|WikiProject Cryptography / Computer science|
|WikiProject Mathematics||(Rated Start-class, Mid-priority)|
I agree with the comment that the article suffers the defect of being too mathematical at the expense of intuition and, on that account, perhaps only mathematicians have a chance to understand it. I invite the commentor to lend a hand to clarifying the meaning and importance of key concepts and also point out what are mere assertions. Skinnerd (talk) 14:17, 9 April 2013 (UTC)
Most of the Math in the article attempt to distil rather simple concepts into a mathematical domain. This make the article difficult or confusing for all but mathematicians in the field. On the other hand: the meaning and importance of the mathematical key concepts -- and their difficulties -- is not discussed or even mentioned. The article also bravely states some unprovable facts, that don't have any meaning in the real world. The interesting part is the difference between what has been proven, what is believed or assumed to be true, and how things work in practice. (End note from TRNG98.se)
This page is very badly written and there are several mistakes and inaccuracies in it. An expert in the field should have a look at it.
Attempted to Improve
I have tried to improve the wording without changing the math (which I don't consider myself qualified to do). My edits reflect my best understanding as a non-expert who nonetheless has some slight acquaintance with the theory of pseudo-random number generation. I note that a previous contributor has said the article contains several mistakes and inaccuracies. It may be the case that some of these have been taken care of in rewording. If not, they need to be attended to if this article is to stand alone. I also note the article's use of the term obliviate; this term needs to be defined and I am not knowledgeable enough to do it. Also a better job needs to be done of defining what is meant by error as used in connection with an extractor. Skinnerd (talk) 20:44, 23 October 2012 (UTC)
I also added a reference to extractors as unbiasing algorithms because this is a term that is encountered in some (older?) literature. Cited reference to Gifford. Skinnerd (talk) 23:58, 23 October 2012 (UTC)
To make this article more clear all of the symbols in the following definition should be defined:
Definition (k-ERF): An adaptive k-ERF is a function where, for a random input , when a computationally unbounded adversary can adaptively read all of except for bits, for some negligible function .
RFC 5869 and the related paper (http://eprint.iacr.org/2010/264) seem relevant here. The paper indicates that it is important that the extraction and expansion are independent. I'm not qualified to comment on the quality of the work, but if both the paper and my understanding thereof is correct, then this independence is important to someone implementing an extractor (e.g., one shouldn't use SHA-256 both to extract and to generate output, or use HMAC-SHA256 with the same key for both operations--or, for that matter, use an unkeyed construct like SHA-256 for extraction at all). Wecotanoxa (talk) 12:17, 10 October 2013 (UTC)