Augmenting Long-term Memory

Memory is one of those cognitive capacities that educated people have been encouraged to dismiss as a lesser skill since roughly the 1960s — the age of open-book exams and the assertion that knowing where to look something up is better than knowing it. Nielsen's essay is an extended, empirically grounded attack on that consensus, and it reads differently from most self-help pieces about flashcards because the author is a serious scientist who has been embarrassed by how much better Anki made him at understanding things he thought he already understood.nnThe first key argument is the chunk theory of expertise. Drawing on Adriaan de Groot and Herbert Simon's chess research, Nielsen explains why expert chess players see positions differently from beginners: they perceive large, meaningful chunks (constellations of pieces with known implications) rather than individual pieces. Simon estimated world-class players had learned between 25,000 and 100,000 such chunks. Nielsen's point is that chunking is essentially a form of elaborated long-term memory, and that having more chunks in any domain is functionally equivalent to a higher-capacity working memory in that domain. Anki accelerates the accumulation of chunks.nnThe second and more practical move is the AlphaGo case study. Nielsen needed to understand a cutting-edge deep reinforcement learning paper well enough to write about it for a general science audience. He describes his multiple-pass process: skimming for key terms and easily captured facts, making Anki cards at each pass, gradually building up background context until a final thorough read, by which point the paper was substantially less difficult than it would have been cold. A year later, when DeepMind released AlphaGo Zero, Nielsen found he could read the follow-up paper in under an hour — he had retained the essential understanding without effort. Conventional note-taking would not have produced that result.nnWe picked this because it is the single best argument for taking memory seriously as an adult learning strategy — useful for researchers, writers, students, and anyone who wants to genuinely understand things rather than having once understood them. The question worth sitting with is how many domains you currently 'know' that you would struggle to demonstrate understanding of at an intermediate level without looking things up.