The use of computer programs to automatically access thousands of websites with blinding speed is not a new phenomenon. These programs, referred to as robots, use the same protocols that everyday web browsers use to access web servers to request pages, and then manipulate the data they receive to perform a wide-array of functions. Companies that maintain public search engines such as Google and Yahoo! use robots to build their search databases and run algorithms over the pages they collect to increase the relevancy of searches run against these data. Other companies have built robots that will mine through a website’s price listings to get a complete inventory of what that site is offering for sale and at what price. Robots can also mimic human input on chat boards, instant messaging programs, and other online communities. They can accomplish most any task that a live human could theoretically do on the Internet.
Depending on the nature of the robot, website owners often have different reactions on this automated access. A single robot running on one computer is capable of generating far more requests on a web server in just a few seconds than most people would generate against that server in an entire day. This barrage of requests could cause an inadvertent denial-of-service attack by so loading the server with requests that nobody is able to access the site at all. While this would be an extreme case, it is entirely possible. Companies often find themselves devoting resources to combating this kind of excessive access both by trying to block such requests and by purchasing more servers to handle the requests when such robots manage to pass through whatever screening filters are put in place. However, most attempt to actually block robot requests are at best short-term solutions since it usually does not take long for person running the robot to notice the new safe guards and then program around these restrictions. The company running the website and the company running the robots end up engaging in a little arms race with each side gaining momentary advantages over the other until the other side adjusts and adapts.
Not surprisingly, this sort of behavior by robots has generated its fair share of lawsuits – the most famous of which was EBay, Inc. v. Bidder’s Edge, Inc. Bidder’s Edge was using a robot to generate a catalogue of the items up for auction on EBay, and Bidder’s Edge would publish that information on their own website. The court granted an injunction preventing Bidder’s Edge from accessing EBay’s website under a trespass to chattels theory. The court found that the traffic Bidder’s Edge was generating on EBay’s website constituted only around 1-2% of the total traffic on EBay, which may not have resulted in any damage or interference with EBay’s property rights in their site. However, the court reasoned that failure grant the injunction would encourage other companies to engage in similar practices, which could result in a real interference with EBay’s personal property. In effect, the court allowed the potential snowballing of robot activity to suffice as potential damages that would warrant granting the injunction. Courts around the country have followed this logic in granting similar injunctions.
The threat of snowballing robot activity as potential damage that would justify an injunction under a trespass to chattels theory may have met its logical outer limits in Intel Corporation v. Hamadi. At first glance, this case does not seem all to similar to the robot cases covered under EBay since this case involves a disgruntled former employee of Intel sending out occasional mass emailing to current Intel employees; however, the reasoning behind EBay and this case was essentially the same. The California Supreme Court looked at many of these robot cases and was able to distinguish them from Hamadi’s behavior. Hamadi only sent six different emails to thousands of employees over a 21-month span. This pales in comparison to the millions of requests often generated by robots, and it represented a negligible amount of the traffic in Intel’s email system. Hamadi also removed any recipient from his email list when so requested. The court refused to find that disallowing Hamidi to send the emails would encourage others to engage in similar mass-emailings. The Court was also worried that Intel was more concerned about the content of Hamadi’s emails rather than the emails themselves. Based on all of these factors the Court refused to grant relief to Intel.
The next test for the courts may come when a robot takes an action that splits the difference between these two scenarios. For example, many of the online communities at Yahoo! have robots that comb through various chat rooms and automatically make posts that often amount to little more than advertisements. There are also robots that will add and remove themselves and their virtual profiles from different interest group associations in an effort to keep their names at the top of various “Newest Members” lists. Yahoo! almost certainly objects to the content that these robots produce since many are advertising pornographic websites under the guise of being a real user. While many of these robots can be detuned to the point where any individual robot constitutes a negligible amount of traffic (and often are to run under the nose of many screening filters Yahoo! uses) and it could be argued that they are using the board in a manner consistent with the intended purpose of the website, the aggregate effect of these robots is significant. Only should Yahoo! or another company in a similar position decide to stop engaging in arms race type battle that the website and these robots often engage in to offset each other’s effects and take this battle to the legal arena will we see where the courts will fall on this issue.
