5 ms·

Smaller standard error, significantly less memory, faster execution, equivalent rebalance cost... what's not to like? Thanks, Google: > Google has not applied

• nightcracker 12y ago
  Except that's useless. Realistically you'd need a legal patent clause, because right now Google can change its stance on patenting it at any moment, and make your code violate the patent.
  • noelwelsh 12y ago
    I was taught that patents can't acquired on discoveries that have already been made public. If that's the case Google's publication of this method before attempting to patent it makes it unpatentable.
    • btilly 12y ago
      According to http://en.wikipedia.org/wiki/Public_disclosure http://en.wikipedia.org/wiki/Public_disclosure there is a 1 year period before it becomes unpatentable. However it does seem unlikely that there will be a patent. :-)
    • DougMerritt 12y ago
      There is a 1 year grace period between disclosure and the deadline for patentability. http://www.uspto.gov/web/offices/pac/mpep/s2133.html http://www.uspto.gov/web/offices/pac/mpep/s2133.html
      • gojomo 12y ago
        Grace periods are shorter or non-existent outside the US. Publication date is June 9 of last year, so Google has less than 3 months to change their minds and turn-evil to pursue a patent on this technique.
  • gojomo 12y ago
    Not quite useless. The statement could prevent Google from later enforcing a patent against anyone who'd relied on that statement, via the doctrine of estoppel. And Google can't quite change their mind "at any moment" – more than one year after the first public offer/description of the technique, under US law, it won't be patentable. So by June 9, 2015, if they haven't patented it by then, they won't be able to. (It's likely already too late to patent in other jurisdictions, that don't have such a long public-description grace-period.)
• bch 12y ago
  > Google has not applied for patent protection for this algorithm, and, as of this writing, has no plans to. Rather, it wishes to contribute this algorithm to the community. What they should do (IANAL) is patent it and license it properly. This way it's protected (patented), but the intent is clear and future-proof instead of leaving it in an ambiguous legal gray area.
  • rakoo 12y ago
    > This way it's protected (patented) Protected from what ?
    • HarrietJones 12y ago
      Protected from me patenting it and then suing small companies that use it.
      • dalke 12y ago
        You can't patent something that someone else published.
        • warfangle 12y ago
          funny joke ;)
          • dalke 12y ago
            ... any more than you can patent something which has already been patented. Both require a lawsuit to challenge.