Copyright in Computer-readable Works

copyrighted works. The following estimates are cited merely to shed some light on the magnitude and distribution of photocopying. 3

1. 27.5 billion paper copies were made by photo-copiers and
photo-duplicators in the U.S. in 1967.

2. Approximately 60 percent of the material copied by libraries.
is copyrighted.

the journal-to-book ratio was 9:1;

the majority of items were scientific and technical;

over 80 percent was less than five years old;

5 percent of the publishers produced 40 percent of the material being copied.

Currently, almost none of this photocopying results in an associated royalty payment or license fee.

A similar situation exists with the use of computer data bases. These data bases may contain scientific, economic or statistical data, bibliographic material, or medical and legal information. In many cases the data has associated with it computer software to facilitate access and use. There are a variety of existing agreements by which the creator of the data base collects for its use either directly from the customer or from one or more of the computer system operators who provide access to the data base.

As we described above, in both the photocopying and computer data base areas the economic issues are the comparative efficiencies of free provision versus the implementation of user charges, and the relative magnitudes of the collection and enforcement costs (the transactions costs).4 These costs will depend on whether blanket licenses or per

use licenses are utilized.

The obvious archetypes of the blanket license are those employed by the performing rights societies (e.g., ASCAP and BMI). Here, a clearinghouse is employed to facilitate the contracting arrangements. The proposed Australian copyright royalty collection operation that resulted from the Morehouse decision will operate in a similar manner. (The decision in the Morehouse case was that libraries in Australia are responsible and liable for photocopying of copyrighted works done on in-library copying machines.)

Usually ASCAP's operating costs are less than 20 percent of revenue. The Australian publishers association has predicted that the costs of their monitoring activity, analysis, and transactions will be

approximately A$.01 per page (one Australian cent per page).5 In both systems, a significant part of the cost is the monitoring of usage (what is performed or copied) so that the revenues can be divided among the copyright holders.

In looking for archetypal billing and collection systems for per-use charges, we found two different industries with well-developed and possibly interesting accounting and billing mechanisms, computer "service bureaus" and local telephone operating companies. One large service bureau organization estimated that the costs of monitoring use, accounting, billing, etc., generally are 15 to 18 percent of total costs. On the other hand, Pacific Telephone Company (which has complex multi-message unit charges for local calls) records shows that all accounting operations amounted to only 3 percent of company expenses for 1975. (Both the Accounting Department expenses and total expenses included all current and capital items. See Fig. B.1.)

The greater the amount of information collected, stored, and analyzed, the higher the costs. For example, New York Telephone does not, as a rule, itemize "message unit" calls on either residential or business customer bills. However, now they must provide such a list to the customer on demand if the customer is willing to pay the extra cost ($1.50 for residential customers and $1.50 plus $.25 per each extra page for business customers).6

An important point to remember is that in neither case do these costs include the expense of determining how to pay out the revenues. These disbursement costs will be related to the degree of accuracy required (i.e., sampling vs. 100 percent monitoring) and the frequency distribution of the copyright holders. Recent data from the British Lending Library (BLL) indicates the skewed nature of the frequency distribution.7 Their survey indicated that of 14,967 serial titles, the top 210 titles accounted for 20 percent of the demand for photocopies and the 6,000 least requested titles accounted for the last 10 percent of the demand. The cumulative distribution curve is shown in Figure B.2. Figure B.3 lists the 15 most "popular" titles.

Although the BLL is a "library of last resort" for academic libraries, it is a major resource for the specialized industrial libraries who comprise a majority of their borrowers. Therefore, we can assume that, although the BLL data may not characterize the photocopying in the U.S. in an unbiased manner, the U.S. data will also exhibit a high degree of skewness. Depending on the exact nature of the payment algorithm, this skewness can lead to either lower or higher costs in the distribution of royalties to copyright holders. The existence of a high threshold number of copies per time period -- unless X copies per month are made, no royalty payments are distributed -- coupled with the skewed distribution could reduce transaction costs in the same way that "deductibles" do for an insurance policy. On the other hand, in the absence of a threshold number, quite large samples may be required to capture the copying of the more obscure works.