I understand what a replay attack is, but I keep reading about a pre-play attack. What is it? Is it when someone intercepts an unused message and blocks the sender so they can't finish using it, and then the interceptor tries to "play" the message as if they were the original sender?
Computer-Related Risks by Peter G. Neumann classes preplay attacks as part of the "Playback attacks" category of risks and defines it as "preplay of a not-yet used message", to help understand this a little more I have included two real world examples.
Web Application example; I would argue that a CSFR (Cross-site request forgery) attack is a form of preplay as you do not have to capture the user’s traffic, but instead are using their own session to play a valid sequence of activity.
Kerberos example; An example of this would be if the attacker were able to compute a valid response to a kerberos challenge and send the expected reply before the real user. This could result in the attacker gaining access to the server within the context of the valid user’s rights.
So a pre-play attack is where an attacker is able to guess or compute a valid response to and send the reply before the valid user can. It is different to a replay attack, where the attacker would capture the whole challenge and be able to replay at a later stage.
You can look for some information about preplay in book Protocols for Authentication and Key Establishment (page with preplay mentioned here). One example of preplay is Burmester triangle attack - unfortunately I didn't find much information about that.
Edit - quote from the book:
Preplay might be regarded as a natural extension of replay, although it is not clear that this is really an attack that can be useful on its own. An interesting example of an attack that employs preplay is so-called triangle attack of Burmester which will be presented in Chap. 5