At least that was the theory.
Instead, what happened was: they'd name the game, Steph would say "Oh, I bet you picked the [such-and-such] play, where I [etc etc etc]", before he'd seen a single frame.
Conversation
At first, the commentators thought it was funny. Then they got spooked. Steph's talking about bench player reactions, where everyone on court was, etc
It's really fun to watch. He only partially does it on the first one, but then gets into the spirit
4
32
369
Very similar thing here: LeBron James gets asked by a journalist "what happened next?", and simply recounts _everything_:
3
11
199
[Insert thread about expertise, memory & chunking, w/ examples in domains from tennis to chess to music to math. Mostly, though, it's just fun!]
Here's a beautiful Steph mixtape the NBA made in 2016, the year he was unanimous MVP. Enjoy, it's wondrous!
3
10
163
Fun early papers on this were written about chess by the great Herb Simon (eg citeseerx.ist.psu.edu/viewdoc/downlo ) and by Adriaan de Groot (eg amazon.com/Thought-Choice ).
3
7
106
Those were for chess, but there's now a literature in many domains
A fave: Helga Noice & Tony Noice have extensively studied how actors memorize scripts. They wrote a book about it, but here's a lovely short paper with a quote from Michael Caine I love shafiknbahou.com/uploads/6/2/0/
4
17
214
The rough moral of all this work seems to be: intensively study (eg analysis and post-mortem) what you do; you will start to automatically see higher and higher level patterns ("chunks") if there are any to find; your memory gets better and better.
7
49
382
So what is happening with this chunking? Esp. re: yours and ’s spaces repetition thinking? It feels different?
QC goal(?): don’t let detail forgetting get in the way of going forward. Then you’ll probably chunk on your own?
What does chunk support look like?
1
Part of Simon’s theory was that chunks correspond to long-term encodings of higher-order phenomena. So maybe you can use SRS to accelerate the chunking process by rapidly committing higher-order structure to LTM.
2
3
I think without something like whole-task practice you might structure LTM in chunks that are semantically meaningful but have no correspondence to task usefulness.
1
1
1
Yes, this seems like a real failure mode, and some of my card clusters feel this way. I don’t yet understand what characterizes them vs successful ones.
Luck, or maybe one's prior knowledge? I don't know if research supports this, but expertise might develop as experience promotes useful chunking while useless chunking is discarded like a disproved hypothesis.