I can draw two conclusions about Human Factors (HF) now that I know it is such a complex field. First, people will not consider HF seriously until they can quantify the benefits of paying attention to it. Second, HF is in desparate need of a simpler model.

My first conclusion was that HF cannot be taken seriously until its benefits can be quantified. Convincing a company to apply HF is a little like convincing a company to emphasize customer service. Both HF and customer service are hard things to measure. It is hard to know when you are doing a good job with either. It is even hard to know what IS a good job. Moreover, most people seem to believe that customer service and HF are somehow good for business. The problem is just with the "somehow". How much good is it? Until this question can be solidly answered, probably only big companies that can afford to go on "somehow"-kinds-of-feelings will address HF.

My second conclusion was that HF is in desparate need of a simpler model. Recall that HF deals with human beings which are very complex animals which cannot be modeled in the standard way. The standard way of modeling something is to model the interactions and states of its underlying components. There are too many interactions and states to account for with humans.

A better way to deal with complexity is to look at a bigger scope. Instead of being concerned with humans at a micro level, let's concern ourselves with humans at a macro level. Economics is a good exmample of this. Economics looks at human behavior on a macro level, and it does a decent job at modeling what people do (for example the supply & demand model). If economics tried to look at human behavior on a micro level it would face the same problems of complexity that HF currently faces. I say we should develop a macro model instead of a micro model.

Such a model would abstract the complexities that today's models suffer. Our report contains a list of 50 Human Factors--there are plenty more. Current models of HF try to address human factors one at a time. Rules are developed for how to address each human factor, but with so many human factors these rules tend to lead to conflicts. Instead of concerning ourselves with so many human factors, we can take an empirical approach.

A feedback loop is the key. This model suggest you should develop your product, and then see if people are able to use it well. Use a feedback loop to see what improvements can be made to better address Human Factors.

Here is an example of how this would work with a program like a word-processor. The word-processor would have built into it a mechanism for tallying information about how the user is using the software. It would count error messages, time spent on certain tasks, and all sorts of other interesting things to know. This information would then be uploaded to a server. The server would have analaytical programs crunching through terabytes of usage data from all users of the software. The server could draw conclusions about how well users are using the word-processor. From these conclusions, designers could tell how much time is spent on certain tasks and which actions tend to generate errors. The designers could stream line these elements of the software in their next release. This model frees the designer of the complexities of current HF theory. Current models of HF only speculate about how human beings are, but the feedback model gets to the heart of the matter: how well can users use software.

Finally, the feedback model allows us to quantify the benefits of paying attention Human Factors in design. The feedback model allows us to correlate sales with effort put into Human Factors design. Since initial releases have less HF design, they serve as a basis of comparrison. I am not claiming that this is a perfect measure of the benefits of HF design, but it is at least as good as any other marketing information system.