OK, new rule, people: user interfaces have to be designed. In an ideal world they'd be designed by people particularly skilled or trained in user interface design, but I would accept their being designed at all. The current model of just throwing things together until the needs are set is just not working.
My specifics here have nothing that I know of to do with Microsoft, although computers are involved. In this case it's the self-checkout aisle at a Stop And Shop supermarket I don't usually go to (unless it's Shop and Save or some other similar name, and don't think nigh-generic names aren't another problem) so I can't say whether the issue exists with other self-checkout aisles from the same chain. But I had only two items and the real checkout lines were surprisingly busy and I figured why not?
First rule of having people do things as has been understood for only a hundred and twenty years now, people, so you don't have an excuse unless you come from an alternate history where time-and-motion studies were never discovered: don't send people's attention jumping all over the place against the logical progression of the task. For specific example: if the payment due is on the big touch-screen monitor on the left, and the credit card is getting swiped from the little touch-screen monitor in the middle, and the slot labelled 'RECEIPT' is to the right and above both, don't go expecting people to look down at at a surface perpendicular to the one all the rest of these items are on.
Second rule, people: if you have to solve a problem, like having a customer sign for a credit card payment, they're going to expect to sign on, oh, the touch-screen surface they slide their credit card through, the way they do at every other store that has touch-screen credit card pads. If you want them to sign somewhere else you have to direct attention to the spot to sign. Third rule: people form associations based on familiarity, for example, that an electronic pen will be put near the touch-screen surface that's supposed to be signed on. If you don't want people to sign on the credit card swipe screen, don't hang the pen next to the credit card swipe screen. Put it next to the surface they're supposed to sign.
Also, if attention's supposed to leap from one place to a non-adjacent place, use some kind of motion or animation to bring attention. People are very good at spotting motion, except in the context of web page advertisements. If you have the price-display touch screen or the credit card touch screen sweeping out movement attention is going to go to those surfaces and away from the screen they're supposed to sign. And if the customer is not finding the surface they're supposed to sign, instead of repeating endlessly a message to please sign using the electronic pen, indicate where they are supposed to sign. This can be done using either of the screens showing animations of things like swiper the credit card through the credit card swiper. Waiting for the supermarket cashier supervising all the self-checkout aisles to guide the customer to the tiny countertop screen that has otherwise gone completely unnoticed is an inefficient use of resources.
And by the way, the cashier's admission that ``this happens all the time'' is a warning that your customer was not incapable of handling ordinary things because of his advanced degree in mathematics. It means whatever random collection of Forgettol-saturated pedestrians threw components together until it was a big enough pile to install in your supermarket failed to think about the question ``how do we show customers the way to use this efficiently?'' They'll notice, but not learn from, their failure next time they're stuck behind anyone in line.
Trivia: On 9 August 1960 the Department of Defense released NASA from the agreement by which it would develop only passive satellites, allowing NASA to build useful communications satellites that boosted incoming signals. Source: Something New Under the Sun, Helen Gavaghan.
Currently Reading: Asimov's Science Fiction, August 2009. Editor Sheila Williams.