The idea of the survey is to produce a map of the cave that is accurate, useful for route finding and possible connections (and look pretty). In most caves (and particularly in Kaninchenhöhle), enough information has to be collected for the full survey to be drawn by people who haven't visited all of it. The work involved in redrawing a huge survey each year is enormous, so it is becoming increasingly important to record enough so that a computer can draw it later.
This being expedition caving, time is limited, but don't forget that no-one else may ever come here again, so the idea is to efficiently record as much info as possible. It is best to record all your passage to high accuracy, but if constrained by time and manpower (eg. this is the last chance to survey, on the derigging trip of a cave which is now "finished"), it is better to record all the passage, rather than part of it to a very high standard. Particularly to be avoided is a survey that doesn't connect to the rest of the cave.
You are collecting data to fulfill a number of needs: the actual position of the passage for finding where it goes and possible connections; the shape of it for drawing pretty surveys; the location of possible leads for future exploration; geological info for working out how it got there.
For small caves the survey should be drawn up for publication by those who surveyed it. However, it is important to realise that for big caves (eg. Kaninchenhöhle) the survey will be drawn up by a small number of people, usually in Cambridge. They may not have visited the bit of cave which you survey, and they have no chance to go back to check anything which is unclear. The same applies to the passage description, which is a complex evolving document. Ambiguities in your description may not become a problem for several years, by which time you may no longer be in contact with expo or may not remember anything about the passage.
Surveying ethics. Remember: survey what you find - don't leave it for someone else. Your ability to find new passage without wasting time reexploring stuff seen before depends on those who came last year leaving good documentation. Likewise, future expeditions will be more rewarding if you have finished the job of exploring with a good survey.
Both on the surface and underground, CUCC tries to achieve a grade 5c survey and the information below is the minimum that can be collected in the cave (or on the ground) to achieve this.
LRUD is becoming increasingly important so we can use some of the fancy cave visualisation software which is now becoming available. For a discussion of how to record this data, see Andy Atkinson's article in Compass Points #10.
There are a whole load of hints and tips on how to do the above, written from experience in Austria.