treating the tape as tracks

Imagine ruling your single line of notebook cells into a few horizontal lanes, so each cell now has a top half and a bottom half (or three rows, or more). One lane can hold the original data while another holds your scratch notes lined up directly beneath it. This is the tracks technique: a way to make ONE tape behave as if it had several parallel rows that the head reads all at once.

The trick is purely a relabeling of the tape alphabet. To get two tracks, you let each tape symbol be a PAIR, like (a, X), meaning 'a on the top track and X on the bottom track of this cell'. The head still reads one cell at a time, but reading the cell (a, X) gives it both track values together, and writing replaces both at once. For three tracks you use triples, and so on. Because the alphabet stays finite (you just have more composite symbols), this needs no new machinery: it is still a perfectly ordinary single-tape Turing machine, only with a richer alphabet.

Tracks are a workhorse design technique. A second track is the natural place to keep a counter, a marker showing the head's 'home' position, or a running copy of part of the input, all kept in lockstep alignment with the data on the first track. It is closely related to, but cheaper than, having genuinely separate tapes: a multitape machine has independent heads, whereas tracks share one head and so move together. The honest framing: tracks add no computational power (a plain one-track machine can do anything a multi-track one can), they only make a machine easier to design and describe.