Version control where every single version is guaranteed to be 100% reproducible with 0 effort from any human on 100% of machines.
The core concept of basic/standard version control is:
changes to documents are identified by incrementing an associated number or letter code, termed the "revision level", or simply "revision".
- ..
At the simplest level, developers can simply retain multiple copies of the different versions of the program, and number them appropriately. This simple approach has been used on many large software projects. Whilst this method can work, it is inefficient (as many near-identical copies of the program will be kept around), requires a lot of self-discipline on the part of developers, and often leads to mistakes. (http://en.wikipedia.org/wiki/Version_control)
Note that the concept only explicitly describes identification of versions! This is important because although many other things are implied in the concept, most of those have tended not to be implemented in the common and standard version-control systems; over time, there is a tendency for them to implement more of the implied features, but this is not a monotonic increase (there are many cases of "two-steps-forward, one-step-back" - and even "one-forward, one back").
"Perfect" version control seeks to augment this by placing a concrete stake in the ground to say explicitly what will be supported; there may be different incompatible parallel ways of achieving this, but you can be sure it does achieve it.
It is important to note that the word "perfect" is used to indicate this is a theoretical ideal - it is being used in the academic context rather than the common usage of the word. It is mainly helpful for letting us make comparitive statements about different version-control systems and explicitly describing them in ways that are directly-comparable, rather than being forced merely to list large numbers of features, without making it clear in a simple manner what the user can expect in terms of end-benefits.