For four centuries the Buddhist canon was carried in people's heads. Reciters specialised: one lineage held the long discourses, another the middle length ones, another the monastic rule, and the text was transmitted by group recitation in which a slip could be corrected on the spot by everyone else in the room. It is a robust system as long as there are enough reciters.
The Sri Lankan chronicles say there stopped being enough. Famine and war in the reign of Vattagamani Abhaya thinned the monastic community badly, and the elders of the Mahavihara concluded that a canon held only in trained memory could vanish inside a single bad generation. They assembled at Aluvihare, near Matale, and wrote it down, the discourses and the rule and the commentary with them, on palm leaf.
The decision changed what a canon was. Recited text is corrected by consensus in the moment and leaves no earlier state to compare against. Written text can be copied badly, and the bad copy survives beside the good one. Everything Pali scholarship does now, collating manuscripts, tracing how a reading entered a lineage, arguing about which of two words the compiler wrote, exists because there is a written trail at all.
The chronicles present the move as a response to danger rather than an improvement, and there is a note of loss in how they tell it. Something that lived in the community moved into an object the community keeps. That is a real change and the tradition knew it at the time.