The night the orchestra got smaller and sounded better
The community hall smelled of dust and brass polish. The conductor checked the clock. One rehearsal slot left. There was room to squeeze in more players, but that would steal minutes from the music itself.
For ages, the instinct was simple: want a bigger sound, bring in more musicians. In the world of text-writing computer systems, people did the same, piling on more inner parts, but not giving them much more reading practice.
A few builders treated rehearsal time as fixed, like a hall booking you can’t stretch. They turned two knobs instead: how many musicians, and how many pages to rehearse. In the computer version, that’s system size and how much text it reads while learning.
They kept swapping players for pages and listening for what worked. The pattern was steady: too few musicians and the sound stayed thin; too many and the group never locked in. Best came from balance, not going all-in on size or practice.
One detail tripped people up. The rehearsal plan has to fit the time you’ve actually got. If you plan as if you’ll rehearse longer, you’re still pushing hard when the slot ends, so the final run-through is scrappy. The same kind of mismatch can blunt a text system’s learning.
On concert night, they tried the opposite of showing off. Fewer musicians, but far more pages rehearsed, all within the same total effort. The smaller group played tighter, and it was cheaper to keep running later. Still, better practice didn’t magically stop ugly, biased blurts.