Many of us put our ideas in github or articles, hoping they will be just used.

However, as I have currently a tough lesson, if your idea turns out successful, you can count on patent vultures trying to monopolize it or its basic applications, variations, consequences.

How to protect from it - safely share your ideas? How to defend when protection has turned out insufficient?

In theory, making your idea public makes it "prior art" and should prevent patenting. In practice, from my experience USPTO examiner just checks US PATENTS for KEYWORDS, often with no understanding - making it extremely simple to smuggle something one has found on the internet.

The first big problem is finding out about such attempt - and it seems nearly impossible to find if they don't cite you, don't use some characteristic keywords?

If you have found it and it is not too late, you can submit e.g. third-party observation (good tutorial for USPTO).

In my situation there at least two ongoing attempts for basic application of my ANS entropy coding. One is in combination with the most basic statistical modelling (Markov): that probability of a symbol depends on the previous symbol. It was explicitly mentioned in my papers, there was much earlier implementation, used in popular CRAM DNA compressor (complaint by its author) ... but still literally description of this DNA compressor was recently given second Notice of Allowance by USPTO. The second attempt is by Google trying to monopolize image/video compression application of ANS, even though I have originally suggested it them and helped them in this adaptation for three years through their public discussion forum (my complaint).

What are other options while defending your work from patent vultures?

Update: With help of many like Ars Techica and EFF, the Google patent was rejected. Nice page with stories and materials for people in such situation: Patent Pandas.

  • The last claim amendment that apparently overcame the examiner's concerns added a limitation involving an escape code to change to a different probability table. There is quite a bit of prior art on the record in this case including several non-patent documents. I see, in particular, that the applicant cited one your papers from 2013/2014 as well as a 3rd party observation from the U.K. Is this case a good example of the system working properly or not properly.
    – George White
    Commented Sep 19, 2017 at 18:30
  • @GeorgeWhite, you are right - I have taken a closer look and situation is a bit more complex: this patent covers simplified behavior of e.g. 33-year old well known PPM compressors: while the original use varying order and escape codes to switch between multiple probability distribution tables (also using large alphabet entropy coding: originally Huffman or arithmetic, for which ANS is now commonly used as alternative), this patent application uses fixed order and switches between two tables.
    – Jarek Duda
    Commented Sep 21, 2017 at 11:19

1 Answer 1


One hopes that should Google actually obtain a patent in this area, it will be a minor "improvement" that should be easy to circumvent. I don't see how they can block you and others from practicing your technology as you have already implemented it. As for your original question of how to prevent such patents in the future, one approach would be to file for a patent yourself. You don't need to actually see it through to a grant but you might want to. Then, it becomes prior art that is within the various patent offices systems. Since you are an academic, your institution may be willing to pursue this for you. If you do obtain a patent, you can grant licenses on any terms you want including free. In any case, I would work with the Electronic Frontier Foundation to oppose any such bogus patents in the future.

  • Google patent is extremely broad - written to prevent competition from using ANS in image/video compression. It's just straightforward adaptation: takes textbook method and replaces coding with mine, especially for transform coefficients - exactly as I have written them in their public forum 2 years earlier. They didn't disclose this direct communication in the original application, but was pointed by ISA reviewer (plagiarism): th.if.uj.edu.pl/~dudaj/ISAopinionGoogleANS.pdf Could you maybe elaborate on EFF and other organizations? How they operate and which to choose?
    – Jarek Duda
    Commented Sep 19, 2017 at 21:55
  • 1
    @JarekDuda I have no personal experience, but the EFF is very active fighting bad patents and they have an experienced legal staff. As for Google's application, claims in applications are often very much broader that what gets granted. Hopefully if anything gets granted it will be narrow. Based on your published work, it would be very hard for Google to exercise the patent even if it is granted.
    – Eric S
    Commented Sep 19, 2017 at 22:03
  • 1
    @JarekDuda As I said, I have no personal knowledge of working with any of these organization. That said, I did find a few more possibilities: openinventionnetwork.com, pubpat.org, ffii.org. Ironically I used Google to search for them.
    – Eric S
    Commented Sep 20, 2017 at 14:57
  • 1
    @JarekDuda As a last suggestion, you may want to get some press coverage. I would recommend contacting arstechnica.com or slashdot.org.
    – Eric S
    Commented Sep 20, 2017 at 18:37
  • 1
    @JarekDuda Arstechnica is widely read in the tech community.
    – Eric S
    Commented Sep 20, 2017 at 20:07

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .