A hotel I recently stayed at scheduled elevator stops rather than relying on the traditional elevator heuristic we’ve used since the dawn of technology (Press button, get next elevator going in your direction). That got me thinking. It’s 2014. Why are we still using suboptimal elevator scheduling heuristics? They didn’t do it this way on The Jetsons, did they?

The elevator from The Jetsons
The hotel uses an innovation from the Otis Elevator Company. You enter the floor you want to go to instead of pressing the up or down arrows. There wasn’t much of a learning curve. The floor you entered will almost immediately pop up on one of the screens above one of the available elevators. You then wait for that elevator to arrive and take you to your destination. This minimizes waiting – the potential time savings in a skyscraper are enormous, especially at the tail of the distribution.
I do not know the algorithm used by the elevators. It was certainly a heuristic as opposed to, say, an optimal policy found by dynamic programing, but the heuristic seemed to work very well. I never had to wait longer than 5 seconds for an elevator.
Interestingly enough, some people find efficient elevators annoying. I found a YouTube video about the elevators called “The highly annoying elevators at Chicago’s Swissotel” (see below). Interestingly, the main complaint in the video is that you cannot change your mind about what floor you want to go once you are riding in the elevator (something that has never happened to me ever – I know where I am going!). Another potential complaint is that the first elevator to arrive is not necessarily the one that will take you to your destination. That is an issue with the old school elevators, too (the first elevator may be going up when you are going down). I would have guessed that in general people would like to wait less, but it seems like people would tolerate longer waits rather than embrace change.
Of course, the problem with elevators is the complaints, not the waiting time. If a new elevator paradigm is too disruptive, then maybe it’s not an improvement even if waiting times go down. But I am a fan of this particular efficient elevator system.
Would you like to see more efficient elevators?
January 15th, 2014 at 9:54 am
I found the video’s explanation unconvincing.
First of all, I rarely, if ever, change my mind about which floor I’m going to; so infrequently, in fact, that if I were in one of these elevators when that occurred, I wouldn’t mind the slightly longer time it would take to take a second ride to my correct floor.
The second argument is that it’s somehow unnerving to be in an elevator with no floor buttons. Well, that just strikes me as odd. At least the important ones are there—door open, door close, and alarm… Since I’ve already selected my floor, what’s the problem?
So I guess you can put me down as a vote for an efficient elevator!
January 15th, 2014 at 9:57 am
Michael, I couldn’t agree more!
January 15th, 2014 at 10:18 am
Huh, interesting. I often think about elevator heuristics (usually every time I’m in an elevator, actually). I stayed at a hotel once where I’m pretty sure each of the three elevators stayed at different floors — one was always at ground floor, one was always at floor 10, and one was always at floor 20 in a 30-story building (or something like that, I’m making up the exact numbers). It still used normal floor-selection buttons inside the elevator, but I remember being very surprised at how little time I spent actually waiting for an elevator.
Of course, I was reverse-engineering what the elevators were doing, since I couldn’t see the state of all the elevators at all times, but based on the five or ten times I rode, that was my best guess as to the algorithm.
January 15th, 2014 at 2:49 pm
If you’re riding down from the 99th floor to the first (or vice versa) and you’re suddenly afflicted with claustrophobia, incontinence or something equally urgent, you’ll miss the ability to foreshorten the ride.
I suspect, though, that the main issue is distrust. When an elevator stops at your floor, going the correct direction, and you’re told not to get on it, it seems intuitive that the Democrats /NSA/elevator gods are conspiring against you.
January 15th, 2014 at 3:06 pm
In addition to minimizing waiting time, there must be a secondary objective function here — e.g., energy consumption. I wonder how well the new method does from that point of view.
January 16th, 2014 at 7:15 am
Some years ago I took a course on behavioral decision making, which was mainly on prospect theory. Since then, I believe that not everyone is interested in minimizing the average waiting time, or maximizing the utility function. Maybe “having the chance to change the decision” is more important than the waiting time.
Economics is mature enough that now we have behavioral economics, operations research also should start considering behavioral aspects of humans decisions.
When I red this post, I immediately remembered that joke about an operations research professor who wanted to optimize his wife’s work at the kitchen. He did optimize the procedure, but reduced the joy of life. This elevator system also may decrease the average waiting time, but it may also decrease the trust that people have on this system (as Paul mentioned) or the joy of elevator riding for average people.
January 16th, 2014 at 7:51 am
I heard the following story some time ago, and I am not confident if this story belongs in Fiction or Non-Fiction. But here it is:
A friend worked for a company who was tasked with designing such an algorithm for some hotel, because guests at this particular hotel had been complaining about the long waiting time for elevators. Eventually a better, more efficient system was designed (like the one you describe), but guests were not happy, for the same reasons you listed above.
Basically, they found it unnerving that the inside of the elevator did not have any buttons. (What if I change my mind and want to go to the bar instead of the pool? What if I forget my pool towel in my room & and want to go back for it, do I really have to ride all the way to the pool deck and then ride all the way back to my floor? etc.)
So the hotel took out the efficient elevator system and replaced it with the old elevators. In so doing, they had to redesign their elevator waiting area. They made it all fancy and put up mirrored glass and a big chandelier. The elevators themselves took just as long as they always had.
But no more guests complained.
Why? Because instead of standing around on one foot and then the other, they’d spend the time adjusting their hair, makeup, tie, slacks, and preening themselves in the mirror while they waited.
January 16th, 2014 at 3:05 pm
The Danish writer, Piet Hein wrote a “Grook” entitled “The thinking elevator”
The thinking elevator, so its makers proudly say,
Will optimise its action in an almost human way,
And truly the resemblance,
Is uncomfortably strong,
It isn’t merely thinking,
It’s even thinking wrong
From “motes and beams (Grooks 4)”
January 20th, 2014 at 11:23 am
A further thought – what happens in the lobby? You press the button for an elevator, other guests arrive just as “your” elevator arrives. Human nature means that they go straight to the elevator with the open door, without checking where it is going. (It needn’t be the lobby … but on other floors, people might have learnt the rules.)
It reminds me of a section in a book “The design of everyday things” where the author had a rant about a designer who put hot taps above cold – the almost universal rule is for taps to be side by side, and preferably, hot to the left, cold to the right.
January 20th, 2014 at 4:28 pm
@Kate Owens: you are correct. I’ve heard that story (by Richard Larson) but it contradicted the concept of my blog post, so I left it out (-:
Hari Balasubramanian included that story in his article on queuing that came out today:
http://www.3quarksdaily.com/3quarksdaily/2014/01/some-thoughts-on-the-science-of-queueing.html
March 5th, 2014 at 7:30 am
nice post about elevator and I love this cartoon, every time i watch this cartoon enjoy a lot. Thank for this post