What you say is all correct, although I suppose people might be able to get at least a rough poll of voter preferences if they actually care. Assuming people don't know enough details about others' preferences to vote strategically is basically security by obscurity so I wouldn't want to rely upon it.
As a matter of methodology, I think we should separate the process of identifying problems in our voting system and gauging their severity from the process of looking for ways to improve it. This makes it much easier to stay objective. Warren D. Smith wrote some code to measure the performance of various voting systems (including various Condorcet systems) under various simulated conditions. Cheers, --Barak.
