Talk:One-time pad

This is an external article. See Talk:One-time pad/Permission.

Disagreement with external article

First, since the under the heading "bogus one time pads" was changed at CZ, the article is no longer completely external.

Second, while it is said that the invulnerability is "easily" proved, neither the proof is given here, nor is Shannon's proof cited. CZ doesn't insist that everything be sourced, but in a case like this, when sources are readily available, it seems reasonable to have it. Shannon is the original, but there are other reasonable sources.

Third, and speaking again to sourcing, give examples or citations, not just generalities about marketing and stream ciphers. In a few minutes of searching, I was able to find, admittedly with some shock that the patent was granted, "US Patent 6337910 - Method and apparatus for generating one time pads simultaneously in separate encryption/decryption systems", which is not generating one-time pads, but exchanging seeds for pseudorandom number generators. See http://www.google.com/url?sa=t&ct=res&cd=2&url=http%3A%2F%2Fwww.patentstorm.us%2Fpatents%2F6337910-description.html&ei=GwuVSNqrMYye8gTq9ZCvCg&usg=AFQjCNGvxb-Q4DbT1dwZVWc7j9eZLqMWJQ&sig2=I36umttpebshrYwg5TlFRQ

There is a quite interesting discussion in some joint lecture notes from MIT and UCSD at http://www.cs.ucsd.edu/~mihir/papers/gb.pdf, with Chapter 3 variously pointing out that some truly natural phenomena can be autocorrelating and thus weak one-time pads, and some arguments, which I want to reread in detail, about a claim that some pseudorandom number generators can be proven "that a generator that passes the next-bit test is perfect in the sense that it will pass all polynomial-time statistical tests."

I am concerned, as a Computers Workgroup Editor, that this article would not meet the CZ:About criteria for eventual approval of "authoritative, error-free, and well-written as encyclopedia articles are expected to be" The current version might develop into such, but it needs much work. Howard C. Berkowitz 21:00, 2 August 2008 (CDT)

Howard - given the current state of the United States Patent System, are you truly shocked? I know I'm not. America's patent system is in shambles - patents are granted left and right for work that has prior art that can be easily found, or the patent's name doesn't accurately describe what is being patented, as you've found. See [1] for more articles illustrating my points Eric M Gearhart

Such comments are sort of apropos to this article, as in my last few dealings with the Patent Office and our own counsel, decisionmaking seemed to be governed by random numbers. Going back to when I worked for a firm with in-house patent counsel, they didn't want to do the one for my team's development, because the stars at the lab next to corporate had gotten patents for what they insisted was the Only Way to do something. More recently, the venture capitalists were not going to give us a new infusion until we had certain biomedical monitoring patented. Ours had a very flexible, fault tolerant, distributed architecture that let the control centers be anywhere. The closest issued patent, which did much less for the patient, had hard-wired multidrop connectivity that could go a few hundred meters, but they called it packet-switched, said the packets could be moved by other methods "obvious to those skilled in the art". Our attorney said that once a patent was granted, the burden of proof that the "obvious" was clearly hand-waving, and we would be considered to infringe by using a networking technology, well known, but about 20 years ahead of theirs.

One-time pad decisionmaking, I supposeHoward C. Berkowitz 15:30, 5 August 2008 (CDT)

The possibility of pseudo-random methods that are adequately random

We currently have a section with that title. It contains good material, but I do not feel it belongs in this article.

Certainly there are such techniques; any stream cipher could be described as a pseudo-random method that is adequately random. However, they are not one-time pads, so I don't think they should be discussed here. Move that text to stream cipher? Somewhere else, like random number? Sandy Harris 22:10, 4 August 2008 (CDT)

I think, with appropriate caveats, it does belong here.

All pseudo-one-time pads, such as BBS, may not be snake oil. Rather than move the entire subject, it's more appropriate to have at least a brief note with a link. In general, I have a sense that your preference is to move something out completely, where mine is to have a summary with a link elsewhere. There should be a compromise.

I still want to get rid of it. I think everything that needs to be said here about pseudo-random methods is said toward the end of the "proof" section. Sandy Harris 14:58, 23 March 2009 (UTC)

Also when linking to a more detailed article, there are strong reasons to have wikilinks in the other direction. There are several reasons to do this. It may give ideas to a reader who came from another direction. It may suggest new articles. Also, the more cross-linked we are, the more likely we are appear high in a Google searchHoward C. Berkowitz 10:12, 5 August 2008 (CDT)

Yes, and stream cipher does link here. Sandy Harris 14:58, 23 March 2009 (UTC)

FAQ link

There's a good FAQ on this topic [2] from a well-known player [3]. I'd say it should obviously be linked to. However, I'm not sure if I already put it in once and someone deleted it or if I just overlooked it, so I'll ask here before adding it. Sandy Harris 03:12, 23 November 2008 (UTC)