Given that one can collect a paper ticket anywhere, I'm confused at how this actually helps.
Several ways:
For one thing, when you've just made the change the person will have been used to getting an e-Ticket and will turn up at the exit barriers before realising that they've got a ToD code. At this point the RPI can pounce: "Now Sir, can you tell me why you've arrived at Cardiff Central with an uncollected ticket from Cardiff Queen Street, despite Queen Street having barriers?" Anyone who has genuinely come from Queen Street will have printed their ticket before entering that station (and let's face it, who would do Queen Street to Central as an isolated journey rather than walk?)
Report to the digital fraud team, they request the purchase history from the retailer and can see the journey from A to B and the journey from Y to Z. Using the footage captured by the RPI they now have a face to look for on the CCTV to prove that the same person did make both journeys (with the timings at entry and exit station conveniently coinciding with the journey time by train, funny that) rather than having bought one of the tickets for a friend or relative.
Alternatively the fact that someone collected the ToD at a different station to the origin of the journey (whether it be a TVM on the inside of the barriers at the destination, or somewhere completely different) can be flagged up to the digital fraud team who can then look at the purchase history for suspicious patterns (funny how the station the ToD from Y to Z was collected at just happens to be "A", with a journey from A to B made immediately after). Obviously collecting a ToD at a different station from the one booked is usually perfectly legitimate, but we're talking about flows almost no one would make as an isolated journey by train, such as Cardiff Central to Cardiff Queen Street.
I must admit that I'm not entirely clear why it's better than automatically flagging up eTickets between Y and Z, but it's produced results so who am I to argue?