One solution to the establishment of a liability law is to simply not create any law at all: in each case a judge would determine, ad hoc, whose fault it is. However, in many cases the concept of fault is not entirely clear. Furthermore, the failure to establish a standard would cause confusion and uncertainty amongst the industry. This confusion might hinder investment as well as allow companies to produce highly unreliable software.
Central to the creation of any liability law is the concept of incentives. The goal is to create the right incentives for all the parties involved. The creation of an optimal incentive scheme will lead to optimal behavior. The optimal behavior under a software liability law is as follows: software producers are producing software that is both safe and affordable to consumers. In other words, there is a natural tradeoff between safety and affordability. If making risk-free software entails raising the price of the software to astronomical levels, then consumers would probably prefer taking the risk with the buggy software instead of spending a fortune on the error-free version. The goal of a software liabililty law is to induce producers to produce at this optimal mix of safety and price.
This section uses economic analysis to analyze the liability laws. The law and economics field has build a comprehensive framework from which to analyze liability rules. Though the framework is economic in nature, it is built upon the ethical system of utilitarianism.
What are the ethical foundations of the concept of liability? Ethics can be associated with liability in two ways, one normative and one empirical. Normatively, we can judge a liability rule as "ethical" if it places responsibility on a party that we feel should be held responsible for a given situation. Liability is fundamentally about assigning responsibility. Empirically, we can judge liability rules in a economic/utilitarian framework. In this sense, a choice of a liability rule is ethical if it maximizes social utility. This second approach may appear to be imprecise and too general, but the growing "law and economics" literature has generated important conclusions about difffering liability rules.
The purpose of this section is to give an overview of the Òlaw and economicsÓ literature with respect to liability. The ideas presented here are invaluable when thinking about software liability rules. Furthermore, these ideas are prominent in the legal thinking of many judges today; any discussion of software liability must include these general ideas.
What follows is an economic model of product liability. The explanation of the model will be intertwined with its application to software.
The goal of a product liability rule is to maximize the total wealth of all parties involved. An optimal product liability rule should maximize the sum wealth of the producer and all the consumers.
This does not necessarily mean that we want producers to produce 100% safe products. For example, most people would prefer a very safe car that cost $30,000 over a perfectly safe car that cost $300,000. People often voluntarily assume risks because it makes them better off to do so. There is an optimal level of safety that will maximize the hapiness of both the producer and the consumers. The goal of any liability rule is to induce the producer to produce at this optimal level of safety.
How can we identify the optimal level of safety. Fortunately, the wonderful forces of the market will go along way to helping attain this balance. The way the market takes care of this is simple, but we must make one assumption: both the producer and the consumer have perfect knowledge about the risks associated with the given product (they know the probabilities of failure and the losse associate with failure). Given this knowledge, they accordingly make decisions about which products to buy. For example, a consumer of software might go through decision calculus as follows: Product A is $100 more expensive than Product B, but there is a 50% chance that Product B will fail and cost me $400 in damages. Therefore, in the long-run it is cheaper for me to bring product B.
Thus in a market with perfect information, it is easy to arrive at an optimal level of software safety. Unfortunately, the "market" with respect to product safety is highly deficient. The classic market assumes that the producer and consumer have perfect information and knowledge about the product. This assumption is far from true in any real-world market and this assumption is especially problematic when dealing with a complex entity like software. Accordingly, we turn to the issue of product information.
Central to creating the right liability rule is the concept of information. There are two important questions:
1. How much information does the producer know concerning the safety of the product?
2. How much information does the consumer know concerning the saftety of the product?
If we assume that both the producer and the consumer have perfect information about the product, we arrive at an interesting result: it doesnÕt matter which liability rule is chosen because all will result in the optimal level of safety. This is very easy to prove mathematically, but a verbal explanation will suffice. Consumers know what level of safety they want, and they also know the level of safety associated with the product. Consumer's will therefore factor the risk they face in with the actual face value of the product. Consumers will therefore prefer the safer of the two comparable products. In order to sell products., manufacturers are forced to make safer products. Because consumers have perfect information, manufacturers are forced to internalize the harm they create.
Under imperfect knowledge/information, problems arise. If consumers don't know the reliability of the product, then producers have no incentive to make the product liable. Thus, we now need a liability rule that makes the producer want to produce safe products.
This section tries to answer the following question: who should be held liable?
The real question is this: who has the power to prevent the harm from occuring? If a car's brakes fail while the automobile is being driven 200 miles per hour down a slippery road, should the car manufacturer be held liable for the failure? Of course not, for the driver was not taking an appropriate level of care. Liability rules need to be flexible to the issue of differing levels of care. Holding the manufacturer liable in all cases will cause the consumer to behave inefficiently. After all, why should the consumer exercise an appropriate level of care when the manufacturer will be completely liable for any mishap?
The law and economics literature provides fascinating insights into which one of these rules are optimal under different situations. The choice of rules hinges on the following key factors:
Information: If there is any product whose inner workings the average consumer knows nothing about, it is computer software. Software engineering is beyond the realm of most people. However, making the software company liable for all software failures might discourage companies from attempting new and innovative projects.
Negligence: On the other hand, attempting to define negligence for the act of software development is naturally problematic. One related concept to negligence is that of proximal cause: how related was the programmer's error to the failure as a whole? Was there intervening circumstances, beyond the control of the programmer, that triggered the failure?
The goal of liability laws is to induce proper behavior on behalf of both the consumer and the producer. With respect to the producer, this means producing safe and reliable software up to an optimal point. With respect to a consumer, this means using the product with appropriate care and caution. Any efficient liability rule must meet these criteria.