"If we assume that everyone in US cities with >500,000 people makes two elevator round-trips per day, the US is losing out on 270 days of worker days per day or $13 million per year. Each elevator used wastes $160 per year. If your elevator was installed x years ago and your contractor charges you y per hour and takes z hours to invert the buttons, should you inver them? Return ((25-x)160-yz >= 0);"

Sorry but this is absolute garbage. I hate when people take a meaninglessly insignificant value, and then extrapolate it out to come up with a ridiculous dollar value, to try to prove a point.

270 worker days because of elevator buttons? Imagine how much the US loses because people sit on the toilet for 2 mins longer every day because they play Words with Friends. I assure you that lost productivity completely outweighs any time spent figuring out which buttons to press on an elevator.

Another error on your part is that you even said yourself that familiarity with the button layout will improve the time. So basically, if the same person is using the same elevator over and over again, it can't improve much further after a few attempts. It's only the first 10 times that the person actually might take more time, and after that it has reached the optimal level.

Sorry to be blunt, but you should abandon this project. It's not useful at all.

hey steve, first off - thanks for reading the whole way through and in such detail. i completely agree that there are more important matters in the world. the real point of the piece is at the end when i talk about the importance of evaluating all the experiences that you encounter and how you approach the design thereof. the dollar value may not be the most scientific of conclusions, but just trying to give a tangible sense of what the effects are. it's also somewhat light-hearted, in case that wasn't clear.

in terms of further pursuit of the project, didn't have any plans for the buttons in particular except to underscore the importance of user experience design

Wonder how this changes when you take basement levels, strange starting floors, or different popular floors into account? My current building has three basement floors, which are below the ground floor on the button panel, my Uni building's ground floor was actually floor 2 (it was built on a hill!), and the local shopping center has the bottom floors dedicated to parking, so the usual pattern is reversed: 50% of the traffic goes to the top floor to access the shops.

If you start counting microsecond savings, shouldn't you also account for the fact that people looking for a Lobby button in a real elevator (and not on a website) would actually be looking at the bottom of the panel, because that's where a Lobby button typically is. If the button is not there, then a person will have a WTF moment, followed by a mental stumble, followed by a guess that the Lobby is actually on the 2nd floor, followed by time needing to look the panel over and so on. Even if the button is at the bottom, but of a different shape, size or if it has a LOBBY on it, it's still a hassle, because that's not what's expected of it. If you ever been to Vegas with its humongous "Casino Floor" elevator buttons, you should be able to relate to this with ease.

The whole thing is basically an http://en.wikipedia.org/wiki/Spherical_cow example. It is silly.

this is accounted for towards the end of the blog by way of insisting on the importance a real-world use case experiment and determining how people adjust to the configurations over time

Having two separate lobby buttons which take you to the same floor is extraordinarily confusing to me. I had to make sure I wasn't misunderstanding something because it seems like an obviously bad idea. But no, that's the plan:

"Adding a big lobby button at the top will make it faster to identify and hit the lobby button, while persisting the existing lobby button on the bottom."

The clear solution is to make the existing lobby button twice as wide, right? I'd have been very interested to see the results of a layout like that.

That said, elevator buttons are in need of redesign. I have never in my life hit the "hold open" or "close now" buttons because something about the triangles confuses me about which is which, in the short time I have during which they'd take effect.

It sounded good in my head...I guess the results showed that having two lobby buttons was not the right way to go! Glad to hear you support the redesign :D

Well good for you for trying an experiment about it.

You might like the book The Design of Everyday Things which is about issues like confusing elevator buttons.

I just flipped through my copy, actually, because I wanted to quote a passage at you, but I couldn't find it so I guess I was wrong: I had thought one principle of good design was a one-to-one mapping of interface features to functionality. Two buttons doing the same thing would break this concept, but maybe it's not a principle after all.

Hey thanks for the shout out! I was wondering when you were going to post results.

There are some really cool visualizations in the linked opani page: http://opani.com/ryan/elevator-analysis/results/#key=elevato...

I think the most relevant part of this for web folks is the gamification aspect (obviously its what I was interested in in your previous post). I'd love to see a version of this where you show 50% of users "play 20 times to get an easter egg" and give the other 50% no incentive. How much does it actually increase engagement?

haha of course, thanks for the active commenting on the last thread! i think that's a good example, but it seems pretty clear that your hypothesis was correct :D

Well its an obvious insight. But the actual increase in time spent on page is what would be interesting.

We have the session lengths in terms of time as well -- all the data is on git

But you had the gamification stuff in place for everyone right? So there'd be no control group.

i cant respond to your note directly, but true, a further experiment could be done to analyze how much of an effect the gamification had, but the existence of the effect is clear

Elevators should have touchscreen panels allowing an easy change of interface.

Physical buttons with braille are probably a bit easier for blind users. You could have the elevator read out the label of each button on the screen as it was dragged over, or use voice recognition. However, since a decent number of lifts have braille labels, a blind user would expect to be able to run a hand over the buttons, guess roughly where there button should be, and read the braille label - and once they'd found that button, they could then remember its physical position (two buttons from the left, three from the top) for future use.

Physical buttons also require very little power when not in use, and their cost scales relatively well regardless of number of buttons - whereas a touchscreen panel three or four feet high for a very tall building would be pretty pricey. Touchscreens do have the advantage of no moving parts, so they're waterproof and have no internal wear and tear over time (assuming a sufficiently tough protective screen), but monitor illumination dims then breaks fairly quickly, and any protective plastic/glass shield over the touchscreen could get pretty scratched up.

I believe a haptic tablet experiment is in order...