The problem isn't too much choice. If the service were able to say, "Here is an all-around good choice. You could save a maximum of 13% of the price of the tr

The problem isn't too much choice. If the service were able to say, "Here is an all-around good choice. You could save a maximum of 13% of the price of the trip by shoving your schedule around, etc."--summarize the choices and evaluate them the way a competent professional human would do it--then Adams could choose quickly and rest easy knowing he got a fairly good deal. This is a hard thing to make a computer do--maybe it's an easy problem and Orbitz are just idiots, but I doubt it. Perhaps they could do a moderately good job with a moderate amount of work. Point is, the problem is that the computer finds a bunch of choices and doesn't know which one is an all-around good default choice, what differences are important (and thus should be presented to the user) and which ones are irrelevant (and should be summarized or ignored). This can manifest itself as a symptom in various ways. The computer could just spit out all possible choices, which is apparently what it does right now; then the user sees "too many choices". The computer could pick 15 choices (the first 15 cheapest, or alphabetically first, or the ones with indexes equal to integers from a random number generator) and only display them; then users would probably see "a few really bad choices, and I know there are others, but the damn computer pretends they don't exist!" Would users be happier with that? I think leaving the choices there is the most graceful way to fail. For the other things--the watch, the movie theaters--the problem is that they're trying to cater to people with different desires, with the same product. The watch--they could make a watch that tracked running distance and displayed the time and did nothing else, and a separate watch that let you record laps (and displayed the average time!) but had no GPS, plus a third watch with all features and a fourth watch that only displayed the time. This would require some diverse manufacturing equipment (dunno how much of a pain that would be; perhaps not much), 4x the shelf space at stores that sell watches (this would definitely be a pain), 4x as many entries in watch catalogs, and so on. Maybe this would be worth it, if enough people would be willing to "pay 50% more" for it. But, of course, this means a bunch more choices as well, just at a different time. They could also just make one kind of watch and throw all their other customers out the window. This is like the "solution" of throwing out all but 15 arbitrary choices on Orbitz. I doubt if Adams would be pleased if they chose to make only the "it records laps, but no GPS and no tracking running distance" model. Maybe they could run a service: "You tell us what features you want on your watch, and we'll send you a watch that has only those features (and a correspondingly easy interface to them)." Dunno, has anyone tried that? Anyway, the problem is how to provide good products (where simplicity of use is a good feature) that are useful to many people who want different features on the product, when the product's nature (having only a few buttons) makes it hard to provide simplicity along with all the other features. "Too many choices" is the symptom. The problem is something else, and providing too many choices is what someone thought was the best way to punt on that problem. If you ask for fewer choices, either you're asking for some nontrivial work--to solve the underlying problem in a way no one's thought of before--or you're asking for a different way to punt, likely a worse way.