I say yes–and I suspect the yes may be significant. In uncovering the problems with the University of Washington’s reporting of its startup activity for the past two years, another issue has surfaced. Let me explain.

The University of Washington claims SNUPI as one of its flagship startups. And, yes, SNUPI has raised $7m in investment and appears to be a happy company. Here is the UW’s account of SNUPI (from December 2012):

SNUPI Technologies [broken link–now WallyHome], a new UW spin-out company that takes advantage of the wiring already in homes to create a wireless sensor network, launched today with $1.5 million in funding from Madrona Venture Group, Radar Partners and the founders. The technology, also called SNUPI (Sensor Network Utilizing Powerline Infrastructure), was developed at the University of Washington and Georgia Institute of Technology by Professors Shwetak Patel and Matt Reynolds and others.

Co-founded by Jaech, Patel, Reynolds and UW graduate student Gabe Cohn, the company will use the financing to build the team, and to design and build the hardware and software implementations. SNUPI Technologies is based in Seattle.

It’s just that SNUPI appears to have its roots in research conducted at Georgia Tech. Here is an excerpt from a GeorgiaTech story [the story apparently has been disappeared from the web]:

Georgia Tech at the Forefront of the Internet of Things

Tech is at the forefront of the IoT trend with the development of WallyHome. The technology was first researched at Georgia Tech’s Aware Homeby College of Computing Professor Dr. Gregory Abowd, and his PhD students Shwetak Patel and Erich Stuntebeck. While working on power line-based technologies, they discovered Wally’s underlying wireless technology. WallyHome uses a home’s electrical wiring to detect environmental hazards. It monitors moisture, temperature and humidity changes. Sensors are placed with appliances or in hazard-prone spaces and Wally alerts users of impending disaster or damage as soon as it occurs. (WATCH: See how WallyHome works)

“We use a completely different approach to wirelessly sending data from our sensors to a central hub,” said Patel. “By using the electrical wiring in your home as a large antenna, this allows us to dramatically reduce the power consumption of the sensor while increasing the range. Consequently, we can produce wireless sensors that have whole-home range and last 10+ years.”

After receiving his PhD in Computer Science from Georgia Tech, Patel became the first Tech alum to win a MacArthur Fellowship. He also started two companies using Georgia Tech research, including SNUPI Technologies, the company behind the WallyHome system.

******

There are now five issued patents at Georgia Tech involving Abowd and Patel.

Patent            Assignee
8,788,191      GTRF and UW
8,494,762     GTRF
8392,107      GTRF
8,334,784     Belkin
8,094,034    GTRF

There are also multiple pending applications, some of which have been published. One appears to also be filed with Belkin (same law firm as the issued patent). One may be filed jointly by Georgia Tech and UW. UW’s inventor Patek Shwetak figures in a number of other published applications. One set goes to Belkin, one set to Microsoft (now issued, and remaining applications to either UW or Georgia Tech. What is not clear from the applications and issued patents is how UW and Georgia Tech are managing the later patent work. No doubt there are other patents as well, depending on how the platform develops.

As an aside, it is challenging to get a picture of what is going on–applications with the same title have different claims and assignment paths, the inventor teams vary, two universities are involved, developing different lines of claims, there may be an inter-institutional agreement in place between UW and Georgia Tech, and both UW and Georgia Tech apparently have done deals with SNUPI, but what inventions are involved, with what scope of rights? What improvements are also obligated? What software, data, and other non-patent stuff is also involved, and how has that been handled by UW and Georgia Tech?  What about patent applications that have been filed but are not yet published? All we get from the universities is publicity–not a clear statement of where they stand. It is all but impossible to construct a clear account of where things stand. When faculty were not filing patent applications willy-nilly, things were much clearer: published stuff was in the clear, without institutional proprietary claims and prospective monopoly positions.

The point here, however, is that both UW and Georgia Tech appear to claim SNUPI as a startup. If both include SNUPI in their count of startups for FY13, then anyone adding UW’s and Georgia Tech’s startup numbers together will get an inflated count, since SNUPI would figure twice. Similarly, a string of patent applications with both institutions providing co-inventors also would be counted twice. Without a list of disclosure titles, application numbers, and patent numbers, and without a list of startup companies, it is impossible to add any AUTM survey figures for invention disclosures, patent applications, issued patents, or startup companies together to arrive at total activity counts across institutions.

Consider Ennaid Therapeutics. UW counts Ennaid as a startup for FY 14, without apparently a shred of evidence for doing so. Ennaid reports licensing technology–apparently core–from Tulane and Rockefeller universities. Within a year it is licensing more from Florida Gulf Coast. Do these universities count Ennaid as one of “their” startups?

How widespread is duplicative reporting? How many inventions are reported to universities with joint inventors from other organizations? How many patent applications filed? Patents issued? Companies started? It’s hard to say. But I bet it is over 10%, and maybe even as high as 20%. Once you slip licensing so that it does not have to involve a specific patent family but can count each divisional, CIP, continuation, and foreign application as a new license, there can be a heap of duplication.

Given the fragmentation of university research–many teams working in the same areas, often encouraged by government funding to collaborate–and given the desire by university administrators to report big numbers representing “potential” economic impact, it is easy to see how there could be duplicative reporting. Of course, the omission of backing data for applications, patents, and startups conveniently works in favor of higher counts. No one looking at AUTM survey data would recognize the duplicate activity. Even AUTM’s detailed guidance for what counts as a startup does not exclude multiple counts. Every university willing to rationalize that a startup was “formed specifically to license and develop the technology being licensed” will count it as their own.

Notice the ambiguity in AUTM’s guidance: the “technology” can be something broad, such as “using house wiring to detect what electrical equipment is on.” The company can be started to license this “technology.” And any number of universities may have inventions that fit within this definition. The AUTM guidance does not require that the startup company has to know about the inventive work at any given university when it starts; thus, any number of universities could be in line to claim the startup as their own. AUTM does not even limit the startup in time. The company could form years before the license, as far as AUTM is concerned. What AUTM does not say–but should have–is that the company should be counted in the fiscal year it forms, not the year in which it obtains the license (and it may never get a license).

The use of “technology” is itself suspect. The expected scope would be “patentable invention owned by the university or its agent.” And one would expect that the license is exclusive. Otherwise, any new company acquiring from a university non-exclusively, say, a software program under license, or a biomaterial under an MTA, or a data set could be classed as a “startup” for the AUTM survey. That is, universities would then be counting every new company that they had touched with some tool or service. While that may be nice, it’s hardly supportive of a rhetoric of startups created to develop a specific invention made in university research. Once the definition is sloppy, then folks willing to game the system to pump up their own status, like those at the University of Washington, are only too ready to do so.

But it would appear that AUTM data, if combined without removing duplicates, overstates activity. AUTM appears to have no quality controls in place to audit incoming survey information. AUTM puts the compiled data behind a paywall. AUTM does not require sufficient information (patent application numbers, issued patent numbers, names of startups) so that others could remove duplicates before assessing overall activity.

It appears that AUTM licensing survey totals overstate the number of patent applications, issued patents, and startups–perhaps by 20%. AUTM survey information is used to influence public policy. Here is the summary of AUTM’s 2012 survey. There is no indication that AUTM has removed duplicates. Indeed, their survey is set up so it would be impossible to find duplicates in their data. As the government encourages collaboration and offers more grants to fund multi-institutional centers, AUTM’s survey numbers will include more duplicates. That is, AUTM’s licensing survey could claim growth in activity when the underlying fundamentals are actually contracting, with each university counting disclosures, applications, and issued patents with joint inventors from other institutions, as well as counting as their own startup every newish company that happens to have done a deal with them for “technology” that the company was formed to develop, and there for license.

The conclusion: AUTM’s licensing survey overstates national tech transfer activity–by how much, we don’t know, and likely will never know. Clearly, however, the survey is defective and AUTM should take action to correct things.