> So, just to play the devil's advocate: are NPEs by definition patent trolls? I can't think of a counterargument, but maybe someone else can?

There are lots of counter-examples. E.g. Mojave Aerospace Ventures (MAV): http://en.wikipedia.org/wiki/Mojave_Aerospace_Ventures. It's the patent-holding company that owns all the patents for Spaceship One. It's a vehicle to intermediate between the major investor, Paul Allen, the inventor, Burt Rutan and Scaled Composites, and the various companies that will be commercializing the technology, namely the Virgin Group and any special-purpose joint-venture entities created by it.

MAV (and ARM, mentioned in a sibling comment), is a great example of why NPEs exist.[1] There is a lot of value in being able to take the products of expensive R&D, like the Spaceship One effort that won the X Prize, and being able to package that into a set of property rights owned by a holding company that can transact in those rights. It allows a division of labor that's very hard to achieve otherwise. And division of labor is a good thing, because it allows everyone to focus on their core competency.

For example, in the case of Mojave, you have the separation between the people doing the R&D (Scaled Composites), the people bankrolling the effort (Paul Allen and other investors), and the people doing the commercialization (Virgin, etc). In the case of ARM, you have a separation between the people doing the R&D (ARM), the people making SOCs using the basic cores (NVIDIA, TI, AMD), the people manufacturing the chips (Samsung, TSMC, GF), and the people using the chips in finished products. It's economically valuable to facilitate this sort of separation.

[1] Indeed, MAV is a better example than ARM, because it's purely a holding company. It does neither the research nor the commercialization, but rather exists to facilitate the involvement of the independent investor in the whole process.

Granting your point about non-trolling benefits to NPE corporate structures, doesn't this also give MAV the ability to troll while protecting the assets in Spaceship One and the other corporate structures responsible for the actual development? In other words, legal action taken by MAV to enforce their patents would still seem to fit the colloquial usage of "patent trolling".

Also, universities. Relevant paper:

"Are Universities Patent Trolls?", Mark Lemley, 2007

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=980776

From the abstract:

>Finally, I think we can learn something about the raging debate over who's a patent troll and what to do about trolls by looking at university patents. Universities are non-practicing entities. They share some characteristics with trolls, at least if the term is broadly defined, but they are not trolls. Asking what distinguishes universities from trolls can actually help us figure out what concerns us about trolls. What we ought to do is abandon the search for a group of individual companies to define as trolls. In my view, troll is as troll does. Universities will sometimes be bad actors. Nonmanufacturing patent owners will sometimes be bad actors. Manufacturing patent owners will sometimes be bad actors. Instead of singling out bad actors, we should focus on the bad acts and the laws that make them possible.

I think the canonical example of a NPE is ARM. They don't fabricate their own chips, but they license out their designs.

'Practicing' doesn't mean manufacturing physical objects. ARM spends money producing things its customers value: processor designs at various levels, and software.

Given that the Acorn RISC Machine came out 28 years ago, I'm pretty sure someone could build a patent-unencumbered processor for that instruction set if they really wanted to go it alone.

ARM is not an NPE because many of the designs they license are copyrighted as well as patented. If all of their patents were struck down they still would have a viable business, although their revenue would be significantly reduced.

Many software companies have business models similar to ARM's, yet we don't call them NPE's.

That's true, but I think you can also differentiate between ARM and the more trolly NPEs in the basis of how much ongoing original work they do. Not all trolls are NPEs either. Look at Apple. I wouldn't say the whole company is a troll, by any means, but they've certainly done some trolling.

By that definition even Qualcomm (top 5 semiconductor company) is an NPE as TSMC fabricates their chips. ARM licenses both RTL (code) and GDSII files (the files that a foundry such as TSMC uses to freate masks for fabrication) just as you would license software. They can hardly be called a NPE.

I would say no, a NPE becomes a patent troll when rather than publicizing and trying to license the technology they developed to companies that might find it useful they wait for their developments to be independently invented or unknowingly copied and then sue the infringer. Or they patent a technology used by another group but which hasn't already been patented, and then sue the inventor (which has happened to groups I've been a part of twice).

In general it's hard to come up with useful inventions without a concrete problem, so often good NPEs are in communication with their potential customers even before they start work.

They arent definitionally "trolls" but they cannot be Using the patents (deriving value from their being patents in the fist place). So it's still a useful metric.