ed rights to Management Company Entertainment Group, Inc., but had granted proceeds to, say, Virgin Vision, Inc., how would Lender A even know about this grant if Producer was not forthcoming?

That is the easy case. At least one can ask the Producer and hope for a complete answer. The Copyright Report for Latino also shows that Management Company Entertainment Group, Inc. granted rights to Orion Pictures Corporation. Now, assume Lender A wants to extend credit to Orion based on the value of its assets. Lender A wants to know what prior interests exist with regard to these assets. Orion may know that it obtained rights from Management Company Entertainment Group, Inc., but it may know nothing about, say, a separate assignment of proceeds to Virgin Vision, Inc. Even if Lender A asked and Orion answered honestly, Lender A might not find out. If Producer is unforthcoming or simply unavailable, Lender could not search by debtor with regard to prior grants of proceeds because it does even know who the prior debtor are.

By splitting filings of rights from royalties, the Proposed Federal Intellectual Property Security Act seriously undercuts the ability IP owners to engage in asset based financing of IP rights. Licenses of intellectual property rights can be either exclusive or non-exclusive. For exclusive licenses, the licensee typically undertakes further exploitation of the work through sublicenses or disposition of copies. It is crucial for a senior lender to know that is security interest against the licensor's rights is prior to and entitles it to royalties from sublicensees, and equally important for financiers of sublicensees to know about prior security interests. This is similar to the position of the lenders against office buildings. The permanent lender wants to ensure that its mortgage against the property has priority against the leases of space in the building, and a junior lender against the property or a leasehold estate needs to know about prior mortgages. Non-exclusive licenses, on the other hand, are typically granted to end users, such as a merchant who uses software in its business. In that case, it would be helpful if a floating against the debtor attached to the non-exclusive license. Yet under federal law non-exclusive licenses are not assignable, so even if the lien did attach, it could not be assigned to the lender on foreclosure in any case. The Proposed Federal Intellectual Property Security Act does not even address this issue. What it does is undercut the ability of exclusive licensees to continue the traditional financing method that they need in an attempt to enable a financing method for non-exclusive licensees that they can not use.

The federal filing system, at least for copyrights, looks to the work as the main determinant of value. From this perspective, it quite correctly indexes filings against the work, and it supports well "vertical" asset financing. The decisions in Peregrine and AEG are correct from this perspective. The Proposed Federal Intellectual Property Security Act only looks at financing from the inventory model. It criticizes Peregrine and AEG for not supporting this methodology. This criticism is misplaced. The proper course is to develop a filing system that supports both methods. B. The Proper Approach: The Relational Data Model

The previous section gives examples of where a mixed system will fail. But proper analysis requires more than a list of problematic cases. We must also ask why the system fails. The answer is a faulty data model. Essentially, the mixed system proposes a hierarchical data model with two roots (technically, a network model). Such a data model cannot ensure consistency or integrity in the data model itself. Let's see why.

1. The Hierarchical Data Model

Consider a simple example, which we might call the "Rolodex Problem." A Rolodex creates a hierarchical database arranged alphabetically and indexed either by individual ("entity") or by company ("object"). A hierarchical database is often represented in a tree-view as follows:

In this case, the database is indexed by individual. One searches the database by "walking the tree" from the root down each branch until the desired data node is reached

As anyone who uses a Rolodex knows, there are problems with this approach. You receive a business card from a client. How do you file the card, by individual name or company name? If you file by individual name, what happens when you want to visit a particular company and all your contacts there? The only way to find all the contacts is to search through all the cards. What happens if an individual leaves a company? If you throw away the card, you lose information about the company. What happens if you file everything by company? How can you find an individual if you forget the company name? Maintaining two databases, one for individuals and one for companies, doubles the filing work, and risks having inconsistent databases (lack of data integrity). The fact is there is no satisfactory solution to these problems using a hierarchical database. The model itself is inadequate to the task.

2. The Relational Data Model

The proper solution to the Rolodex Problem is to implement an entirely different data design, the relational data model. Unlike hierarchical models, which are based on an ad hoc data model, the relational model is based on a consistent underlying mathematical theory derived from predicate logic. In simple terms, all data is expressed in tables consisting of rows and columns. These tables are related through a key column that uniquely identifies each row. The Rolodex Problem is solved by maintaining three tables like this:

One table holds the data for persons, another for companies. A third table relates persons to companies. The Person/Company Table identifies that John Jones (P1) and Ann Tones (P2) both work for Acme Co. (C1). Pete Clones (P3) works for Widget Co. (C2). Now, if we want to visit Acme Co. and find all the employees who work there, we just search the Person/Company table for all entries for Acme Co. (C1), and for each entry look to the corresponding Person key and use that to find the entry in the Person table. If Pete Clones moves to Blackacre Co., we simply update the entry in the Person/Company table to indicate the change. We do not need to delete the entry for Widget Co. when Peter changes jobs.

It should be clear that the Rolodex Problem is identical to filing security interests in intellectual property. There are really two different indexes in use. One can record against the owner of the rights, i.e., the person. This is the approach taken in under Article 9, which indexes security interests against the debtor. Alter

natively, one can index against the work (company), i.e. the IP rights. This is the approach taken under federal law, for example in the Copyright Office, where one records in reference to a registered work. Both state and federal law use single index, hierarchical databases. Both of them lead to the problems commonly found in hierarchical systems. Neither one does or can provide a complete solution by itself.

The Proposed Federal Intellectual Property Security Act adopts the worst solution of all. It requires maintaining two hierarchical databases, with all the problems in that data model, without any method of determining which one prevails in case of conflict. This is like maintaining two Rolodexes with separate data in each one, such as phone numbers in one indexed by name and addresses in the other indexed by company, without any methodology to ensure consistency between them.

It is pointless to argue the merits of inventory financing (Article 9 approach) over asset financing (federal IP approach). Each works fine for its own financing model. Neither one works well for the other. But the universe of IP financing requires we use both. We cannot address this universe by tinkering with either hierarchical model (e.g._choosing either state or federal system) or by simply decoupling them (Proposed Federal Intellectual Property Security Act). The underlying data design is wrong. We must move to an entirely new data model.

C. Implementing The Relational Models

In a relational design, we need to maintain two separate tables (registers), one for persons (owners, transferees, secured parties, etc.) and another for IP interests (copyrights, etc.) The Copyright Office already maintains two such registers now: a work register, and an author register. In concept, we need to amend the Copyright Act to allow constructive notice from filings in the author register, and then give the Copyright Office authority to create a relational database that relates filings between the two. We should start with the Copyright Office as the first step in order to make sure the system functions correctly and then as necessary roll out the system to the P.T.O.

As a design matter, we would not be working on new ground. The World Intellectual Property Organization maintains an International Register for Audiovisual Works.5 This International Register has been set up on a proper relational model, with two registers, a “person register” and a “work register" with systems to relate filings. W.I.P.O. has already established forms and procedures for its use. The current Registrar of Copyrights spent close to a year at W.I.P.O. working on the International Register and is quite familiar with how these systems should be established. There is of course no need to adopt any of the W.I.P.O. rules or forms, and I am not advocating that we do so. The point is that there is a wealth of knowledgeable talent available at the federal level to implement a proper system.

We should identify in principle what we want the system to do (data design phase) before constructing the legal rules (coding). I suggest the system should do the following:

Single, National System: Article 9 envisions separate state registers. But intellectual property rights under federal law are national in scope. Thus, we need a national database to deal in national rights. It must also be a single database. We cannot maintain two separate databases and ensure data integrity. This argues that the filing structure must be a single, unified federal system.

Relational Data Model: The federal system must implement the relational data model. In other words, the system must allow filings against the work (IP Rights) as well as filings against persons. This will require maintaining a separate "person" index. The system must maintain the relationship between them. The work of the System Manager (e.g. Register of Copyrights) is to maintain the database.

Constructive Notice As to Persons: As a legal matter, filings in the person index must also impart constructive notice to establish priority against subsequent transferees. There may need to be a "birth certificate," like a registration certificate, identifying the first filing for a person. This system would greatly simplify such matters as filing corporate name changes, mergers, etc.

Floating Liens: The system should allow for filing floating liens. This could be done by filing against an individual in the person index. The system would then attach the filing to all registered works of the debtor.

After-Acquired Property: The system should allow for filings that apply to afteracquired property. The Copyright Act now allows for transfers of works to be cre

5 For political reasons we need not discuss here, the U.S. has not adhered to the Treaty establishing this Register, although many countries have.

6I was a member of the U.S. Delegation to the Diplomatic Conference that established the International Register, and also spent considerable time working on the project.

ated, although there is no constructive notice effect until the work is registered. This becomes a problem when advancing funds to create a work, such as for software development or motion picture production. Again, filings in the person index can accommodate this approach. We may want a time limit, such as the filing only applies to works registered within X years of the filing date.

Others may have additional suggestions as to how the system should operate, but I suggest that this is the direction discussions should take. Please note that the Copyright Act only has a federal filing system for assignments and exclusive licenses. Thus, this discussion only relates to them. Different considerations may apply to non-exclusive licenses.

CONCLUSION

Intellectual property rights have become the center of the American economy. The IP industries are the faster growing segment of the economy and the engine of new job growth and wealth creation. Continued development in this area requires a modern system for facilitating secured financing.

The relational data model is the modern data model. It is the basis for the vast majority of business databases. It is implemented in all the office suites (Access, Approach, Paradox) as well a products from major business suppliers (Oracle, Sybase, Microsoft SQL Server). It is the only data model grounded on a solid mathematical foundation.

A modern financing model for IP Rights should be grounded on a modern data model.

Mr. COBLE. Thank you, Mr. Brennan. That bell indicates we have a vote on. Ms. Chasser, why don't we hear from you and then I will go vote and come back if that is okay.

STATEMENT OF ANNE CHASSER, DIRECTOR, OFFICE OF TRADEMARK & LICENSING SERVICES, OHIO STATE UNIVERSITY, ON BEHALF OF THE INTERNATIONAL TRADEMARK ASSOCIATION (INTA)

Ms. CHASSER. Thank you, Mr. Chairman. The International Trademark Association appreciates the opportunity to appear before the subcommittee to comment on the need to reform the system governing security interests and trademarks. America's fastpaced economy is motivated by ideas, information, and technology. To put it another way, it is driven by our intellectual property. Therefore, many consider intellectual property, including a company's trademarks, to be assets in the same manner as a building, a piece of machinery, or a work of art.

Trademarks are often a company's most valuable asset. As a result, trademarks can become collateral to finance a new venture or launch a new product line. In these circumstances, the lender holds a security interest in the trademark. The UCC is not totally effective when dealing with the security interest in a federally registered trademark. The problem lies in the interplay between the State-codified UCC and the Lanham Act. Among trademark practitioners and legal scholars, there is uncertainty as to whether the Lanham Act preempts the State UCC. The UCC states that article 9 does not apply to security interests if the parties' rights to the property are governed by a Federal statute. It also states that if the Federal statute does not address this, then article 9 of the UCC may indeed be looked to for an answer.

So we turn to the Lanham Act. It does not directly address security interests per se. However, it does contain provisions allowing for assignments of trademarks to other parties to be recorded. When all is said and done, the Lanham Act has generally been interpreted not to be a Federal statute that preempts article 9, al

though the case law is far from being consistent on this issue. The result is widespread legal uncertainty for intellectual property owners and also for purchasers of business in which intellectual property is an increasingly valuable part of the transaction.

The practice, which has evolved among trademark practitioners, is to record the security interests at the State level under the UCC and also record a copy of it with the U.S. Patent and Trademark Office. Unfortunately, this practice does not always work due to the ambiguities of the recording security interests under section 9 of the Lanham Act. In fact, this practice has often led to disastrous outcomes, the loss of trademark rights and their value as collateral. What this amounts to, Mr. Chairman, is a troubling lack of certainty as to how to perfect a security interest in trademarks. In other words, how do you put other parties on notice that the interest in the trademarks exist?

This ambiguity has negative repercussions for trademark owners, lenders and potential purchasers.

Mr. COBLE. Ms. Chasser, if you will, hold that thought. Let me proceed to the floor.

You all stand easy in the interim, and I will return imminently. [Recess.]

Mr. COBLE. I apologize to you all. The best laid plans of mice and men go awry. The vote that was to have been about 30 minutes from now was accelerated. I thought, rather than come back and return to the floor, I would just stay over there.

So thank you all for waiting. Ms. Chasser, are you still holding that thought?

Ms. CHASSER. I am holding that thought.

Mr. COBLE. You may continue.

Ms. CHASSER. We were talking about the ambiguities in the recording of security interests.

There is one promising answer, and that is a proposal to establish a national uniform recordation system for tracking security interest in trademarks. Under this approach, security interests are filed under section 9 of the UCC in the applicable State, but then they are also supplemented by a new notice filing at the Federal level. This allows potential purchasers to check only the Federal database rather than resorting to guesswork or conducting timeconsuming searches of all fifty States for a UCC filing.

This is a straightforward method that utilizes the UCC apparatus already in place and is familiar to lenders, borrowers and purchasers in the commercial world. Yet it also integrates the existing Federal registration system.

By reconciling these two established systems, this approach will provide an effective mechanism for recording interests in trademarks.

The benefits for all parties of a central database will not be realized, however, until the 3-month grace period for filing an assignment under section 10 of the Lanham Act is eliminated. This will encourage prompt recording of security interests so that prospective purchasers of the trademark will have timely notice of security interests in the property.

To conclude, Mr. Chairman, I want to thank you again for this opportunity. INTA remains committed to working with you and