use every single day at work.