Very little can make you crazy faster than reading about NYC construction prices
Great article, you should read every word. Goes into the details of why the projects cost so much (scope bloat), and then the cause (no in-house capacity to sweat the details, leading to many multiples of cost from contractors with every incentive to drive scope higher)
NEW: Why does it cost $100 million to put elevators into a subway station? Which became an exploration of what all is in those projects. Which became an exploration of how and why the MTA's old and bad planning pathologies continue to persist: www.curbed.com/article/subw...
They’re taking a huge bite out of the MTA’s budget.
Long horizon reasoning is fascinating, and how we can enable it is the motivating drive for why a lot of people — myself included — are so interested in robotics and AI. Points2Plan was a recent paper promising great long-horizon reasoning: itcanthink.substack.com/p/paper-note...
"Strong generalization to unseen long-horizon tasks in the real world"
So did any pop-sci social science finding survive? Out: power-poses, Zimbardo prison experiment, robbers cave experiment, willpower depletion, "hand-washing makes you intolerant", etc. etc. What's left?
Life would be a lot easier if occasionally the skies parted and a giant hand gave us a thumbs up when we're on the right track
Teaching is so great for sneakily learning all the stuff you didn't quite understand the first go-round
The day someone makes it easy to slowly unveil equations in powerpoint or keynote is the day lectures get a lot better
Making Large Language Models into World Models with Precondition and Effect Knowledge arxiv.org/abs/2409.12278 I'm an RL guy. To me, a world model maps (state, action) -> state' So let's make LLMs do that. Then we can build planning and reasoning algorithms on top of them.