Seoul Cafe Learning: A Snapshot — Epoche B2
Thirty seats, twenty-eight laptops, three o'clock on a Tuesday An iced americano in a café on the third floor above the Myeongdong subway exit costs 4,800 won, and the price includes the table for as long as one cares to keep it. At three on a Tuesday afternoon there are about thirty people in the room, every seat but two has a laptop on it, the place closes at eleven, and there is no evident mechanism by which anyone will be asked to leave before then. Korean has a word for the people this arrangement produces — 카공족 ( kagongjok ), roughly 'the café-study tribe' — and the coinage of a word is itself evidence that the phenomenon is settled rather than occasional. Four things in that room deserve explanation, and they turn out to be unrelated to one another: what the price is actually paying for; what the woman opposite is doing with a tablet and a paper notebook at the same time; why four students annotating one document have said exactly one sentence aloud between them; and why a room containing thirty people is quiet without being silent. The last of these is the one that unsettles, and it is best taken last. What 4,800 won actually buys The café is not really selling coffee. It is selling coffee bundled with an unpriced quantity of seat-time, and the bundle's economics can be worked out from the numbers already given. The revenue a seat earns is what the occupant paid divided by how long the seat was occupied. A customer who buys one americano and leaves after half an hour yields $\frac{4800}{0.5} = 9600$ won per seat-hour. A customer who buys the same americano and works for four hours yields $\frac{4800}{4} = 1200$ — one eighth as much. A customer who arrives at three and is still there at closing has occupied the seat for eight hours and yields $\frac{4800}{8} = 600$, one sixteenth. Those divisions assume one drink per person, which is what the room shows. Multiply out and the whole floor, thirty people at 4,800 won, takes 144,000 won for the afternoon; whether that covers rent above a Myeongdong subway exit is not something a postcard can settle. What the arithmetic does settle is the direction of the pressure: any café in this position has an incentive to raise revenue per seat-hour, and the levers are few and predictable — a posted time limit at busy hours, a minimum order per person, fewer power sockets, harder chairs. So why is none of that in force at three on a Tuesday? Because the relevant cost of an occupied seat is not its share of the rent but what the café loses by not selling it to somebody else, and at three on a Tuesday there is nobody else: two seats are empty, so the next customer can be served regardless. An occupied seat costs nothing until the room fills. It may even earn something, since a café with thirty people in it advertises itself to the street in a way an empty one cannot. The bundle is generous precisely where and when it is cheap to be generous, which is why the same chain will post a ninety-minute limit at the weekend. The woman with the tablet and the paper notebook The woman opposite has headphones on and is repeating English sentences under her breath into a tablet, then writing the same sentences out by hand in a paper notebook. It looks inefficient — two passes where one would do, and the slower medium chosen for the second — and there is reason to think both choices are deliberate and defensible. Saying the material aloud exploits what memory researchers call the production effect. Colin MacLeod and colleagues established the phenomenon in a series of experiments in which participants read some words aloud and others silently, and later recognised and recalled the spoken ones considerably better. Their account of why is that production makes an item distinctive: reading aloud adds a motor act and an auditory trace that the silently read items do not have, giving the memory system something extra to retrieve. The effect is robust and it is not subtle, which is a good reason to mutter into a tablet in a public café despite how it looks. Writing by hand is the more contested choice, and the essay should be careful here rather than confident. Pam Mueller and Daniel Oppenheimer reported that students who took notes in longhand outperformed students typing on laptops on conceptual questions about a lecture, and proposed the mechanism: typists can transcribe nearly verbatim, whereas the hand is too slow, so the writer must select, compress and reformulate as she goes — and it is the reformulating that does the learning. That is a real result and a plausible mechanism. It is also a result that has not held up cleanly: a subsequent replication and extension by Kayla Morehead, John Dunlosky and Katherine Rawson, using a larger sample, did not reproduce the advantage for longhand. The honest position is therefore narrower than the popular version of the finding: transcribing without processing is a poor way to learn, and handwriting is one way of preventing transcription, but the pen is not magic. One thing the woman is not doing is multitasking, and the distinction matters because it reverses the judgement. Dividing attention between two different tasks at the same moment degrades both. She is not doing that: she is encoding the same material twice, in sequence, through two different channels — articulation, then handwriting. Each pass is undivided. What looks like doing two things at once is a single thing done twice by different routes, which is the opposite of the failure it resembles. Four people, one document, one sentence spoken aloud Behind her, four students share three laptops and an e-reader and are annotating the same document simultaneously. Watching them for some time, one sentence has been spoken aloud: 'who deleted my highlight'. That single utterance is diagnostic, and the framework that makes it so comes from research on shared software. Carl Gutwin and Saul Greenberg, studying groups working together through shared digital workspaces, set