The accuracy issue...is it real?

Because of the laboratory and "scientific" origins of process automation, we tend to pay a lot of attention to issues like accuracy. A device that is more accurate is inherently better. But is this really true? I submit that it isn't true. While there are some devices, and some applications, where the maximum obtainable accuracy is critical for proper process control and manufacturing performance, those applications are relatively few. What is really necessary is repeatability, durability, simplicity, and ease of use and service. Sometimes, these are antithetical to maximum obtainable accuracy. I know this is partly heresy. I know that field device vendors like to differentiate their devices by accuracy. I also know that David W. Spitzer (http://www.spitzerandboyes.com) and I have spent the last several years writing books (the Consumer Guide series, available from http://www.isa.org and amazon.com) that rank field instruments, and we're continuing to do so. One of the rankings we use is accuracy. This is not to say that I am being hypocritical, because accuracy is a valid measure. But when we look at the process of the future, we need products that have significantly more durability and simplicity than those of the present day. My ideal sensor works reasonably well, is highly stable and repeatable, doesn't cost a lot, and is easy to operate. Its cost makes it easy to service, or use as a discardable sensor. So when am I gonna get one? Comments? --Walt Boyes