News for hackers. Stuff that matters.
Re:Scientist Finds PageRank-Type Algorithm from the 1940s by frisco on Thursday August 22, 2013 @02:11AM (#6255424)
What a silly article. PageRank is based on eigenvector centrality[1], which is one of a number of well-understood centrality measures that have been around for a long time. Google's innovation was noticing the importance of the link structure of the internet and knowing to apply EC to it - not inventing the graph theory underlying EC.

Edit: and knowing to apply EC to that graph structure.

[1] http://en.wikipedia.org/wiki/Centrality

Re:Scientist Finds PageRank-Type Algorithm from the 1940s by nullc on Thursday August 22, 2013 @02:17AM (#6255451)
What they patented, however, is more broad than that still:

1. A computer implemented method of scoring a plurality of linked documents, comprising: obtaining a plurality of documents, at least some of the documents being linked documents, at least some of the documents being linking documents, and at least some of the documents being both linked documents and linking documents, each of the linked documents being pointed to by a link in one or more of the linking documents; assigning a score to each of the linked documents based on scores of the one or more linking documents and processing the linked documents according to their scores.

Re:Scientist Finds PageRank-Type Algorithm from the 1940s by 6ren on Thursday August 22, 2013 @07:20AM (#6256413)
Patent claims usually begin with the broadest claim possible, then narrow it in following claims. If the first is found invalid, the other claims may still succeed. Thus, it makes no sense for the first claim to be certain to stand on its own; strategically, it should be over-broad.

From a computer science/mathematics point of view, this is also a convenient way to factor out commonalities of several claims, instead of duplicating them ad nauseam. For example, in the PageRank patent, claim 1 is immediately followed by six dependent claims giving further detail on it (as do ten other claims scattered through the rest): https://www.google.com/patents/US6285999

Thus, it is inaccurate to treat the first claim as "what they patented".

Re:Scientist Finds PageRank-Type Algorithm from the 1940s by throwawaykf02 on Thursday August 22, 2013 @10:29AM (#6257349)
While it's true that the independent claims try to be as broad as possible and dependent claims narrow them down, each claim is still evaluated -- and must be valid -- on an individual basis. And if infringement of a patent is to be determined, you'd start with the independent claims and work your way down, as those have the broadest scope.

So, yes, it does make sense to say that the patent claims what nullc quoted. It's only slightly inaccurate in that the patent also claims a lot more, but the independent claims define the broadest scope.

Re:Scientist Finds PageRank-Type Algorithm from the 1940s by 6ren on Thursday August 22, 2013 @10:41PM (#6261393)
Oh dear, you and nullc are right, and I am wrong. My apologies. This patent was granted, and so they do have a patent for claim 1 (and also for the other claims). I was confusing it with a patent application.

It's surprising to me that even the cursory prior art search of examination didn't turn up this idea, since it's so simple, doesn't use any of the meat of the invention described, and the idea of ranking by citation was well-known, e.g. for academic papers. I think applying that to hypertext links is clever and insightful... but not patent-worthy (like many other patents). Upon litigation, I think claim 1 might not stand. But regardless, the patent was granted, so they do have a patent on claim 1.

Thanks for taking the time to correct me.

Re:Scientist Finds PageRank-Type Algorithm from the 1940s by pudquick on Thursday August 22, 2013 @03:04AM (#6255573)
.. And yet still not invalidated by this prior art (if you were to stretch the concept) due to key phrases like 'computer implemented method'.

Computers in the 1940s were things like ENIAC and more likely to be involved in computing firing tables or cryptanalysis.

Now - if you were to look at Kleinberg's HITS instead you may possibly have a better argument, but I'm betting Kleinberg's prioritization method is still technically different enough than PageRank for that patent to stand. Given that Brin and Page actually reference the paper, I'm sure they've had patent lawyers make sure it wasn't infringing.

Re:Scientist Finds PageRank-Type Algorithm from the 1940s by mayank on Thursday August 22, 2013 @02:28AM (#6255480)
> Google's innovation was noticing the importance of the link structure of the internet - not inventing the graph theory underlying EC.

Actually, I'd say that Google's innovation was applying the eigenvector centrality computation to the Web graph, not discovering any particularly meaningful property (the "random surfer") of the Web graph.

I'd call it an extremely skillful application of an old algorithm to a new problem domain.

Re:Scientist Finds PageRank-Type Algorithm from the 1940s by spullara on Thursday August 22, 2013 @02:29AM (#6255484)
We shouldn't allow patents on old X applied to new Y domain.
Re:Scientist Finds PageRank-Type Algorithm from the 1940s by gfodor on Thursday August 22, 2013 @03:44AM (#6255716)
You just gave a fairly abstract definition of "innovation."
Re:Scientist Finds PageRank-Type Algorithm from the 1940s by walid on Thursday August 22, 2013 @11:19AM (#6257649)
Nope, @spullara just gave a definition of prior art :|
Re:Scientist Finds PageRank-Type Algorithm from the 1940s by gfodor on Friday August 23, 2013 @04:33AM (#6262353)
The point being the difference between "innovation" and "prior art" is a tricky one, since all innovation builds upon previous knowledge of some kind.
Re:Scientist Finds PageRank-Type Algorithm from the 1940s by wicknicks on Thursday August 22, 2013 @05:05AM (#6256002)
Exactly. Page and Brin cite some papers (or atleast Kleinberg's paper does -- which inspired Pagerank) which deal with ranking authors based on their citation graph. This article makes no sense whatsoever.
Re:Scientist Finds PageRank-Type Algorithm from the 1940s by floobynewb on Thursday August 22, 2013 @04:08AM (#6255811)
Damn you beat me to it.