Protecting against plagiarism may involve specific steps as explained below, in addition to action in accord with the discussion above in section ETD Guide:Universities - Plagiarism.
First, there is software available to detect plagiarism, or, in simple terms, similar copies. Stanford's SCAM software, and other software developed in Australia and other locations, discussed in the digital library literature and elsewhere, is available. This software can compare two documents, finding sections that are identical or similar (e.g., up to within simple substitution changes). If such software were to be widely run, it could ascertain for each new work if there were an ETD previously submitted that is very "close". While determined authors might defeat such software, more refined software could be developed to highlight cases where even such protection efforts were involved.
Second, it should be noted, as stated in section ETD Guide:Universities - Plagiarism, that if a work in made widely available, it is more likely that someone, seeing a new ETD that involves plagiarism, would detect that situation. In effect, as more and more ETDs are made accessible, there is a larger community monitoring abuses.
Third, since it is known that ETDs are often read by many more people than would read a paper dissertation, there is a strong psychological pressure to discourage plagiarism. One aspect of this is that students are aware of dire penalties if plagiarism is detected - they may be removed from their degree program and forever disgraced. Another aspect is that faculty working with students on ETDs, being aware that many might read the work, are likely to be more diligent than with paper works with regard to checking the validity and quality of results reported. In other words, it becomes more likely that faculty will carefully study ETDs that have their name on them as advisors or examiners. In particular, it is exceedingly unlikely that a student would be able to write about work they had not done, without that being detected.
In summary, detection by machines or people, and the threat of severe penalties, are likely to discourage students from even considering plagiarism with regard to ETDs.
Next Section: Naming standards: file names; unique Ids