agreement for the final negotiations, so I hope that it will be given very serious consideration. Consider the alternatives. I, for one, do not consider many more bilaterals a very reliable alternative.
Intellectual property rights have been becoming an increasingly important part of AT&T's competitive strategy. With the continuing increase in worldwide competitiveness, this trend seems likely to continue.
I am going to take the liberty here of offering a few comments on some of the matters discussed earlier.
First let me make one other observation. AT&T does take into account the strength of intellectual property protection in making investment decisions in particular countries, and although the level of protection is not always a go-no-go decision maker in and of itself, it is an important element. In some instances it has been the sole determinant of an investment decision. When AT&T was asked to license UNIX in Brazil, we pointed out to the management of the company the level of software protection then available and the request was denied.
Of course, evaluation of the level of protection is heavily dependent on the age of the technology to be put into a particular country. That is alluded to by Mansfield in Chapter 5, who said you cannot simply look at gross investment numbers. You need to get underneath them and see what they represent. For example, in the semiconductor world, we measured the generation of technology by the width in microns of the lines etched on the chip. To go back a few years, if you were licensing 2.5-micron technology, you were going to have one degree of sensitivity to the level of IPR protection, but an entirely different sensitivity if you were licensing 2-micron technology.
I would like to clarify, if I may, several allusions that were made earlier in the conference to the European Community directive on computer software suggesting that it somehow gave recognition to or was a form of sui generis protection. It quite clearly is not. The EC directive, if you read it, says that it supports protection of computer software both by copyright under the Berne Convention and also under the patent laws. So instead of suggesting it provides support for a sui generis procedure, I think it is just the opposite.
Finally, one speaker suggested that since there have been some bad copyright decisions in the computer software area, and I fully agree, the only solution is sui generis protection, because these decisions show the copyright system does not work for software. To me, it only shows that some judges do not understand copyright law and I think we have a very effective appellate process for taking care of that. So, before we look to sui generis, let us first try the appellate process.