I. Introduction

The Internet is a relatively new way to conduct commercial exchanges. [1] In the span of a little more than five years, the Internet has gone from a curiosity to a necessary component of any business plan. [2] Electronic commerce (e-commerce) on the Internet has developed so quickly that it has left the law lagging behind. [3] The result is that courts are increasingly faced with Internet related problems that are not only beyond their technical understanding, but also strain traditional legal doctrines.

One example of this is the application of trademark infringement doctrine to e-commerce. Trademarks can be used in e-commerce in a variety of ways. A trademark could be used as part of the domain name of a web site. [4] A trademark could also be used on the face of a web page. [5] Finally, one could use a trademark in the metatags [6] of a web page. [7]

One of the issues this Article seeks to address is whether, in any of these three instances, the holder of a trademark has a claim for contributory infringement against an Internet Service Provider (ISP) that gives the infringing party access to the Internet. The second issue this Article addresses is whether allowing contributory infringement suits against ISPs is good public policy. Part II of this Article will explain how the Internet works and the application of trademark infringement law to the use of trademarks in e-commerce. Part III will discuss contributory infringement. Finally, Part IV will discuss the public policy implications of allowing suits for contributory infringement against ISPs.

II. Background

A. How the Internet Works

[p67] The Internet is an open network of interconnected computers.[8] Host computers called servers operate as the links in the Internet network. [9] These links are independent of each other in that a problem in one place on the network can be routed around by connecting through other servers on the network. Each server is capable of hosting hundreds of web sites. [10] Each web site's first page is known as its homepage, and this homepage has a unique numeric address or URL. [11] There are two ways to access a web site on the Internet. [12]

One way is to enter the web site address, or domain name, into a web browser such as Netscape Navigator or Microsoft Internet Explorer. These browsers read a text web address entered by a user such as www.bu.edu as a numeric address, but browsers allow users to use text terms rather than numbers to access web sites. [13] The theory being that web users will have an easier time remembering text terms related to the content of a web site rather than a long string of numbers. [14] The second way to access a web site is by using a search engine such as Google [15] or Hot Bot. [16] These search engines respond to a text query entered by the user such as Boston University. The search engine then creates a list of web sites that most closely match the search terms. [17]

Whether a web site is accessed directly or through a search engine the information the web site contains is transmitted on the network through the use of a common language called hypertext mark-up language (html). [18] This [p68] universal language allows web users to view the graphic and text displays created by the web page designer. [19] This language is read by the computer and is not seen by the web user. [20] Part of this hidden html code is metatags. [21]

1. What Are Metatags

There are different types of metatags contained in the hidden code of web pages. [22] The two most common types used by web site designers are keyword and description metatags. [23] For instance, the keyword metatag for omahasteaks.com [24] lists the terms "steaks, omaha, cooking, gourmet, food, steak, gifts [and] meat." [25] These are terms that clearly describe the content of the web site. When a web user enters any of those terms, a search engine that uses metatags to categorize web sites will return a list of web sites that includes omahasteaks.com.

The metatag description of a web site contains a textual description of the content of the web site. For instance, the metatag description on the Omaha Steaks web site reads: "Omaha Steaks offers premium heartland quality steaks and meats, as well as seafoods, and desserts. Free burgers when you order from our site." [26] A search engine may display this description along with the web site address in the query list that results from a web search. [27]

2. How Search Engines Work

A search engine has three parts. First, the search engine software uses what is known as a spider to actively travel the Web seeking out new web sites and reporting its results back to the search engine database. [28] The database contains information on all the web sites the search engine has cataloged. [29] Second, the database catalogs web sites according to the software program of the search engine. Some search engines give little or no [p69] weight to metatags, while others will rank a web page higher on a query list when a keyword appears more than once in the metatag. [30] However, not all search engines use metatags to categorize the Web. [31] Many search engines rely on scans of the individual pages of a web site for key words as well as the domain names of those web sites in order to categorize a given web site. [32] No matter where a trademark might be used on a web site it is likely to be picked up by a search engine spider and become part of the categorization of that web page. [33]

The third part of the search engine is the search function that responds to the keyword queries of users. [34] This software searches the database and returns a list of the web sites it has cataloged that most closely respond to the user query. [35]

The limitations of this system are two-fold. [36] First, this system of locating a discrete web site is limited by the search engine database. [37] Obviously, the larger the database the more likely it is to contain the web site that the user is seeking. Second, this system is limited by the accuracy of the user's keyword input. [38] Anyone who has used search engines before knows that the use of general terms can result in a query list that is literally thousands of entries long. [39] The savvy web user knows that even a very specific inquiry can result in a number of responses that are irrelevant to what the web user is seeking. [40]

If, for instance, one were looking for the Omaha Steaks web site and entered the search terms Omaha Steaks the resulting query list would contain 27,200 responses. [41] The first site on the list is the Omaha Steaks [p70] web site, and most of those in the top ten are related to the sale of Omaha Steaks. [42] However, a search for the term steak returns nearly 900,000 web sites, and the Omaha Steaks web site is eighteenth on the list. [43]

Most users will not venture beyond the first ten or twenty responses to a search. [44] Web site designers are keenly aware of this, and therefore they have a strong incentive to get their web site listed in the top ten of a keyword query. [45] Furthermore, this simple example shows that the use of trademarks on the Web can help to narrow the scope of search results. Those who use trademarks on their web sites increase the likelihood that people searching for a specific product will be able to locate their web site on the Web. [46]

3. Why Web Page Designers Manipulate Search Engines

Web sites are created for commercial purposes and must be visited by customers to be effective. Designers are aware that most Internet users locate web pages through the use of search engines. [47] This gives unscrupulous web designers an incentive to manipulate search engines by including irrelevant terms in their web pages. In this way web designers hope to increase traffic on their web site in order to increase the sales of products and services.

A second reason for manipulating search engines is that many web sites sell advertising. As with television advertising, the rates charged on the Web are tied to the number of viewers. [48] On the Internet the number of viewers of a web site is measured in "hits." [49] A hit is an individual visit to a web site. [50] Thus, even if the web user only visits a web site for a moment the web site records a hit. Increased hits to a web site result in increased advertising revenues for the web site.

[p71] Given these monetary incentives it should come as no surprise that some web sites will resort to the unauthorized use of trademarked terms in their web pages in order to attract web users to their web site. [51] To understand why manipulating search engines in this way is infringing behavior it is important to understand the basics of trademark law.

B. Trademark Law

A trademark is defined as: "any word, name, symbol, or device, or any combination thereof" that a person uses in commerce to identify goods. [52] Trademark law is really just a part of the broader law of unfair competition. [53] As a portion of unfair competition law, trademark law seeks to encourage the proper conduct of businesses in the economic market place.

The theory behind allowing protection for trademarks is that it allows consumers to more easily identify the products they seek. [54] Consequently, it also is thought to prevent the palming-off of goods of inferior or different quality as those of the trademark holder. [55] Trademark creates a property right in the holder of the mark. [56] The holder has an incentive to invest in the mark by advertising as well as simply by creating a quality product that people will wish to purchase again and again. [57]

By granting some limited protection for trademarks, the holders of trademark rights are encouraged to build goodwill in the minds of consumers; and consumer search costs are reduced because goods are more easily identifiable. [58] The trade off, however, is that the holder has some right to exclude the use of his mark by others which can be anti-competitive. [59]

1. Levels of Trademark Protection

Trademark law allows for varying levels of protection depending on the strength of the mark. [60] This is important because the trademark holder's ability to exclude others from using his trademark is directly linked to the [p72] strength of the mark. That is, a strong mark will get more protection than a weak mark. There are four categories of trademarks. A trademark can be generic, [61] descriptive, [62] suggestive, [63] or arbitrary or fanciful. [64]

2. Trademark Infringement

The primary way a trademark holder enforces his rights is by bringing suit against alleged infringers. The Lanham Act of 1946 created the federal statutes governing trademark infringement. [65] Under the Lanham Act one infringes a trademark when, without consent, they use another's mark in [p73] commerce in a way that is likely to cause confusion to consumers. [66] The key statutory elements of infringement are: use in commerce and likelihood of confusion. Use in commerce generally requires the sale, or at least the shipment, of goods having the trademark on them. [67]

When a court seeks to determine whether there is a likelihood of consumer confusion they rely on a balancing test that includes a number of factors. [68] These factors are: (1) strength of the mark; [69] (2) proximity of the goods; [70] (3) similarity of the marks; [71] (4) evidence of actual confusion; [72] (5) marketing channels used; [73] (6) type of goods and the degree of care likely to be exercised by the purchaser; [74] (7) defendant's intent in selecting the [p74] mark; [75] (8) likelihood of expansion of the product lines. [76] This is a non-exhaustive list of factors that courts may consider. [77] Because this is a balancing test no particular weight is assigned to any given factor. [78]

These are the basic elements of the infringement analysis. Courts have identified a number of legal theories regarding when consumers may be confused. [79] For purposes of this Article only two of these theories are relevant. The first is the traditional theory of direct infringement discussed above. The ideal example of this type of infringement occurs where similar products bearing confusingly similar trademarks are sold side-by-side. The second theory courts have relied upon in the Internet context is initial interest confusion. [80] Here, the consumers confusion as to source of goods occurs before they see the products side-by-side on a shelf. [81] Rather they are lured, if you will, into a store on the promise of certain goods being present there only to find the goods of a different producer. [82]

When a trademark is used in a domain name or on the face of a web page it is easy to see both used "how the mark has been" in commerce and the potential for "likelihood of confusion." [83] When a trademark is used in a metatag there is some debate on the issue of consumers confusion. [84] This is primarily because web users never see a trademark contained in a [85] Some commentators claim that it is the search engine that is in fact being confused rather than the consumer. [86] Other commentators make the more practical observation that consumers are well aware that search engine results contain a number of listings that are irrelevant to their intended [p75] search. [87] The distinction between trademarks in metatags and in domain names, or on the face of web pages, will become important when assessing the public policy of allowing for contributory infringement suits in each context.

3. Fair Use

A trademark is a property right that allows the holder some measure of exclusive use of the mark. [88] The fair use defense is a limitation on this property right. [89] It allows others to use a protected mark for certain limited purposes. [90] Under the Lanham Act, a defendant claiming fair use must prove three elements. [91] First, a defendant must prove that their use of the mark was not as a trademark. [92] Second, the defendant must prove they used the trademark in "good faith." [93] And finally, they must establish that their use of the mark was descriptive only. [94]

The simplest example of fair use is comparative advertising. Here a seller makes reference to a competitor's product in order to make a comparison between the two. Because it is often impossible to refer to a competitor's product without using the trademark, the fair use defense insures that the holder of a mark does not have complete monopoly use of a given trademark term. The existence of a fair use defense to trademark infringement is important in the context of contributory infringement suits because at some stage in the process, leading up to and including litigation, someone must consider the availability of the defense. The public policy issue that will be discussed later is: Who should bear the burden of determining whether fair use is a viable claim in a given case? In connection with that question, one must consider the affect on commerce of placing that burden with either the trademark holder or the alleged contributory infringer. [95]

C. Trademark Infringement on the Internet

There are three different ways a trademark could be infringed on the Internet. The following discussion is intended to point out the legal theories relied on by courts in each of these three instances. Arguably the most clear [p76] cut example of trademark infringement is where similar goods are sold side-by-side with similar trademarks used to identify them. Here, the potential for consumer confusion is most obvious and its affect is most direct on the producers of the respective goods, as the sale of one of the goods necessarily means the loss of a sale for the other producer. Some courts have resorted to the theory of initial interest confusion to find infringement in instances where, at least at the time of purchase, the consumer is not confused about the source of the product they are purchasing.[96]

This theory falls a little left of center in terms of common law understanding of what trademark (or unfair competition law) was intended to protect. This is important because the subject of this Article is third party liability. As will be explained, third party liability requires that there is an initial infringer. [97] If the initial infringement is attenuated at best, then the argument, at least from the public policy standpoint, against holding a third party liable is strengthened. [98]

1. Domain Names

A domain name is the text version of a web site address. [99] The domain name appears at the top of the page in most browser programs and is readily seen by the user. [100] If a web user does not know the domain name of the web site they wish to visit they may choose to use a search engine. Many times however a web user will simply assume that the web page of a company is the name of the company followed by ".com." [101] A domain name serves much the same function as a sign on a storefront -- it identifies goods and services in commerce.

In Brookfield Communications v. West Coast Video, [102] the Ninth Circuit Court of Appeals applied a direct infringement analysis to a case involving domain names. [103] The plaintiff claimed that the defendant had registered a domain name that infringed one of the plaintiff's trademarks. [104] On the basis of its analysis the circuit court granted a preliminary injunction against the defendant's use of the allegedly infringing domain name. [105] By way of analogy to the world of brick and mortar storefronts, this case involved two [p77] stores using the same sign on their storefront. Given the close connection between a domain name and a storefront sign the analysis in this case is not surprising. It is clear that, barring the assertion of a valid defense, the unauthorized use of a trademark in a domain name is infringing.

2. On the Face of a Web Page

The use of a trademark on the face of a web page could be analogized to using another's trademark in an advertisement or on a billboard. [106] As one might expect, courts have applied a direct infringement analysis to these types of cases as well. [107] In Goto.com v. Walt Disney Co. [108] the circuit court found that the defendant was guilty of infringing the plaintiff's trademark by using it on the face of its web page. [109] This involved the application of a direct trademark infringement analysis. [110] Trans Union v. Credit Research [111] involved the defendant's unauthorized use of the plaintiff's trademark on their web site. [112] Here again, the district court applied a direct infringement analysis. [113]

These cases both involved the use of logos on the face of web pages. This is easily analogized to using a logo in an advertisement, meaning that application of contributory liability in these instances could be analogized to/with traditional cases of trademark infringement. There is a strong link between common law torts, like unfair competition, and statutory trademark infringement. [114] As will be discussed later, the genesis of contributory liability in trademarks is in the common law. [115] Therefore, when the alleged trademark infringement can be easily analogized to common law forms of unfair competition, it is easy to take the next step and find contributory liability.

However, the use of another's trademark on the face of a web page will not always result in a finding of infringement. The fair use doctrine allows one to use a trademark for descriptive purposes and not be considered an infringer. In Playboy v. Welles, [116] the district court found that the [p78] defendant's use of the "playboy" and "playmate of the year" trademarks on the face of her web page were fair uses. [117] The defendant was a former playmate of the year and thus the district court found her use of the term was descriptive and a fair use. [118]

Cases such as Welles establish that the mere existence of another party's trademark on a web page is not necessarily infringing. This should create some concern when we begin to consider third party liability. Is it reasonable to think that ISPs will understand the fair use doctrine and are they willing to take the chance that they could be wrong? Is it perhaps more reasonable to think that the risk averse ISP will demand removal of any trademark that does not belong to the web site owner in order to avoid potential litigation?

3. Metatags

Trademark infringement in the metatag context is not as easily analogized to traditional forms of trademark infringement. This is primarily because metatags are not seen by Internet users. Trademark is primarily concerned with consumer confusion, and when the trademark is unseen by the Internet user the link between the use of the trademark and consumer confusion is attenuated. [119] Because of this, courts have resorted to a particularized theory of infringement called initial interest confusion. [120]

In Brookfield, the circuit court used a hypothetical situation to illustrate the concept of initial interest confusion. [121] The hypothetical involved Blockbuster Video putting up a billboard on the highway that said "West Coast Video" next exit. [122] When the consumer arrives at the next exit they find only a Blockbuster video store. If they choose to go to Blockbuster they will do so knowing that they are patronizing Blockbuster. However, the concern is that they took the exit off the highway because they wanted to patronize West Coast Video. In Brookfield, the circuit court found that the defendant's use of the plaintiff's trademark in the metatags from its web site was trademark infringement under this theory. [123]

4. Questionable Nature of the Initial Interest Confusion Theory

[p79] Application of initial interest confusion theory of infringement in the metatag context has been accepted in only a few circuits. [124] The theory has been questioned by a number of commentators. [125] It depends heavily on two factors. First, that search engines are in fact manipulated by metatags. [126] Second, that consumers are confused by the appearance of multiple, sometimes irrelevant responses, when they do a keyword search with a search engine. Both of these premises are questionable.

The theory that search engines generally can be manipulated by metatags is debatable. The implication of the Brookfield court is that metatags will influence all search engines. [127] This is not true. Many search engines in fact do not scan metatags at all. [128] Many others give them a low priority in terms of categorizing web sites. [129] Furthermore, the circuit court admits that when a person conducts a search, the resulting list contains the domain names of the various sites. [130] Thus, the consumer has the opportunity to see both the Blockbuster sign and the West Coast Video sign side-by-side before deciding which store to visit. [131]

This fact leads to the conclusion that consumers are not necessarily confused. Most search engine users are well aware of the fact that any given keyword search is going to result in a number of listings that are irrelevant. [132] A better case for initial interest confusion could be made if, as a result of a keyword search, the Internet user was connected directly to a web site rather than being given a list of web sites from which to choose. It is difficult to see the difference between a search result list on the Internet and a person sitting in the parking lot of a strip mall. In both situations the [p80] consumer is faced with a number of options and must take some affirmative step to choose one of them. In both cases the consumer knows full well what store, or web site, they are entering. It is debatable whether the initial interest confusion theory is applicable to search engine manipulation. [133]

With respect to trademark infringement in domain names or on the face of a web page the analogy to direct trademark infringement is strong. There can be little doubt that the statutory prerequisites of use in commerce and consumer confusion can readily be shown. [134] The consumer confusion in these cases is direct. The Internet user views the domain name or the web page with the infringing trademark, leading to confusion as to the source of the goods or services the domain name or web page represents. [135] The initial interest confusion theory relies on a middleman -- the search engine -- to induce the confusion. [136] Because consumers are well aware that search engines do not always produce relevant results the validity of the initial interest confusion theory in the metatag context is debatable. [137]

If we proceed on the assumption that the reasonable consumer can be confused by search engine results then the debatable nature of the initial interest confusion theory is less troublesome. This is because search engines take into account not just metatags, but also the text on the face of web pages and domain names when they produce result lists in response to keyword queries. [138] A trademark on the face of a web page is just as capable of manipulating a search engine as a trademark used in a metatag. [139]

This leads to the conclusion that what the Brookfield court was really after was stopping search engine manipulation, not the theoretical infringement that takes place when a trademark is used in a metatag. [140] [p81] Manipulating search engines looks like unfair competition. The public policy question of whether the law should prevent this kind of activity can be debated. [141] However, there is a threshold question of whether this manipulation actually leads to consumer confusion.

There is no doubt that some people may be confused as a result of search engine manipulation. But the standard for liability for trademark infringement considers the reasonable consumer. [142] If the hypothetical reasonable person is not confused by search engine manipulation there is no reason to allow a suit for infringement. [143] Consequently, there would be no reason to allow third party suits. [144] The linchpin that allows the initial interest confusion theory to operate in the Internet context, and would allow extension of that theory to third parties, is the finding that consumers are actually confused by search engines.

The reason not to extend such liability to third parties is simple -- the theory is too attenuated. [145] The alternative theory is that, with respect to contributory infringement, it makes no difference whether initial interest confusion is a good theory. As a practical matter, the use of a trademark anywhere on a web site has the potential to manipulate search engines, and therefore, in theory, confuse consumers. [146] If most web sites are located on [p82] the Web through the use of search engines, then third party liability should not depend on how the search engine is manipulated, only that it is and that this is the intent of the web designer.

As a general matter, in the likelihood of confusion analysis the intent of the defendant to confuse is often times enough to find infringement irrespective of the other seven factors articulated in Sleekcraft. [147] One might argue that Brookfield was simply an outcome determinative decision. The circuit court, having found intent or bad faith on the part of the alleged infringer, set out to find a theory that would allow them to hold against that party.

If Internet users are in fact confused by search engine results, then whether the trademark was used on the face of the web page or in the metatags should make no difference. [148] The result is still the same. The better argument would seem to be, that absent compelling evidence, that a reasonable consumer would be confused by search engine manipulation, the initial interest confusion theory should not apply.

III. Third Party Liability

A. General Theory

Contributory infringement has its origin in the common law tort cause of action for vicarious liability. [149] The theory is that a third party can be held liable for the tortious conduct of another if that third party purposefully assists the first party in the infringement despite having no direct control over that party. [150] Contributory liability is distinguished from vicarious liability because the later requires the third party to exert some control over [p83] the direct infringer. [151] For third party liability of any kind to exist, there must be a direct infringer. [152] That is, one could not hold the employer liable without first proving that his employee was responsible for some tortious act.

In the realm of intellectual property, there is a statutory cause of action for contributory infringement in patent law. [153] While in copyright the cause of action maintains its common law roots, Congress has created statutory provisions outlining when a cause of action may be brought against an ISP. [154] However, there is no statutory cause of action in the Lanham Act for contributory infringement of trademarks. [155] Because trademark law has common law origins, one might sue for contributory infringement on the tort law theory of vicarious liability. [156] Some discussion of the scope of contributory infringement in patent and copyright law may help inform the debate about its proper scope in trademark.

B. Patent Origins of Contributory Infringement Theory

Patent infringement is governed by statutory law. [157] Under the statute, "[w]hoever actively induces infringement of a patent shall be a patent infringer." [158] A contributory infringer is one who "offers to sell or sells . . . or imports . . . a component of a patented [product] . . . knowing the same to be especially made or especially adapted for use in an infringement." [159] The standard for contributory infringement is "knowing," i.e. actual knowledge. [160] As will be discussed later, this is a stronger knowledge requirement then has been demanded in the trademark context. [161] In addition, the patent doctrine is limited to products. [162] As will be discussed, in the trademark context, the doctrine of contributory infringement began with products but has since been extended to include services. [163]

C. Contributory Infringement in Copyright

[p84] Contributory infringement in the copyright context is controlled by the Digital Millennium Copyright Act (DMCA), [164] at least for the purposes of this Article in considering ISP liability. Prior to the passage of the DMCA, courts had found third parties liable for contributory infringement in the copyright context. [165] In fact, the leading case in this area seems to have prompted congressional action. [166]

In Religious Technology Center v. Netcom, [167] the district court addressed the issue of whether "a computer bulletin board service (BBS), and the large Internet access provider that allows that BBS to reach the Internet, should be liable for the copyright infringement committed by a subscriber of the BBS." [168] The district court admitted there was no statutory authority for third party liability in the copyright context. [169] Rather, it contended that vicarious liability is found "in virtually all areas of the law." [170] Finding "contributory infringement is merely a species of the broader problem of identifying the circumstance in which it is just to hold one individual accountable for the actions of another." [171]

The rule the district court offered is that liability for contributory infringement is established when the defendant "with knowledge of the infringing activity, induces, cause materially contributes to the infringing conduct of another." [172] In Netcom, the plaintiff notified Netcom of the alleged infringement. [173] For the district court, this letter was possibly enough to give Netcom the requisite knowledge. [174] Netcom argued that the [p85] alleged infringement must be unequivocal before they can be considered to have knowledge. [175] The district court found this too strict a standard, but offered that a defendant's claim of lack of knowledge might be found reasonable in certain instances, even though copyright infringement had actually occurred. [176]

On the issue of whether the defendant had 'induced, caused or materially contributed' to the infringement, the district court found that Netcom did not completely relinquish control of their network when they granted access to a customer. [177] The district court stated that it is fair to assume Netcom could have taken simple measures to prevent further distribution of the plaintiff's copyrighted works. [178] The implication of the focus on the control element is that an ISP is responsible for the content of anything on a web site it hosts. [179] While in this case the infringing material was on a bulletin board and arguably easy to remove, this rationale would not extend to a web site where potentially the only way to excise the infringing material would be to close down the entire web site. [180]

Limited to these facts, the implications of the Netcom decision could be devastating to web-based publishing. If any person publishes material on the Web containing a copyright notice, that individual's ISP could be liable for contributory infringement, if the holder of that copyright provides some form of notice to the ISP. In this case, a letter appears to be enough, but the district court does not spell out the notice requirements. [181] It is also unclear what a copyright holder would have to do if the material did not contain a copyright notice. [182] A similar situation exists in the trademark context. [183]

D. Contributory Infringement in Trademark

[p86] There is no statutory rule regarding contributory liability in trademark law. [184] Courts have, as in the copyright context, fashioned a rule based on the common law. [185] However, unlike in the copyright context, contributory infringement in trademark law is not limited by a statute like the DMCA.

1. Origins

In Inwood Laboratories v. Ives Laboratories [186] the U.S. Supreme Court held that:

if a manufacturer or distributor intentionally induces another to infringe a trademark, or if it continues to supply its product to one whom it knows, or has reason to know is engaging in trademark infringement, the manufacturer or distributor is contributorially responsible for any harm done as a result of the deceit. [187]

The Inwood case involved the manufacturing of generic drugs. [188] Inwood was producing a drug in generic form after the Ives patent on the drug had run out. [189] However, Inwood made its pills look exactly like those of Ives. [190] The U.S. Supreme Court found that this could be contributory infringement if Inwood knew pharmacists would take advantage of the fact that the pills looked alike and fill prescriptions for Ives' pills with the cheaper version produced by Inwood. [191]

Inwood involved a product and is therefore closely analogous to the patent theory of contributory infringement. [192] However, the Inwood concurrence points out that the rule established by the majority expands the knowledge requirement normally applied in the patent context. [193] There, contributory infringement required actual knowledge that others would use [p87] the product for infringing purposes. [194] The Inwood formulation of the knowledge requirement says that the contributory infringer must "know or have reason to know." [195] This effectively means something less than actual knowledge can suffice. [196]

2. Expansion

Inwood was expanded to cover services as well as products by Hard Rock Cafe v. Concession Services. [197] In that case, the plaintiff brought a claim of contributory infringement against the operator of a flea market that sold space to the defendant. [198] The circuit court agreed that the flea market operator was like a landlord. [199] As such, they had a common law duty to control persons they permitted on their premises. [200] The circuit court distinguished this case as one not involving vicarious liability because the defendant and the infringer did not "have an apparent or actual partnership, [or] have authority to bind one another in transactions." [201]

The case was remanded, however, on the issue of the flea market operator's contributory liability because the lower court had not made sufficient findings to satisfy Inwood's "know or have reason to know requirement." [202] The lower court had simply found that the flea market operator was "willfully blind" to the possibility that infringing activity was taking place. [203] This finding appears to be based on the operator's failure to take precautions against infringement, rather than any findings as to its state [p88] of mind. [204] While the circuit court agrees willful blindness is akin to actual knowledge, they offer that the lower court has not pointed to facts that establish willful blindness. [205]

The circuit court further offers that the flea market operator has no affirmative duty to take precautions against the sale of infringing material. [206] The circuit court claims the reason to know portion of the Inwood rule simply means that the defendant is required "to understand what a reasonably prudent person would understand, [and] it does not impose any duty to seek out and prevent violations." [207]

This falls short of the proper analysis. In Inwood, the U.S. Supreme Court did not require Ives to prove that Inwood had actual knowledge of the infringing behavior of the pharmacists. [208] Rather, it seems the "reason to know" addition to the rule was there specifically to allow Ives to succeed despite Inwood's lack of actual knowledge. [209]

It is disingenuous of the Hard Rock court to say this portion of the test imposes no duty on the flea market operator. [210] It was a necessary element in Inwood. How can one say that a reasonably prudent person would know the activity in front of them is infringing, but then say they have no duty to investigate further? This implies that so long as the third party distances themselves from the infringing behavior, and takes steps such that they will not gain actual knowledge that infringement is taking place, they are free from liability. [211] This is exactly the kind of behavior the willful blindness standard is intended to prevent. [212] This behavior is exactly why courts have equated "willful blindness" with "actual knowledge." [213]

The real issue created by the "reason to know" standard is the degree of investigation required by the reasonably prudent person who reasonably [p89] believes infringing activity is taking place. [214] When the lower court found that the flea market operator in Hard Rock had not taken "reasonable steps to detect or prevent" [215] the infringing behavior, the lower court was saying they did nothing when surely they were obliged to do something. [216] The lower court is saying the "reason to know" standard simply imposes on the party with reason to believe infringing behavior is taking place, the obligation to undertake a reasonable investigation to confirm or deny that belief. [217]

In Fonovisa v. Cherry Auction, [218] the Ninth Circuit Court of Appeals applied both Inwood and Hard Rock in another case involving a flea market operator. [219] There is no mention in the Fonovisa decision of Hard Rock's distinction regarding the "reason to know" standard. [220] Clearly, this is factually the same case as Hard Rock. [221] Yet this court recognized that the Inwood standard does in fact impose some duty on third parties to investigate trademark infringement, saying that "a swap meet can not disregard its vendors' blatant trademark infringements with impugnity [sic]." [222] After Fonovisa there is a viable cause of action for contributory infringement against a service provider. This could be extended to an ISP.

3. Application to the Internet

Very few cases are applying these decisions to the Internet context. In Lockheed Martin v. Network Solutions, [223] the district court found there was no viable claim for contributory infringement in a case involving an allegedly infringing domain name. [224] This decision is case specific for two reasons. First, the defendant Network Solutions, Inc. (NSI) only registered the trademarked domain name. [225] They did not directly profit from the use of those trademarks, unlike the defendant flea market operators in Hard [p90] Rock and Fonovisa. [226] Second, NSI is a unique defendant because they have a government contract for the registration of domain names. [227] The case might be different if it involved an ISP that allowed a web user to make use of an infringing domain name in commerce. [228] The ISP, unlike NSI, shares in the profit made through the use of the infringing domain name just as the flea market organizer in Hard Rock profited from the sale of T-shirts. [229]

In a footnote in Lockheed Martin, the district court hypothesizes that contributory liability as applied in Hard Rock and Fonovisa would be unlikely to apply to an ISP. [230] The district court theorizes that those cases do not apply to Network Solutions because Zeran v. America Online, [231] had noted that Congress had created tort immunity for ISPs in 47 U.S.C. § 230(c). [232]

In Gucci v. Hall, [233] however, the district court found this argument unavailing, pointing out that in 47 U.S.C. § 230(e)(2), Congress also said that any limitation of liability was not to be implied where such limitation would "limit or expand any law pertaining to intellectual property." [234] In fact, the district court cites the legislative history of section 230 that indicates that its purpose was only to immunize ISPs against "defamation and other non-intellectual property, state law claims arising from third party content." [235]

The Gucci case involved an allegation of contributory infringement against an ISP because of the infringing activity taking place on a web site that the ISP hosted. [236] This case is a recent development and has not been followed by other courts. However, it opens the door to contributory trademark infringement liability for ISPs. The remaining question then is whether this is good public policy.

IV. Public Policy Considerations

[p91] The rationale for creating trademark protection can be found in the balancing of a number of competing interests. On the one hand, legislators are reticent to grant monopoly powers to a business and, on the other hand, they want to make it easier for consumers to find products in a store. [237] The balance struck by trademark law is to grant a sliding scale of monopoly protection depending on the degree to which the mark can be identified with a particular product manufacturer and therefore serve the interest of lowering consumer search costs. Allowing trademark holders to bring suits for infringement gives them the ability to protect their trademark. This consequently serves consumer interests in that if there are fewer unauthorized uses of trademarks, there is less chance that consumers will be duped into buying inferior or different products than they intend.

The question is whether allowing trademark holders to sue third parties on a contributory liability theory serves the interest of lowering consumer search costs. On the one hand, such a rule is sensible because ultimately it is the actions of the third party that may most seriously damage the value of trademarks to consumers. An individual brick-and-mortar store may infringe a trademark, but the scope of that infringement is limited by the physical realities of the world. The same act on the Internet has a much wider scope and therefore more potential to damage the value of the trademark being infringed. However, in order to infringe a trademark on the Internet one must gain access to the network. [238] Thus, the need for a web host such as an ISP.

Allowing third party suits could allow trademark holders to put a stop to the large volume of infringing activity possible on the Internet by cutting it off at a single convenient point. Without such a rule, trademark holders might simply have to allow some infringing activity to take place if the costs of investigation and litigation are more than the benefit of curtailing the infringing activity. Contributory infringement suits would also allow trademark holders to get at a number of small scale infringers in a cost-effective manner.

While third party suits create a cost advantage for trademark holders, there is a concomitant cost disadvantage for ISPs. In a world where direct infringement is the only cause of action, trademark holders must go to the expense of not only suing infringers, but also the expense of finding them. When contributory liability is a cause of action, a disincentive for trademark holders to conduct searches for infringers and at the same time ISPs are [p92] encouraged, if not forced, to take over that searching activity to avoid potential litigation. In determining how best to balance these costs, it is instructive to see how Congress dealt with contributory liability in the area of copyright law.

A. DMCA Safe Harbor Provisions

The Senate judiciary committee report on the DMCA pointed out that a number of cases had found ISPs liable for contributory copyright infringement. [239] Rather than rewriting the law, the committee chose to create "safe harbor" provisions for certain common ISP activities. [240] The DMCA contained a limitation on liability for contributory infringement in the realm of copyrights. [241] The statute reads:

A service provider shall not be liable for monetary relief for infringement of copyright by reason of the storage at the direction of a user of material that resides on a system or network controlled or operated by or for the service provider, if the service provider:

(A) (i) does not have actual knowledge that the material or an activity using the material on the system or network is infringing;

(ii) in the absence of such actual knowledge, is not aware of facts or circumstances from which infringing activity is apparent; or

(iii) upon obtaining such knowledge or awareness, acts expeditiously to remove, or disable access to, the material;

(B) does not receive a financial benefit directly attributable to the infringing activity, in a case in which the service provider has the right and ability to control such activity; and

(C) upon notification of claimed infringement as described in paragraph (3), responds expeditiously to remove, or disable access to, the material that is claimed to be infringing or to be the subject of infringing activity. [242]

In practical terms, an ISP is unlikely to have "reason to know" given the volume of material they deal with. However, the statute allows that an ISP can be made aware of alleged infringing activity through notification. [243] Once notified the ISP would certainly have reason to know of infringing behavior. In that instance the statute requires them to take steps to remove the material. [244]

[p93] The notification provisions in the statute require the holder of the copyright to identify the copyrighted work, [245] identify the material alleged to be infringing and where to find it. [246] The copyright holder must also assert that they have a good faith belief that the use of the material is infringing. [247] Finally, the entire notification must state that it is true and accurate under penalty of perjury. [248]

Nothing in the statute requires the ISP to make any judgements regarding whether the allegedly infringing material is a fair use. [249] If the holder of the copyright is willing to assert a good faith claim of infringement, then the ISP is required to take steps to remove the material. This operates as a sort of preliminary injunction in that the alleged infringer will now have a chance to asserts their rights in court. The ISP can avoid liability completely and as a practical matter has no affirmative duty to search for potentially infringing material on its system. [250]

The statute makes good sense for two reasons. First, it puts search costs on the holder of the right. The holder of a copyright is obligated to incur the expense of enforcing his rights. The ISP has no affirmative obligation to seek out potential instances of infringement. [251] Rather, they have only the obligation to take steps to eliminate infringement once they become aware of it. [252]

Second, the obligation for determining whether the alleged infringement is in fact infringement is placed on the holder of the right. [253] They are the ones who must attest, under penalty of perjury, that the alleged copyright infringement is in fact infringement. [254] This obligates the holder to weigh [p94] the possibility of a fair use defense or to determine if in fact the alleged infringer is a licensee. [255]

Under this statute, Congress was willing to allow some non-infringing material to be excised from the Web on the basis of a good faith belief that it is infringing. [256] This leaves final determinations of fair use to the courts where they belong. These provisions impose the same threshold knowledge requirement in the copyright context that Inwood does in the trademark context. [257]

The House report explains the knowledge requirement, saying it "is met either by actual knowledge of infringement or, in the absence of such knowledge, by awareness of facts or circumstances form which infringing activity is apparent." [258] The 'awareness of facts or circumstances' language is described as a "red flag." [259] The report says "a service provider need not monitor its service or affirmatively seek facts indicating infringing activity. However, if the service provider becomes aware of a "red flag" from which infringing activity is apparent, it will lose the liability limitation if it takes no action." [260]

The House describes the determination of awareness on the part of the ISP as both a subjective and an objective test. [261] That is, a court is to determine first whether the ISP was subjectively aware of the infringing behavior, and then objectively whether those facts would lead a reasonable person to believe infringement was in fact taking place. [262]

This is clearly the same standard articulated by the U.S. Supreme Court in Inwood. [263] The difference is a formalized notice procedure. With respect to the DMCA however, if a notice to an ISP does not meet the statutory requirements, no action on the part of the ISP is necessary. [264] Under the [p95] Inwood rule since there are no formal notice procedures, any notice might be enough to obligate the ISP to take some action. [265]

B. Distinguishing Trademark and Copyright

Although trademark and copyright are both forms of intellectual property, they are distinguishable both in whom they protect and the overall public policy goals they seek to reach. Because of these differences, one must consider whether it is appropriate to treat contributory infringement in the copyright realm in the same way as one might in the trademark context. Copyright law exists primarily to promote the wide dissemination of information. [266] Copyright law covers a wide variety of works so long as they can be "fixed in a tangible medium of expression." [267] It creates legal protection for such works on the theory that if creators have some legal protection for their work they will be more likely to allow public access to it. [268] This serves to increase the level of knowledge and information available to everyone. [269]

Under one theoretical underpinning of copyright law, copyrights primarily benefit the holder of the right while creating an ancillary benefit to the public. [270] Conversely, trademarks primarily benefit consumers while [p96] creating an ancillary benefit to the trademark holder. [271] A key distinction between copyright and trademark law is that the copyright holder can benefit from the law without serving the purposes for which the right was created. [272] Copyright law has no requirement that the copyright holder actually disseminate the information or art subject to the copyright. [273] A copyright holder may effectively sit on his rights. [274] Conversely, for the trademark holder to maintain his rights he must continue to use the trademark in commerce. [275] Failure to use a trademark in commerce can result in abandonment and loss of trademark rights. [276] While a trademark can potentially last forever, that long life comes at a price to the holder.

In considering whether copyright and trademark law should be treated in the same fashion with respect to contributory infringement, it is important to consider the ease with which infringement can be spotted. Copyright rights are more broadly based than trademarks. The copyright holder not only has rights in the actual words that make up a manuscript but also in derivative works based on the original. [277] There is no analogous right in trademark.

However, the fair use doctrine applies to both copyright and trademark. [278] This means that even in cases of infringement that might appear clear, there may not in fact be infringement. In trademark law there is the additional complication that confusingly similar marks can be used simultaneously if they are used in separate product markets or even for the same products if used in different geographic locations. [279]

[p97] There are a number of dissimilarities between copyright and trademark law leading to the conclusion that they should not necessarily be treated in the same way in terms of contributory infringement.

C. Rationale for Allowing Contributory Infringement in Trademark

The best reason to allow a cause of action for contributory infringement in trademark law is that it will help to promote the goals of trademark law. If through the mechanism of contributory infringement the law can help to prevent consumer confusion and reduce consumer search costs, then it is a good rule.

The greatest benefit to trademark holders is that they would be able to prevent infringement on a large scale without incurring as many costs. The World Wide Web is a vast marketplace. New web sites are added to that marketplace everyday. For the individual trademark right holder to police the entire Web for infringing behavior would be an enormous cost. Unlike localized infringement that can occur in the brick-and-mortar world infringement on the Web is potentially far more harmful to business.

It would certainly be much easier for those allowing access to the Web to police for infringing behavior than for the individual trademark holders to do so. Many large companies like Playboy Enterprises have the capability and motivation to conduct such policing on their own. But the vast majority of trademark holders do not have such resources. The result for them is that a certain degree of infringing behavior has to be tolerated. While this rule may result in some loss of the ISP's business, the larger concern is the degradation of trademarks generally. [280] If wide spread infringement is allowed to become commonplace, the value of trademarks to consumers is reduced.

D. Arguments Against Allowing Contributory Infringement in Trademark

The primary argument against allowing a cause of action for contributory infringement of trademarks is that it imposes search costs on ISPs. [281] Trademark rights, while serving some public good, are generally thought to be personal rights of the holders. [282] As such, it is the responsibility of the holder of the trademark to enforce it. [283] That means, the trademark holder [p98] must actively seek to protect the mark from infringement. [284] Laches theory has been applied in some instances when courts found that a trademark holder sat on his rights too long before asserting an action for infringement. [285]

Placing the obligation for discovering infringing behavior on the ISP increases the strength of trademark rights. That is, we can assume the ISP will conduct a cost benefit analysis of the cost of potential litigation versus the cost of losing some business. If the cost of litigation is higher then the risk, a neutral ISP should opt for forcing the web site it is hosting to remove the potentially infringing trademarks.

In all likelihood, this means larger corporations with the resources to bring costly lawsuits will be able to disproportionately influence the use of their trademarks by others. Whereas the small company without the means to bring a costly lawsuit will have to suffer not only infringement of their marks [286] but will potentially be prevented from making fair use of other marks. Many potentially infringing uses may in fact be fair use.

A reduction in the fair use of trademarks means that the property right created by the trademark is stronger. Fair use is part of the trade-off of trademark law. While patent law creates an almost exclusive monopoly for patent holders, [287] trademark law allows only for a limited monopoly. [288] Fair use is part of that limitation. A reduction in the level of fair use of trademarks means a corresponding increase in the strength of trademark rights. [289]

[p99] Fair use in trademark law has an important benefit for consumers -- it allows for product comparisons, among other things. [290] If product comparison is curtailed, particularly between the smaller company and the large corporation, consumers do not benefit. This not only creates a barrier to market entry for smaller companies, but it creates a situation in which, at least in e-commerce, larger companies need not work as hard to improve their products for fear of losing market shares to newer companies.

Although search costs are increased for ISPs in this scenario, the increase serves a public good. It is not simply that they are spending time and money doing something that rights holders would otherwise do themselves. Rather, it is also that they are performing a public service by helping to ensure the ability of trademarks to continue to reduce search costs for consumers. [291]

V. Conclusion

This Article seeks to establish that, under current precedent in the federal courts, there is a cause of action for contributory trademark infringement against an ISP. This cause of action is limited to the infringing use of a trademark on the face of a web page and in the metatags of a web site. This Article has argued that application of contributory infringement liability in the metatag context, where the theory of infringement is based on initial interest confusion, stretches trademark doctrine too far. The possibility of contributory liability on the basis of such a theory should encourage courts to consider the validity of the initial interest theory in the first instance.

This Article has sought to deal with whether contributory infringement in the trademark context is good public policy. The use of trademarks on the Web could be vast, and placing the burden for policing cyberspace on ISPs demands too much. The notice requirements spelled out in the DMCA should be applied by statute to trademarks as they have been in the copyright context. [292] This would grant trademark holders the ability to [p100] enforce their marks use on the Web in instances where direct action against the infringer is either too costly or inefficient.

The state of trademark law has yet to reach the extent that copyright law had prior to the passage of the DMCA. However, the Gucci decision has opened the door to the same types of suits in trademark law that Religious Technologies did in the copyright realm. Contributory trademark infringement suits have the same destructive potential for ISPs as they did in the copyright realm. Congress should seize the opportunity to limit contributory infringement for trademarks on the Internet.