Research is an extremely cyclic process.
Later stages might necessitate a review of earlier work.
This isn’t a weakness of the process but is part of the built-in error correction machinery.
Because of the cyclic nature of research, it can be difficult to determine where to start and when to stop.