As expounded elsewhere in this blog, I built a database of full-text U.S. patents and patent applications and some search tools for doing some data mining in it. As described in this post, I used it to extract a corpus of patents and applications that appear to represent in-house inventions of Intellectual Ventures, and organize them by subject matter.

Intellectual Ventures (“IV”) is the intellectual property powerhouse founded by Nathan Myhrvold after he left his position as Chief Technology Officer of Microsoft. IV receives a lot of criticism in the press and the blogosphere, often being portrayed as the mother of all “patent trolls” (a pejorative term originally popularized by IV co-founder and vice-chairman Peter Detkin, who was in house at Intel at the time). IV owns a lot of patents — according to its own published figures, it has acquired 70,000 patents since its founding, and filed 3,000 patent applications on the inventions of its own team of inventors. (Not everyone thinks this is a good thing.)

No one knows exactly what technologies are covered by all those patents. Like most patent aggregators, IV usually sets up corporations or LLC’s to hold IP rights that it acquires. Current law does not require buyers or sellers of patent rights to disclose the identities of the “real” owners — presumably whoever owns or controls the corporation holding the patents. This was briefly a matter of intense debate, with no less than that pinnacle of patent scholarship, the White House, chiming in with an executive action. In early 2014, the U.S. Patent Office proposed “real party in interest” rules that would have required disclosure of the “real owners” at least while an application is pending, but the PTO has now decided to hand off that particular hot potato to Congress.

It’s much easier to figure out who owns what with “in-house” inventions, however, because U.S. law does require the inventors to be named. So it turns out to be quite feasible to get a fairly good picture of what technologies IV considers important enough to set their network of inventors to work on. One of the ways that IV adds to its IP portfolio is by filing on ideas generated via brainstorming sessions with a carefully selected group of experts. Of course, patent applications are typically not published until 18 months after filing, so there’s an inherent time lag, but it nevertheless seemed interesting to inquire what fields of technology IV considers compelling enough to spend their own patenting budget on.

Such questions are, or should be, of considerable interest to anyone hoping to make money by invention things. One of the most important challenges that inventors face is figuring out which problems are worth working on. Ray Kurzweil, by any measure one of the smartest and most successful inventors around, puts it this way in his book The Singularity Is Near:

l realized that most inventions fail not because the R&D department can’t get them to work but because the timing is wrong. Inventing is a lot like surfing: you have to anticipate and catch the wave at just the right moment.
My interest in technology trends and their implications took on a life of its own in the 1980s, and I began to use my models to project and anticipate future echnologies, innovations that would appear in 2000, 2010, 2020, and beyond. This enabled me to invent with the capabilities of the future by conceiving and designing inventions using these future capabilities.

To identify the IV-originated patents, I began with the one person who I could be sure was an IV inventor: Nathan Myhrvold. From my database of full text patents from 2007 through 2013, I assembled a corpus of 1,686 applications and 734 patents in which either Nathan Myhrvold appears as an inventor or at least two of his 10 most common co-inventors appear as inventors, or that are assigned to any of the companies appearing as assignees on any of the foregoing. (It turns out IV does not appear to use different assignee companies for most of their in-house filings — by far the most common assignee shown was Searete LLC and its various misspellings).

I then clustered the titles by keyword, removing duplicates, and followed up with some rearranging by hand to classify the patents into categories by subject matter.

(This is all another example of a task that is done fairly easily with some simple python scripts on a full text database, but that would be somewhere between seriously tedious and impossible on a typical patent search utility. Just saying.)

The resulting data is here. It consists of a text file (all_titles_by_category.txt) listing only titles, which are grouped by subject matter categories; and a text file for each subject matter category, containing the title, abstract, number, and other metadata for each patent or application in the category. The categorization isn’t perfect — I didn’t want to make a career out of this, so I went mainly by title, and anything I couldn’t categorize by title (or in a few cases abstract), went in the “miscellaneous” category. But my impression is that it provides a reasonable picture of where IV was concentrating its efforts during the time period in question.

Apart from the miscellaneous category, which contains 370 titles, the largest categories (in descending order by number of titles) were:

  1. Nuclear reactor related — 81 titles
    representative title: System and method for annealing nuclear fission reactor materials
  2. Substance delivery — 74 titles
    representative title: Acoustically controlled substance delivery device
  3. Frozen particles — 68 titles
    representative title: Compositions and methods for administering compartmentalized frozen particles
  4. Vehicles — 63 titles
    representative title: Hybrid vehicle qualification for preferential result
  5. Flexible electronics — 62 titles
    representative title: E-paper display control based on conformation sequence status
  6. Virtual world — 62 titles
    representative title: Disposition of component virtual property rights
  7. Power — 62 titles
    representative title: Optical power transmission system and method having co-propagating control signal
  8. Computational systems — 61 titles
    representative title: Computational systems and methods for health services planning and matching
  9. Blood and lymph — 57 titles
    representative title: Device, system, and method for controllably reducing inflammatory mediators in a subject
  10. X-ray related — 44 titles
    representative title: Ionizing-radiation-responsive compositions, methods, and systems

 

 

Leave a Reply

Your email address will not be published. Required fields are marked *