is a blog about design, technology and culture written by Khoi Vinh, and has been more or less continuously published since December 2000 in New York City. Khoi is currently Principal Designer at Adobe. Previously, Khoi was co-founder and CEO of Mixel (acquired in 2013), Design Director of The New York Times Online, and co-founder of the design studio Behavior, LLC. He is the author of “How They Got There: Interviews with Digital Designers About Their Careers”and “Ordering Disorder: Grid Principles for Web Design,” and was named one of Fast Company’s “fifty most influential designers in America.” Khoi lives in Crown Heights, Brooklyn with his wife and three children.
Among many modern innovations that we have at the new Times building , we’re using a fancy method of managing elevator ridership. Rather than having riders push a directional button (up or down) take the first elevator car that arrives — which is what practically every other elevator bank in the world does — ours takes a different, more ambitious tack.
At our building, a rider pushes a button on a keypad before getting on an elevator to tell the system what floor she’d like to go to. The system then directs her to a specific car which, in theory, will also carry other riders going to that same floor. The idea is to get riders to their floors faster by ‘batch processing’ them, so to speak, rather than serially processing them.
We Had to Break the Elevator in Order to Improve It
This is a potentially very useful rethinking of the traditional elevator usage model. In the old Times building, for instance, elevators would often makes stops on as many as ten of the fourteen floors, and you may work in an office building yourself where the elevator experience is similar. In terms of lost productivity alone, it’s a worthy goal to try defeat this inefficiency.
There are some usability problems, though. The interface for directing riders to their cars flashes its information too quickly, especially given that the idea of waiting to be told by a computer which elevator car to go to is an unfamiliar activity for most riders. As well, there are no buttons inside of the elevator cars (which also means you can’t change your mind once you’re on), only a digital read-out that indicates what floors the elevator will stop at — but with no indication of what floor the car might currently have its doors open to. As a result, lots of riders get off at the stop before their intended destinations, which isn’t fun.
Elevator Pitchin’
Judging by most of the conversations that happen in and around the elevators, this new system is not exactly a popular hit. I must admit, I’m currently not a huge fan of it, but I tend to think that most complaints are owing to the not insurmountable usability problems I described above. Hopefully, those will get sorted in short order so that we can see if this batch processing method of elevator travel does indeed work.
However, it would have saved a lot of people a lot of trouble had there been a skilled interaction designer on the design team for this system from day one, which there clearly was not. Which is my point, really: it’s not advisable to try to reinvent universally understood user experiences like elevators without excellent, excellent design help.
Where I interned last summer, they had a similar system, and it was very hit or miss. Sometimes, the elevator would stop at 5 or 6 floors before getting to yours and it would be a huge annoyance, but other times you would zip right up. It was always entertaining seeing people using it for the first time, however–they’d stare for a while until they would get the nerve to ask someone how to get to the floor they wanted.
Joel on Software also pointed out another usability bug — riders who are unfamiliar with the building’s elevator system might rush to an opening elevator thinking they’ll be able to select a floor within only to be taken for an unexpected ride.
I think its a great idea to try to re-think elevator efficiency, but until this type of system becomes the norm, they should have some fall-back controls in the cars themselves. You could even do something like punish users who step into cars without pre-selecting by making the car go to their stop last regardless.
My office building in Hong Kong has the same. It’s fun to watch visitors puzzle out the system for the first time (I don’t speak enough Cantonese to help most people. I do a pretty mean charades routine, though)
I think for us the idea comes down to transparency. Do we trust the computer to make a smart decision and get us to our floor quickly? This is hard to do, especially when it’s not telling us the other facts it uses to make its calculations.
Our system has a nice perk: the front desk / security team can punch in your floor destination for you, from the comfort of their chairs. They can recognize you from the street, punch in your floor, and have the fully-automated elevator ready when you walk in – and you can whiz up to your twenty-third floor office without touching a single button.
Yeah these elevators have been out for quite a while now, just never really seen in the wild. I think it is a good improvement over the traditional elevators, however does pose many usability issues like you mentioned.
Would it be handy to have a ‘traditional’ elevator next to these as well?
Living in a town where buildings infrequently break the 3 story mark I can only marvel at your fancy vertical world of wonders. That said, imagine that I wouldn’t want anything to change from this day forward. The system is rather odd and can be a challenge. I think I should revel in that.
A new elite clique of elevator ninjas should arise. Capable of knowing just where and when to be to catch the series of elevators that will move them most swiftly from floor to floor. They might lead a bewildered colleague through a series of elevator rides bearing no resemblance to a logical system that is in fact a secret code to the wonderful building they call home for a good portion of their day.
I think working there each day would be the most fun thing in the world. And I would never want the system to be improved one iota. Even though I may never gain this bellhop knowledge of the elevator arcane necessary to reach my desired floor I think it would be nice to just bump into folk as I aimlessly wander up and down.
Jay: We have express elevators already, as the Times has twenty-plus floors in the building. The problem is that even within the grouped floors (e.g., one thru fourteen), there’s a tremendous amount of traffic frome one to another, because of the nature of the business.
Each of the problems described above (except for the one where the elevator’s internal display doesn’t show the current floor number, that seems like a terrible ommission!) seem to only apply to novice users of the system. Once you’ve learned how to use the system, it’s probably super efficient not just for the individual, but for the overall efficiency of the companies working inside of the building. Don’t you think so, personally?
It’s like the NY Subway MetroCard machines — they suck for first-time users, but for everyone else we should be glad that the UI doesn’t put too much emphasis on guiding newbies — since 99.9% of every session with the machine is probably by people who already know how to use it intimately. This is all in line with Alan Cooper’s rule of thumb to design the best core user experience for the intermediate user, not the newbie.
Back to the elevator example: Having “override” buttons inside of the elevator completely defeats the whole purpose of the new design. It’s kind of an absurd suggestion, really: It reverts the whole system to the old way, allowing cars to arbitrarily go to floors inefficiently. Maybe a skilled UI designer was involved in this decision?
Also, compare these complaints to the problems already existant in traditional elevator design — in particular the long, long waits — and I can’t help but think that the new design is far superior.
Since the building is brand new (and not yet fully occupied), it’s impossible for you to accurately compare the real-life performance of old vs. new style elevators (given the same footprint and same number of people working in the building), but I would imagine you’d be singing a different tune if the system was reverted and you suddenly had to wait 5 minutes for an elevator to arrive to pick you up.
One thing i LOVE about the Times’ elevators is that the digital display indicating if the car is going up or down is *inside* the elevator — this means that instead of having (say) 400 of these displays (ten per floor) there are only 10 of them, making it a lot easier to maintain them and of course making it financially feasable to make them very fancy.
It sounds good in theory, so if they work out those kinks it should be rather successful. My only other issue that I am imaginaing while thinking about this system is that depending on how far away the elevator carts are from one another that could also be quite the hasle.
Also knowing human nature we all need a last minute chance of changing our minds… Especially women…
Chris: Actually, the elevators right now sometimes do in fact take 5 minutes or more to arrive. I’ve waited for as many as ten.
That photograph you linked to was taken by a friend of mine; I’ve never seen an error quite as bad as that but I have seen other errors, and I’ve seen some wonky behavior from the system as a whole.
Also, the digital readouts being only inside the elevators are problematic, as they can’t be easily seen from any vantage point the way they could be on the outside.
These are among some of the other usability problems that I omitted here in the interest of staying optimistic.
I think though that I’ve reached intermediate status, and I’m not as much of a cheerleader about its usefulness as it sounds like you are. I don’t mean to offend, but your take is kind of the novice take on the system.
For instance, I still think that not providing override buttons inside the elevator is silly. People change their minds, it’s a fact. It’s also a fact that the vast majority of people know where they’re going to go, and they’re not going to change their minds that often. Nor will the social climate allow people to regularly jump on an elevator and hijack its express path by pushing on an un-scheduled floor.
It’s a system full of interesting ideas. And while I think the problems can be sortred out, I just wouldn’t advise anyone to be a cheerleader for it until they use it day in and day out.
How efficient does every moment of out lives have to be? It’s bad enough we have to go to work. Can’t we just enjoy a little randomness in our elevator ride?
J.Ro
@peterme:
I personally had the displeasure of staying a week at the Times Square Marriott Marquis and experiencing their “Smart” elevators firsthand back in April of this year. The system, it seems, is not quite ready for primetime, or at least, not ready for a major conference such as the one I attended at the hotel that week.
It was not uncommon for one to wait, as Khoi experienced with his building, upwards of ten minutes for one’s car. More perplexing and frustrating than that, however, was when the system would lock up as it became overloaded with requests, and keep users from punching in their destination. When locked up, we the attendees saw it take 15 minutes at one point to come to its senses, despite the shafts having been broken into groups of floors. This inconvenienced we attendees as well as the poor hotel guests who had to put up with this on top of the rest of our hubbub that week.
As Khoi summed up with his last post, “I just wouldn’t advise anyone to be a cheerleader for it until they use it day in and day out.” Word.
We have a similar system here at the Ministry of Education building in Singapore. Despite the large number of signs telling people to input their destination, we still get a lot of newcomers hopping on to open elevators then discovering that they can’t choose their destination from within.
The one improvement we have here is that when the elevator stops at a floor, a computerised voice reads out the floor number, along with the large digital display on the top of the door.
It’s hard to know when to close to the door, or if “everyone” is in.
I worked for an elevator control manufacturing company for years writing dispatching software. Destination based dispatching (which is what this blog is describing) was something on my implement schedule when I left 2 years ago.
I went through the entire requirement and design process and usability was definitely one of my top concerns. From my research and my experience in elevator dispatching, such a system is only optimal when the majority of riders arrive in the lobby of a building, typically first thing in the morning and after lunch. In all other traffic patterns normal elevator dispatching, where you just push an up or down button, works a lot better.
In some cases, the destination based interface is only placed in lobbies and everywhere else in a building regular dispatching is used. This too can be a little confusing and has its own usability issues but from a system performance perspective of minimizing peoples wait times (which is the ultimate goal) it works much better.
This thread is very interesting. I have always wanted to hear from the people actually using such systems. This is a big fad among elevator consultants, I constantly saw specifications calling for this system as an option that could be added or removed just in case it didn’t work. It will be interesting to see if its popularity grows. The concept and a couple running systems have been around for over 7 years.
Among many modern innovations that we have at the 
Where I interned last summer, they had a similar system, and it was very hit or miss. Sometimes, the elevator would stop at 5 or 6 floors before getting to yours and it would be a huge annoyance, but other times you would zip right up. It was always entertaining seeing people using it for the first time, however–they’d stare for a while until they would get the nerve to ask someone how to get to the floor they wanted.
Joel on Software also pointed out another usability bug — riders who are unfamiliar with the building’s elevator system might rush to an opening elevator thinking they’ll be able to select a floor within only to be taken for an unexpected ride.
I think its a great idea to try to re-think elevator efficiency, but until this type of system becomes the norm, they should have some fall-back controls in the cars themselves. You could even do something like punish users who step into cars without pre-selecting by making the car go to their stop last regardless.
My office building in Hong Kong has the same. It’s fun to watch visitors puzzle out the system for the first time (I don’t speak enough Cantonese to help most people. I do a pretty mean charades routine, though)
I think for us the idea comes down to transparency. Do we trust the computer to make a smart decision and get us to our floor quickly? This is hard to do, especially when it’s not telling us the other facts it uses to make its calculations.
Our system has a nice perk: the front desk / security team can punch in your floor destination for you, from the comfort of their chairs. They can recognize you from the street, punch in your floor, and have the fully-automated elevator ready when you walk in – and you can whiz up to your twenty-third floor office without touching a single button.
Yeah these elevators have been out for quite a while now, just never really seen in the wild. I think it is a good improvement over the traditional elevators, however does pose many usability issues like you mentioned.
Would it be handy to have a ‘traditional’ elevator next to these as well?
I had a similar experience a few weeks ago, here in Brussels…. very disturbing…
As you state: “it’s not advisable to try to reinvent universally understood user experiences like elevators without excellent, excellent design help.”
Living in a town where buildings infrequently break the 3 story mark I can only marvel at your fancy vertical world of wonders. That said, imagine that I wouldn’t want anything to change from this day forward. The system is rather odd and can be a challenge. I think I should revel in that.
A new elite clique of elevator ninjas should arise. Capable of knowing just where and when to be to catch the series of elevators that will move them most swiftly from floor to floor. They might lead a bewildered colleague through a series of elevator rides bearing no resemblance to a logical system that is in fact a secret code to the wonderful building they call home for a good portion of their day.
I think working there each day would be the most fun thing in the world. And I would never want the system to be improved one iota. Even though I may never gain this bellhop knowledge of the elevator arcane necessary to reach my desired floor I think it would be nice to just bump into folk as I aimlessly wander up and down.
Express elevators are the way to go. I’ve seen a few throughout the city. One bank goes from floors 1-8, another goes 8 – 16, etc, etc.
Seems to solve the problem of hitting 10 floors before you hit yours, and people understand signs that say Floors 1 – 8 only.
Jay: We have express elevators already, as the Times has twenty-plus floors in the building. The problem is that even within the grouped floors (e.g., one thru fourteen), there’s a tremendous amount of traffic frome one to another, because of the nature of the business.
Each of the problems described above (except for the one where the elevator’s internal display doesn’t show the current floor number, that seems like a terrible ommission!) seem to only apply to novice users of the system. Once you’ve learned how to use the system, it’s probably super efficient not just for the individual, but for the overall efficiency of the companies working inside of the building. Don’t you think so, personally?
It’s like the NY Subway MetroCard machines — they suck for first-time users, but for everyone else we should be glad that the UI doesn’t put too much emphasis on guiding newbies — since 99.9% of every session with the machine is probably by people who already know how to use it intimately. This is all in line with Alan Cooper’s rule of thumb to design the best core user experience for the intermediate user, not the newbie.
Back to the elevator example: Having “override” buttons inside of the elevator completely defeats the whole purpose of the new design. It’s kind of an absurd suggestion, really: It reverts the whole system to the old way, allowing cars to arbitrarily go to floors inefficiently. Maybe a skilled UI designer was involved in this decision?
Also, compare these complaints to the problems already existant in traditional elevator design — in particular the long, long waits — and I can’t help but think that the new design is far superior.
Since the building is brand new (and not yet fully occupied), it’s impossible for you to accurately compare the real-life performance of old vs. new style elevators (given the same footprint and same number of people working in the building), but I would imagine you’d be singing a different tune if the system was reverted and you suddenly had to wait 5 minutes for an elevator to arrive to pick you up.
One thing i LOVE about the Times’ elevators is that the digital display indicating if the car is going up or down is *inside* the elevator — this means that instead of having (say) 400 of these displays (ten per floor) there are only 10 of them, making it a lot easier to maintain them and of course making it financially feasable to make them very fancy.
Of course, there’s always the problem of the crashed elevator!
It sounds good in theory, so if they work out those kinks it should be rather successful. My only other issue that I am imaginaing while thinking about this system is that depending on how far away the elevator carts are from one another that could also be quite the hasle.
Also knowing human nature we all need a last minute chance of changing our minds… Especially women…
Chris: Actually, the elevators right now sometimes do in fact take 5 minutes or more to arrive. I’ve waited for as many as ten.
That photograph you linked to was taken by a friend of mine; I’ve never seen an error quite as bad as that but I have seen other errors, and I’ve seen some wonky behavior from the system as a whole.
Also, the digital readouts being only inside the elevators are problematic, as they can’t be easily seen from any vantage point the way they could be on the outside.
These are among some of the other usability problems that I omitted here in the interest of staying optimistic.
I think though that I’ve reached intermediate status, and I’m not as much of a cheerleader about its usefulness as it sounds like you are. I don’t mean to offend, but your take is kind of the novice take on the system.
For instance, I still think that not providing override buttons inside the elevator is silly. People change their minds, it’s a fact. It’s also a fact that the vast majority of people know where they’re going to go, and they’re not going to change their minds that often. Nor will the social climate allow people to regularly jump on an elevator and hijack its express path by pushing on an un-scheduled floor.
It’s a system full of interesting ideas. And while I think the problems can be sortred out, I just wouldn’t advise anyone to be a cheerleader for it until they use it day in and day out.
Your very own editorial featured a nifty video spot on the smart elevators at the Marriott Marquis in Times Square.
I am just a starry-eyed newbie, of course.
How efficient does every moment of out lives have to be? It’s bad enough we have to go to work. Can’t we just enjoy a little randomness in our elevator ride?
@peterme:
I personally had the displeasure of staying a week at the Times Square Marriott Marquis and experiencing their “Smart” elevators firsthand back in April of this year. The system, it seems, is not quite ready for primetime, or at least, not ready for a major conference such as the one I attended at the hotel that week.
It was not uncommon for one to wait, as Khoi experienced with his building, upwards of ten minutes for one’s car. More perplexing and frustrating than that, however, was when the system would lock up as it became overloaded with requests, and keep users from punching in their destination. When locked up, we the attendees saw it take 15 minutes at one point to come to its senses, despite the shafts having been broken into groups of floors. This inconvenienced we attendees as well as the poor hotel guests who had to put up with this on top of the rest of our hubbub that week.
As Khoi summed up with his last post, “I just wouldn’t advise anyone to be a cheerleader for it until they use it day in and day out.” Word.
I don’t know… how are you supposed to hit on the nice lady that works two floors below yours now?
We have a similar system here at the Ministry of Education building in Singapore. Despite the large number of signs telling people to input their destination, we still get a lot of newcomers hopping on to open elevators then discovering that they can’t choose their destination from within.
The one improvement we have here is that when the elevator stops at a floor, a computerised voice reads out the floor number, along with the large digital display on the top of the door.
It’s hard to know when to close to the door, or if “everyone” is in.
Word!
Hello! Good Site! Thanks you! rjjegaxsans
I worked for an elevator control manufacturing company for years writing dispatching software. Destination based dispatching (which is what this blog is describing) was something on my implement schedule when I left 2 years ago.
I went through the entire requirement and design process and usability was definitely one of my top concerns. From my research and my experience in elevator dispatching, such a system is only optimal when the majority of riders arrive in the lobby of a building, typically first thing in the morning and after lunch. In all other traffic patterns normal elevator dispatching, where you just push an up or down button, works a lot better.
In some cases, the destination based interface is only placed in lobbies and everywhere else in a building regular dispatching is used. This too can be a little confusing and has its own usability issues but from a system performance perspective of minimizing peoples wait times (which is the ultimate goal) it works much better.
This thread is very interesting. I have always wanted to hear from the people actually using such systems. This is a big fad among elevator consultants, I constantly saw specifications calling for this system as an option that could be added or removed just in case it didn’t work. It will be interesting to see if its popularity grows. The concept and a couple running systems have been around for over 7 years.