like. direct manipulation of source tokens, and definition of new functions is done by selecting the tokens that you’d like to be the definition and then giving a name

there’s live feedback throughout

I finally got around to finishing the video. Extremely thought-provoking. It's kinda interesting to think of PANE and Brief as two poles with complementary strengths and weaknesses (with LoGlo in the uncanny valley in between) * PANE shows everything, but it's less dense. A single value is often printed in multiple places on screen. In Brief there's a designated place for showing runtime values, so less real estate is expended on them. * Brief shows a stack at the top level, but once definitions are created they get 'reified' into textual definitions. This is nice and dense, but solves just one level of the problem of not making me simulate in my head. I kept waiting for the author to show some way to "drill down" into a defined word to visualize how it constructs its intermediate results, but that still seems to be an open problem. It's also a little klunky that there are two ways visually to represent the same expression, one with boxes and the other in plain text. * LoGlo seems to have the drawbacks of both. It doesn't have any way to show intermediate stack states, and it also has no way to drill down into intermediate expressions the way PANE does.