Raw Notes from the CRA Workshop on University Software Licensing, Patents, and Industrial Interaction

(Held 9 December 1996 at the Embassy Row Hotel, Washington DC)

Overview Presentation on the Workshop by Randy H. Katz

Session I -- Where are We Today?

Session focus was on the current state of the practice in university-industry interaction and agreements.

Specific issues that are important to address:

Principle that the university research community holds dear: building on each others work. Yet many license agreements, whether drafted by University or by industry, prevent the distribution of software to other academic researchers.

Example source code copyright statement from the University of California: retains copyright for the University, allows extension but requests attribution, focused on research and educational uses. Commercial exploitation normally requires a commercial license agreement with the University. Limited distribution is the exception rather than rule. Companies prefer limited distribution. Important to retain unrestricted use rights for university once licensed to a commercial entity.

What is the actual value of University-developed software? In general, successful transfer of software from University to Industry is rather rare, despite a few notable exceptions such as NCSA Mosaic. University software suffers from quality issues, and generally is not ready for prime time. Exceptions do exist: Berkeley BSD Unix, Ousterhout's Magic, Tcl/Tk. The real product is often the students who built the software, not the software itself (an example is the DB group at University of Wisconsin).

What about conflict of interest issues? Research group develops software, spins off company with students and professors in the lead. Many examples of this. The ultimate arbiters of culture are the department chairs and deans, with responsibility for determining what is or is not acceptable.

Another layer of complexity: academia has a good understanding of how to deal with the Federal government, lots of history and much experience. However, these underpinings don't exist with respect to industrial interactions. In some ways, universities operate at a disadvantage with respect to negotiation with big companies, with their staffs of experienced corporate lawyers. No common principles, approaches when dealing with small or large companies versus small or large universities. Everyone is trying to "get the best deal."

CASI (Colorado state funded program to support university-industry interaction focused on software) IP agreement: funded by industry and state funds, university retains the rights, sponsoring industry obtains first rights to commercialize the software as well as internal rights to use. Industry receives object code but not source. Commercial licenses can negotiate rights to source code and continued support from the developers.

Industry has a difficulty with object code use licenses. Since code is not commercial quality, the main interest is in looking at the ideas embodied in the code. Also, may need to do local fixes to keep the software running.

While it would be nice to have general guidelines, the reality is that agreements are negotiated on a case by case basis.

Further difficulty for university researchers: delays between disclosure to licensing office and the approval to prepare a patent. Too much delay limits ability to publish (though new laws related to provisional patent applications may improve this situation). Licensing office should move more quickly to reach decision and return the IP to the inventor if the University chooses not to license. Note that software has a similar problem of time to market.

There are some examples of university software that is of commercial quality. SUNY Buffalo OCR software developed for the IRS has generated $500,000 in license fees so far. A major problem is how to equitably credit the 300+ students who have contributed to this software. Conflicts of interest problems are possible: students might select research projects to work on based on financial considerations. At SUNY, the software policy is derived from patents policy.

Another example is NCSA Mosaic. Lots of people worked on it, how to distribute license fees among them? Decision that the software belongs to NCSA and represents the creators as a "whole." Licensing fees used to enhance the research environment, not individuals. Netscape was formed, in part, for people like Andressen to get achieve larger compensation than is possible under this scheme. Interesting paradox. UIUC did not try to obtain patent protections for Mosaic. But if they had, there would have been more license fee sharing among the inventors. The university is considering more equitable rules. UIUC obtains over $3 million/year, from software licenses, 50% from Mosaic, most of the rest from health care/biomedical software.

Can't use patent perceptions for software. "Software is different." Must consider net revenue versus gross revenue: half of net intellectual property revenue at UWashington comes from software because of the high costs associated with patents (personnel time to evaluate, filling fees, litigation to enforce patents or defend them). Don't have to make all of the financial decisions up front with software -- get it used, figure out which software is valuable, which are the dogs. Negotiate different licenses after evaluation period. License revenue can be written so that researchers can recover distribution and maintenance costs during the evaluation period.

Back to principles. What does it mean to be in the public interest, faculty interest, and student interest with respect to industrial sponsorship of research? Federally sponsored research is relatively well understood, but this is not the case with industrial sponsorship. Universities do have policies, but these are frequently being excepted.

Few universities have specific policies with respect to copyrights or software (e.g., should software be protected by copyright or by patents)? Can we make recommendations?

Changes to patent laws due to recent GATT agreement. Provisional application associated with first publication, can file patent within one year for a modest application fee of $75.

Does the university own the intellectual property? If patents, the answer is yes. If it is work for hire (e.g., a staff member developing software as part of his/her job), then the answer is also yes. Other cases not so clear: faculty have many traditional rights related to copyrights. What happens if the faculty/inventor changes institutions? Experimental exception of the patent rights -- "intellectual interest" -- IEEE guidelines recommend exception for faculty member/member to use the intellectual property for research purposes only.

Start-ups: want access to intellectual property that came into being over time, such as the fruits of a research project. How are the rights assigned to the participants, such as students involved in the research project.

Industry would like to know what they are buying up front, but this is sometimes difficult in a research project.

Session II -- Where Do We Want to Be?

What are the principles that underlie public good vs. private gain? Are these really opposite?

Nature of the licensing fee structure: royalty-free vs. scaled royalty arrangements. Money plays a significant role that can change the perspective, especially when the perceived dollars are large.

Universities should focus on basic/fundamental research. Focused/directed research yields different objectives for the university.

Share the success among the involved parties and sponsors: developers, industry, university.

Government has an explicit view: inventions supported by federal funding should be commercialized. What about state funds?

How you get a new technology commercialized is not an easy process, full of risk and expenses. High license fees can inhibit successful commercialization, does not show appreciation for the expenses involved in bringing a technology to market.

If company pays salary of university researchers, then expects exclusive ownership of rights of interest to companies. Agreements should identify the "commercializing party." Industry pays $100,000s for the research, now the university wants to receive a royalties stream in addition? Industry says "no way."

University wants to protect itself if technology proves to be a big hit. Deserves participation in the revenue stream Can arrange scaled royalty structures: low fees at the beginning, scaling up if successful.

Many more steps (costs, risk) to commercialize, makes industry reluctant to commit to a royalty stream up front -- difficult to do if you don't know what the amount to be shared will be in the future. Industry would like non-exclusive royalty free licenses.

Leads to the question of industry access to prior technologies held by the university. Need carefully to scope the limits of any agreement.

Different categories of interactions may have different policies: recognize the difference between a small # of students working on a project up to large consortium-based projects.

At least for publically supported universities, regional economic development and job creation is as important as the return of revenue to university.

Rice policies: first priority is wide dissemination of the intellectual property. Second priority is the revenue stream. Preferred method is non-exclusive use rights.

University of Michigan CITI: highest priority is to create and disseminate knowledge. Non-exclusive licensing is the norm. Exclusive arrangements means that money issues come to the fore.

Important to distinguish between two activities

Is contract research a good relationship between University and Industry? Changes the role of the university -- lots of strings attached, does it become "work for hire"?

Principle: don't allow companies to tell the researchers what it is they are supposed to do. Who is creating the agenda--if it is not the researcher, then it should not be done at the university.

Principle: deploy new ideas as widely as possible. This could be achieved by giving it away OR licensing it exclusively. Disclosure issue when it involves private money and personal gain -- conflict of interest? Use the 6 o'clock new criterion: "Will this arrangement make the 6 o'clock news?"

Should consider industrial sponsorship in the context of the mission of the institution. Deliverables are not the same as onerous direction of the research. Industry would like ALL of the IP of the group as a beginning point of negotiation. This is natural but should be resisted.

University Mission -- a complex balance between knowledge creation vs. economic development and industry outreach.

Can't comprise academic freedom, service concept: deploying the results widely.

Contracts generate more problems with large companies than small companies

Do it for the public good

Deliverables: value received for money given; companies are paranoid, competition, might get sued.

If the researchers don't like the terms, don't enter into the agreement in the first place. Burden on company that publications reveal internal/proprietary information. Exclusive relationship means you have to pay more.

Govt funding: unlimited rights for govt funding only, restrictive rights otherwise.

Mixed funding is the norm, not the exception.

How to share information widely, but also allow successful commercialization which often can only be achieved through exclusive access to the technology.

NCSA: 75 partners in recompetition for supercomputing centers. How to manage IP among large scale consortia.

Guidelines: industry research is different -- define the statement of work extremely carefully, as well as the deliverables.

Identify who manages the relationship. What do the students need to know about the proprietary nature of the things they are working on? Researchers don't usually get to see the final contract. Disclosure restrictions not widely known by the researchers. Carefully define the background rights in advance.

Royalty streams: What is the value university will receive from industrial relationship? Fame, attractiveness to faculty, etc. Royalty requirements may make it unattractive for industry to support basic research that does not appear to have a commercial pay-off.

Algorithms are often the key, as opposed to the software itself. Leave flexibility for the university in negotiating software agreements. Internal use of source code may be highly desired by the corporate sponsors.

Consortium agreements: many sponsors, non-exclusive, perhaps royalty free agreements. Visitors -- develop sole technology at the university, company will own IP

Contract: each party wants to get the best out of it -- what about exclusivity. Think beyond revenue stream when considering value to the university. In general, the revenue streams are peanuts. Think of the long-term research relationship, not the money.

Equitable policies and fairness: Suppose that students who want to start a small company with the IP developed as part of their dissertation research. They have less rights as staff then as students to the exploitation of the IP.

High priority principles: academic freedom and the freedom of publication; must be concerned about education in a broad sense, and must have a proper concern for students -- they should not be exploited in this process.

Non-exclusive licenses are ok if many companies participate, paying continuing royalties is painful for companies, and this is part of the reason they find it undesirable.

A grant is a gift, allows researchers to do as you wish. This is distinct from contracts which are work for hire. You can distinguis between these by the level of funding.

Formation of Break-out Sessions

Breakout Group #1. Led by Bill Gear.

Breakout Group #2. Led by Peter Freeman.

What does industry expect from collaborative arrangements? What does institution expect, etc.? What is the benefit we expect (little revenue vs. large grant income)? More creative ways to give industry what they want? Preferential licenses that are still non-exclusive. Right to use. Early access, being involved in the research early on. Principles of revenue sharing.

Identify the potential players in a software licensing agreement

Recognize differences between maturity and sizes of the companies involved when it comes to negotiating licensing agreements.

Breakout Group #3. Led by Ed Lazowska

Focus on the underlying principles of intellectual property rights

Industrial visitors as well as faculty, students, staff. Researchers able to continue developing the technology when researcher moves to new institution.

Some Principles:

NCSA partnership agreements: "safe zones" for ownership; sliding scales of partnership rights -- whole partnerships can exclude competitors from the consortium, but only at a higher level of support.

Breakout Group #4. Led by Rick Adrion

Focus on Multi-University, Multi-Industry Consortium Agreements

The Different Types of Consortia

Set up consistent rules within consortium agreement; consolidate rights and use rights.

Attribution: must establish a clearance mechanism, make proper use of names of parties involved.

Make all participants aware of the risk involved. Three kinds of risk: infringement, breach, and liability. To protect against infringement, it is important to get info to the PIs. To avoid breach, PI must inform staff and students. To be protected against liability, must get idemnification.

Financial interests: must identify what "divide equally" means in advance. Need some kind of conflict resolution mechanism. The method of division is likely to be different for different kinds of aggrements.

Important Points as Identified by Workshop Participants

There is a lack of understanding among the various parties involved of the issues of intellectual property. The education/research mission of university must be tantamount. OTTs can't do there jobs alone -- it is not possible to simply throw the disclosures over the wall. Faculty are too preoccupied with fame.

Treatment of derivative works and source code -- the major challenge is how to do effective technology licensing in such a fast moving field as computer science and engineering.

Entertainment of idea that all IP should be treated in the same way is extremely important. Universities do not now have consistent policies.

Keep it simple -- no single licensing agreement is likely to cover all cases. First and formost, we need an agreement to agree. Too much time is wasted in license negotiation in the face of a very fast moving field.

Appeal to OTTs to take it easy on software. So far, licensing revenue has been low. So cut deals that allow the software to be put in the hands of users as quickly as possible. Renegotiate later if software proves itself to be value.

Software requires a different process than the traditional one that works for patents. The method disclosures are different: patent filing versus source code access. In keeping with the university's role as a disseminator of knowledge, ability to distribute source code is very important.

Software is different. It requires a different process, not necessarily different policies.

Universities must realize that IP is becoming more important. New policies are needed and existing policies should be reviewed. Universities should educate the faculty, staff, and students on the importance of IP and the relevant university policies.

Part of the educational mission of the university should be the obligation to prepare graduate students for their professional careers by exposing them to IP issues.

Rights of the graduate students as developers of software IP are not well served by many universities' existing policies, which are mainly geared towards faculty rights.

It is important to determine the rights of individual contributors towards the development of IP. Their responsibilities, as well as the university's, should be carefully delineated by policy.

Take it easy on software. In many ways "software is a form of publication." Universities should not stand in the way of rapid dissemination of software.

Graduate students rights and opportunities with respect to intellectual property should be carefully described in policy. They need protection and should be well informed of their rights and responsibilities.

Nice, clean, definative definitions; note that the different players have different expectations; manage these/control/modify them in the context of the university mission.

Continue to increase the understanding of expectations of and constraints on all of the players. Foster collaboration on policies among the faculty, administrators, and OTTs.

A general agreement that the situation on most campuses is about the same. "My place is not off scale." It seems that there is no single agreement that is possible, and each agreements strikes a balance among the needs of all of the involved parties.

Important to explain to all the constitituences exactly what the problems and issues are.

Realize that software is different than patents anmd traditional copyrighted academic works. Develop policies that promote innovation and further invention.

Promote development, don't restrict it. Advance knowledge, don't lock it up.

Optimize for the common case, not the home runs, which occur all too infrequently with software.

Attempt to develop a common set of policies for different kinds of IP. If this is not possible, then set guidelines and priorities for IP. Avoid doing it on case by case basis.

Make the primary objective the enhanced collaboration between industry and university researchers. Licensing revenue should be a secondary consideration.

When fortune is the prime consideration, less progress will be made.

Encourage collaborations between industry and university. Remember that the university's primary goal is the creation and dissemination of knowledge.

Make sure to understand the issues confronting each party in the licensing equation. Since the common case is that nobody will get rich from licensing revenue streams, keep everyone's expectations within reason.

The university provides a very useful neutral meeting ground for industry in the context of a university research project. Companies should not take universities for rides. There should be fair collaboration. Specific liasons should be identified. Exclusivity is problemmatic for universities. Licensing revenue is not the only nor the best metric of success in university-industry collaboration. This must be the view held on the industrial side as well.

Income is only a small part of the value of the relationship. It is important to agree to agree in license negotiations. Good to develop policy guidelines. However, you still need to able to make exceptions. Recognize that there are many different ways of disseminating knowledge, such as the successful formation of small companies and the creation of new jobs. Discussion and interaction among the communities represented by the workshop attendees is very good and very productive.

The primary importance of a meeting of this type is the way in which the players (OTT, CS, industry) can come together to discuss these issues. IP rights should be considered under the same conceptual framework whether patents, copyrights, etc. This gives universities a tool for consistent polices.

Shift the funding base of OTTs to something more stable. Technology offices are forced to focus on the quick kill to sustain themselves. This is the wrong incentive. Maket hem a partner in managing university-industrial relationships, not an impediment.

University has to sell its intellectual resources, then the agreements should not diminish that value.

Universities can pursue any activities provided that those related to the deployment and acquisition of innovations are properly disclosed. This is important so that the appropriate individual and institutional rights are maintained. The bottomline is to keep your OTT people informed of what you are doing.

Look at the broad picture of the university's mission with respect to industrial interaction and intellectual property. Importance of disseminating knowledge, fostering industry relationships, fame and reputation of the institution, and recurring royalties. Develop generalized polices that maximize the VALUE -- broadly interpreted -- received under the auspices of university created IP (patents, technology, copyrights, trade secrets, etc.).

Write report as though it was a set of guidelines for agreement. A possible goal for a future workshop might be the drafting of a model licensing agreement.

Next Steps

Last updated 15 December 1996 by Randy H. Katz