The use of that notation implies the number of classes was proportional to number of operations (+/- constant factor), not that it was equal to number of operations. O(n) works both if there's 1 class per operation, and if there's 5 classes per operation.
Frequently when writing O(n), the writer is making a comparison. e.g. "The GC was O(n log n) wrt memory, but now it's just O(log n)." That expands to a lot more text.
Or, to put that another way: there is a constant cost in human parsing complexity when writing "O(n)", but the number of parse-nodes in the text when using "O(n)" only increases as O(n), whereas without "O(n)", it increases with O(n^1.3). ;)
