Defining a theoretical strategy to interpret software in court cases has become a common dilemma as judges are increasingly presiding over cases that involve computer programs. James Grimmelmann, the first Cornell Law School professor based at Cornell Tech, offered a theoretical model to determine the legal effects of computer programs in a public lecture on October 24 at Gates Hall. His talk was the sixth in Cornell University's Tech/Law Colloquium series, which explores emerging technologies and the legal and policy challenges surrounding them.
Interpreting the legal intent of software is a complex process because court cases involving software often do not align with existing laws. As an example, Grimmelmann discussed a case in which two video poker players took advantage of a software bug on a Game King machine in two casinos to change the payout multiplier and increase their winnings tenfold. The players were indicted for hacking into the software on video poker machines in Nevada and Pennsylvania in 2013, but the U.S. attorney later dropped the charges. "The defendants raised a serious question about whether it was actually a crime or not," Grimmelmann said.
Cases such as the video poker prosecution that involve computer programs present a dilemma for the courts. While courts are familiar with wills, contracts, statutes, regulations, and other legal texts, it is not clear how courts should handle texts written in programming languages like C and Python.
"The question that we're concerned with is, what do we do when that text consists of a computer program, when it's software shaping people's rights?" Grimmelmann said.
What differentiates software is that a legal text's legal effects are its only effects. "But software is primarily a functional artifact," Grimmelmann said. "It drives cars, turns lights on, modifies account balances or shows or hides data. What it means legally is derivative of what it does." Another key difference is legal texts are addressed to people, such as citizens, counterparties or judges, and therefore their meaning to people is critical. But software is addressed to computers, consisting of a series of commands to execute, Grimmelmann said.
For software, the legal meaning is grounded in its functional meaning, even though that meaning can be complicated by the fact that computers often have bugs. Nevertheless, most of the time, Grimmelmann said, running a program on different computers yields the same results. He proposed two interpretive strategies for providing a legal interpretation of software: its literal functional meaning, or what a correctly functioning computer would do; and its ordinary functional meaning, or what a reasonable user would understand the programmer to have meant.
"Which of these theories is the right one for a given context depends on the context," Grimmelmann said. "In general, we're going to want literal functional meaning in cases where it's really important to have precise agreement. Ordinary meanings make more sense in cases where users are involved and users are making choices about what to do with the program."
In the video poker case, Grimmelmann said, ordinary functional meaning is a better theory to analyze the players' hacking of the Game King machine. "I think we can all recognize that what the players found was not part of the rules of online poker," he said. "The payout multiplier is a bug that will be fixed in the next version of the software on that machine as soon as the gaming commission approves it."
Whether the hacking is a crime depends on the mental state of the players and whether they intended to defraud the casino, he said. "Whoever programmed the Game King machine probably didn't intend for them to be able to quit and return to the game and change the payout button," he said. "That was a bug, and it was not an authorized use."