Google wants to build maps that customize themselves based on what they know about you

[protected-iframe id=”d4fde4183d533da3701c21bf6a0367ad-14960843-33105277″ info=”http://new.livestream.com/accounts/74987/events/2497095/videos/34069147/player?autoPlay=false&height=360&mute=false&width=640″ width=”640″ height=”360″ frameborder=”0″ scrolling=”no”]
Transcription details:
Date:
05-Nov-2013
Input sound file:
1003.MP3

Transcription results:
Session Name: Reinventing Maps for a Data-Rich Web

Chris Albrecht
Om Malik
Jonah Jones
Bernhard Seefeld

Chris Albrecht 00:00
Question: Can you design a headset that won’t get my wife to laugh at me every time I wear it? It’s big and it has a big thing and I’m like Bobby Brown from way back in the day. Next up we have Mr. Om Malik is coming back to the stage. He’s going to be talking with Jonah Jones, Lead User Experience Designer at Google and Bernhard Seefeld, Product Management Director at Google Maps at Google, talking about reinventing maps for the data-rich Web. Please welcome our next panel to the stage.

[applause]
Om Malik 00:32
Thanks Chris. Welcome. Wow those are green shoes.
Jonah Jones 00:39
Matching as well.
Om Malik 00:42
I don’t know why I sit down and I start noticing things. You have green soles. What’s with this thing? Is this the subtle Android marketing thing you guys are referring to? Because they changed the color to brown now. You know the Kit Kat is all brown? Just pointing out. Welcome. Just a little back story on these two guys. I met them about seven months ago when Google was redesigning their maps interface, first time in seven years or so. We were supposed to talk for 30 minutes and the conversation went on for about two hours and we talked about everything under the sun – the maps, the interfaces, and I thought that would be a fascinating conversation to have at Roadmaps, so I basically pulled some strings and here they are. Welcome guys. Thank you for making time for us. They normally are very hidden inside of Google. They don’t come out much. They just have to make maps work which is a lot of work I’m told. How long have you been working on maps, the two of you?
Jonah Jones 01:48
Both of us joined at pretty similar times but nearly seven years now. We’ve been working on maps for a while. Berny, actually before Google, was working on maps.
Bernhard Seefeld 01:57
Been working on maps for 10 years now.
Om Malik 01:59
You’re Swiss and you’re British and you’re working on a map solution for an American company. What makes you the ideal guys for that product?
Jonah Jones 02:09
We’re working on the map solution for the world. I think both of us quite like traveling and I think it’s nice to bring different influences into the product to make sure that it works for a global audience.
Om Malik 02:23
Let’s talk about when you started working on maps and where maps are today. Take us through the time which has passed from… I remember the day Google Maps launched and was just like, “Wait, you can do that?” It was a pretty amazing, breakthrough feeling at that time. Since then to today, the world is very different. Tell us from your eyes what you think has changed and why?
Bernhard Seefeld 02:50
When Google Maps launched, it was the first time you could actually drag a map. Before you had to click and it would reload an entire page and it’s very slow so you didn’t really use maps. You read it like a paper map. You opened it at that page and maybe zoomed in. It wasn’t an experience where you would go and browse around and discover things. You could drag the map. That was the first time people started doing this. The real thing that took off was the satellite map. This is something you could go anywhere in the world and see how it actually looks like there. That was a big champion over traffic, and got very popular but it wasn’t the first time there was a satellite map available on the Web even for free. It was really the combination of the experience where you can drag around and fly in and fly out with the data that actually made it a successful product. The biggest change after that was the mobile phone. Again you actually had online maps on the phone for a while but it was the same thing. You had to use arrows. It was very slow. Only with the iPhone then, with the smart phone the touch interface the usage really took off. The also with GPS being available. GPS was in the phones before but the telcos lopped it off and you had to pay extra to use it. Only now that it’s basically a feature that everybody expects and can use, that’s actually when the usage really took off and most people here will use it almost every day to not get lost anymore and that was the next big change.
Om Malik 04:18
The funny thing is, when I used to use the Nokia smartphones, the N series and the E series, they had a wonderful mapping application, you talked about not just a location. They told you where to go. If you were walking, it would tell you what route to take. It wasn’t very accurate at times, but it was pretty amazing to see that. I’ve always been fascinated by that concept. We’re still not fully able to realize the potential of maps. It still feels early days. I don’t know what you guys think.
Jonah Jones 04:52
I feel like definitely digital maps are super early days at the moment, because what we’ve basically done is just taken the same map that you have printed and then put it on a computer screen. You can zoom it and tilt it and that’s quite cool. What we’re just beginning to get started with now with the new Google Maps is understanding that it shouldn’t necessarily be the same map for every single person in every single place. The map to get to the cafe is not quite the same as the map of the park that you should show. Maybe your map shouldn’t be the exact same as my map. I think the main power of having a map that’s digital is that we can generate a new map for every person and every place. We’re beginning to scratch the surface of what’s possible with that now.
Bernhard Seefeld 05:33
What changed since we started over the past few years to enable that is that we have so much data and such higher accuracy that we actually understand so many pieces of the world. Before it was all about the things between the places – the roads and you had addresses and vaguely buildings and now we actually know where the buildings are, how they look like, what’s in those buildings. That gets us a semantic meaning of where you actually stand. Latitude longitude that you get on GPS allows you to sort things by distance, which is nice. It’s useful. But now we can actually tell that you are here in this auditorium or that you are walking with the new sensors to some place or that you just arrived at the train station and probably are going to board a train soon. This context comes from combining the location with the data and what’s in the world, like having that map. And now, we can start using that to change the map. We can actually build a whole new map for every context and every person by combining the things we know about you, things we know about the world, things we know about what people usually do when they’re at this place and create a very specific map that nobody has seen before and you won’t see it again because it’s just there for this moment to visualize the data that we have.
Om Malik 06:47
I think when you talk about that, do we need to think in terms of our quantum or think in terms of the tiles on the map? Do we need to think from that context or even that can be reimagined and rethought of what maps should be?
Jonah Jones 07:04
I think maps is just a canvas for the stuff that we know about the world. It should draw the stuff that you care about. You can imagine if you zoom in to the deepest level of the map, we’re going to show you every single business on the map. We’re going to show you the football stadium and all these different things. As you start zooming out, the labels are going to collide, so you need to decide which of these things are going to be important. Traditionally, what we’d have done is we’d have just said, “Okay, for the entire world, we probably know that this restaurant here is more prominent than this coffee shop, so we let this one win”, and the coffee shop doesn’t show. We already had to solve that problem but now it gets more interesting because we know, maybe the places that you’ve rated and reviewed, maybe the places that your friends have been to and they like, maybe we know that there’s a football match happening at the football stadium right now and the score is three-two or something like. We can actually use those signals to draw a map that’s contextually much more relevant to you and choose the right information to show. And similarly, we can say that, whereas on a normal map you might show the most important arterial roads and highways and stuff, but maybe now you click on this tiny little museum down a backstreet, and now we know you’re more interested in that. Now we’re going to make sure we highlight and label that little footpath that leads up there and all these minor streets. The fact that we can re-rank and decide what to show on the map depending on your context and you and a bunch of other signals, is super interesting.
Om Malik 08:35
It seems like you guys are saying the data inside the maps is going to become more and more important than just the idea of the map itself. Is that a fair statement?
Bernhard Seefeld 08:47
I would say the map is one interpretation of the data. It’s the one that’s very visual and you can see that and that’s the one that drives us, and we pick which things are important and so on. But once you know that, it actually means we understand the context here, and that context we can carry on to any other application. We can use it in Google Now. We can use it in many other applications that don’t require a map because you can do this interaction just by then. It’s no longer necessary for you to open a map and understand it. You can do it for longer tasks. You can do a very quick task in a few seconds just based on the context. It’s almost like your location is the query, like you’re at a very specific location, if you’re at the airport ticket counter, you’re probably going to check in, so that’s almost like a query. We could do something as well there.
Jonah Jones 09:33
Or your context, you’re hashing on the time.
Bernhard Seefeld 09:37
That actually happens because we go even further with the accuracy. We know the buildings, we know the businesses, but the next level is really mapping things indoors and knowing what is the semantic meaning of being at the cashier in the store; these kinds of things. It’s just the next level of accuracy that can build into a map that is exponentially richer in semantic meaning that we can help to create a meaningful experiences based on your location.
Om Malik 10:02
Recently, we had the whole chaos with the Apple maps and they would always have wrong data. It seems like you guys are talking about data influencing how the maps are going to work in the future. The fidelity of that seems like that is the gating factor in the success of not just map but any other application out there.
Bernhard Seefeld 10:25
I can tell you, it’s very hard to build a very accurate map. We’ve been working on this for many, many years and everybody thinks it’s easy and hasn’t got the depths you have to go. It’s a lot of work and the biggest problem, our biggest competitor, is actually the real world. If you think about it, it changes so much and keeping up with that is a huge challenge. In the US, it’s about 20% of businesses every year, either close or open, so they’re different. If you have a perfect data set tomorrow, within a year, one in five things will be completely wrong. All the little details change. You have to keep up. You have to build a system that keeps up with that data. That’s the core base challenge. You only see those things being wrong once you visualize them. Only then do you notice, “Oh, this road thing here, we have your accurate list”, but it doesn’t line up with this other piece. So if you build an application that is data-rich, you need to start with at least one of the use interfaces that will guide you, which areas are more important, which are less important? You won’t get 100% perfect at the beginning but that will be the guidance to get a step by step towards this data-rich experience, and then you can add the context. Then you can go and say, “Now I understand the basic layout of the road. Now I can understand your data and then I can understand intersection between those two and that allows me to create a much richer experience.” That’s really where we are right now. We’re at the beginning of this phase. We’ve spent the last seven years or more building up the base state of figuring out how to keep it up to date and now we can, on top of that, build the next level of accurate applications.
Jonah Jones 12:10
I think what’s really interesting about this is it can make the map feel a lot more alive. If there’s the football game going on right now, maybe you can zoom in and you can click on that, and maybe we can tell you what the score is, and that explains why the traffic is so bad going there. We’re starting to do things now, where we can tell if you’ve got a restaurant reservation, so when you click on the restaurant, it will say, “You’ve got that booking for 8:00”, or things like that. It’s got my flights on the map now, as well.
Bernhard Seefeld 12:36
That was fun. The first time I saw the restaurant reservation I had made, without even thinking of maps, it just went to open table, made the reservation, and then opened maps the next time, and it was labeled right there, “Table for four, 8 pm”, and that’s really cool. Obviously, it’s adapted to me and that was the place to go but that was a real-time affirmation of here’s this bit of data that came in and it’s now reflected on my map right here.
Jonah Jones 13:01
That’s just because it was in our Gmail inbox, so you have all the information there.
Om Malik 13:04
Great. When you guys look at the maps, the way you’ve been working on it, the way you guys are talking about data, influencing the future, how do you think people should be thinking about building their products in the future? How is data influencing applications and services in the future? How should we be thinking about that?
Bernhard Seefeld 13:29
The key point really to start with one user interface that drives what kinds of data you acquire, how accurate you need to make which part of the data. Well, you notice, you hit the barrier and you can’t get it more accurate in this dimension, which is inherently fuzzy, then you adapt the user interface to not make promises you can’t hold up. That dialogue, that’s the first step that you have to build a process around. It’s also where you get the feedback from users. Once it’s launched, they will tell you what’s wrong and you can correct it. Some things, people are really good at helping you. Other things, they’re not that good at helping you. You will figure that out. Things, if you’re very strongly to fix your business. Only if you’re very strongly about fixing your business, and you have to create a good process around that. In our case, we’ll find similar things. That’s really the foundation, and you can’t really skip over that. Once you have that foundation, then you can go and watch the users and create extra data sets of information you know about the user or the context of the user. What does everybody else do in that similar situation? That’s another data set. That’s when you start merging, and you get interesting applications.
Om Malik 14:41
The interesting about maps is that it uses just once sensor in just the GPS location as a context. Now our phones have much more essential data coming from them. You can tell when you’re walking or driving or riding a bike. Your phone can tell that. And yet, every time I use Google Maps or whatever – I have to get walking directions or directions for driving. Why? Why do I need to do all those things? It seems like we’re still stuck on using only one sensor in our information slicing and dicing.
Jonah Jones 15:26
This is a super interesting topic, actually. We were talking about this just yesterday in the office. Where do we want maps to be two years out or five years out? You can’t think about, where do we want it to be today? With the current technology, there’s so much more you can do. There’s just a finite amount of time that you have to build these things. But I image with all the sensor data we’ve got, we can do some super interesting stuff. Of course, you’ve got all the battery constraints, which is what normally holds you back with the mobile phones and stuff.
Om Malik 15:55
You’re Google, you can do whatever.

[chuckles]
Om Malik 15:57
Come on. You don’t have to worry about the battery power. It’s the other people who have to deal with that.

[chuckles]
Jonah Jones 16:03
But the signal’s on your mobile phone, you don’t want to be changing [crosstalk].
Om Malik 16:07
But seriously, do you think in the next two, three years, we will actually start to see maps which are more sensor intelligent and applications? What do we need to think about for a world like that from a creation and consumption standpoint?
Bernhard Seefeld 16:25
The location is the one piece of the context that we’re talking about. So location in context of a map intersecting that we know where you are, what the place is. The sensors, they give you extra context. So that will build things up together. This drives a lot of applications. There will be improvements in maps, but there will be even more interesting things coming out of this where we really have a much deeper understanding of what you’re currently up to, whether you just arrive or you’re passing by or where you there for a half an hour? Makes a big difference. So we where to place you at, but that give us context. So I really believe this will be a big change. The latest versions of Android and iOS have support to make this easy for applications to use anyway.
Jonah Jones 17:09
It’s way bigger than just maps. Maybe you’ve got a slightly-elevated heart rate now because you’re on stage [chuckles] and you’re sitting here and stuff. Maybe you’re–
Om Malik 17:19
Are you talking about yourself or me?

[chuckles]
Om Malik 17:21
Because I’m perfectly fine.
Jonah Jones 17:23
This isn’t the time for your mobile phone to ring. Whereas, when you’re off stage and relaxing, that’s a good time to your latest e-mails and stuff. So the more you know about somebody and the more sensors you’ve got, the more interesting stuff you can do.
Om Malik 17:35
I find this very frustrating at times. The phones – the hardware – is so much more ahead of what the experience should be. We still have not figured out the minor details of how to create those applications. So maybe you can tell us what the Google Maps of two years from now looks like.

[chuckles]
Jonah Jones 17:59
I don’t think we can really talk about that. I can talk about all the stuff that I’d like it to be today. I can image that when you zoom into the deepest zoom level of the map, sure, we show you all the businesses and this stuff. But Google is smart enough to know a lot more about the world than that. It knows where I’ve been, where I hang out, maybe it knows where my friends have invited me to a party. It knows a ton more about the world. As I was saying before, it knows the score of that. That football game is going on right now. So it’d just be really nice if we were just smarter about putting everything that we know about the world and plotting that into a place so that you can use that as a way of discovering stuff.
Om Malik 18:35
Dennis Crowley calls it his Hattie Porter Map which tells you everything. When you talk about Google and spending all that time and effort, how many people work for Google? Can you actually even count?
Bernhard Seefeld 18:52
It’s hard to count because it’s not like– it’s a [inaudible] on the maps team. They’re the people who build the cameras that go on streetcars. Do you want to count them? And then, it goes into the where we maintain the business database that’s also used all over Google. So it’s not really a fixed length, but I would say, one and a hundred.

[chuckles]
Bernhard Seefeld 19:16
Maybe a couple thousand.
Jonah Jones 19:16
Less than a million.
Om Malik 19:17
One thousand?

[chuckles]
Om Malik 19:19
You sure about that?
Bernhard Seefeld 19:20
Yeah.
Om Malik 19:20
Do you want to take a help line? Call somebody?

[chuckles]
Om Malik 19:24
I’ve got more phones with me.
Jonah Jones 19:25
We have all the mapmaker users, of course, as well – the people that are contributing to the map. So you can count them too.
Om Malik 19:31
Do you think we will ever actually have maps which will be more like the maps, the way they used to be in the old atlases? The nuances on those were just so much. Just the coloring used to tell so much about the map and the place itself.
Bernhard Seefeld 19:50
That’s actually a fair topic. If you look at the map from the 15th century, 16th century, they were telling a story of the guy who drew the map or the group of people. We heard there’s some lake over there, so you’ve got a rough lake. But they actually go there. There’s a coastline that has an insane detail of rocks, probably because they had to navigate around them. The things people pick to put in the map and choose more details and less details, and here, we ran into a bunch of natives – all these things. That’s their choice and it tells a story. It’s very emotional. What happened over time is we got better at making maps. Every iteration of maps became more neutral. It became more like a dictionary, certain rules on how you pick things. Now we have this common denominator map, a really useful roadmap that reflects the importance of the road for everybody. Now, we are at the point where we can reverse that, and we can move it back into a map that we have an emotional connection. A lot of you guys looked at the map at some point in your life. You can look at it for hours and find new things, and the places you see that remind you of things. It’s a very emotional connection you can have, and that’s on a very generic map. So by building this map that is really adapting to you, where we see where you’ve been, where we know what kinds of places you might go to, that actually moves us back to an emotional map – a map we have an emotional connection to that reflects a real-life connection, and that says, we left those places, a peek in the future, we actually travelled there, and these kinds of things. So that is something that’s really exciting. I think you will see a bit of that in the next two year, but it’s going to be a long process.
Jonah Jones 21:34
There’s an interesting aesthetics part of this as well, which is that you need to balance this application being performant on your mobile phone. You want it to be as beautiful and as stunning as possible. We find out, Man, we’ve got this amazing, new 3-D mesh model for this 3-D building. The engineers are like, No, [chuckles] because they’ve got to figure out how to make this thing work with a good frame. So we’re constantly trying to design the most aesthetically pleasing thing that we can, but within the constraints of the hardware. One example of something that we did was, recently, with the Google Maps, we have a gradient on the outside of the water. So it gets darker as it’s approaching the landmass. It gives the maps this depth and this sense of richness that you used to get in the old maps, but it’s also super useful because it adds to the contrast of being able to see the land forms. But I was not popular with the engineers when I first did the designs [chuckles] for that one. So you’ve got to balance those things.
Om Malik 22:29
But now, we have quad core phones essentially, right? So it’s not like the processing power is an issue or–
Jonah Jones 22:35
It’s still an issue.
Om Malik 22:36
Really?
Jonah Jones 22:36
Yeah. Look at all the data we’ve got in the world. If you wanted to do everything fully textured with beautiful light source and shading, reflections, and all this crazy stuff, your phone would fill over pretty quickly. So we’ve got to– [chuckles]
Om Malik 22:49
–to invent something. It’s Google.

[chuckles]
Om Malik 22:51
You know everything. Jokes aside, the thing which I find interesting about maps and smartphones is, before the smartphones came around, I used to use a regular candy bar phone. And I would remember everybody’s phone number – all my friends. I would remember almost 250 people who were important to me – their phone numbers. I would look at the numbers and– who’s calling. When smartphones came, they just can do that. I forgot that I had that ability to retain phone numbers. Now with the maps, I take a lot of pride in being– I can look at a map, and I can remember the map for the rest of my life. I can navigate wherever I’m going. Except for Paris, I get lost there all the time. I don’t know why. But any city I look at, I can remember it. But lately, because of the phone and the map, I’ve started to lose that ability. So it’s an interesting dichotomy, that we’re losing our human capability to machines in many ways, including these maps.
Jonah Jones 24:00
I want to talk about the napkin map for a second on that. Hold that thought. One of the interesting things that we wanted to do whenever you click on a place, we wanted to draw the equivalent of a napkin map that I would draw you on the back of a napkin – super simplified. This is the street. This is the street. These are the three that you need to care about and not all these other ones elsewhere. Part of that concept was that you should be able to glance at the map for one second, then I take it away, and you just remember the important list. It turns out, that concept is super simpler and actually doing it is a little bit harder in terms of the algorithm. Also, you don’t want to necessarily just throw away all the other information. But we’re trying to do that a little, make it easier for you to remember the personal pieces of information rather than bombarding you with everything. What were you going to say?
Bernhard Seefeld 24:44
Exactly. If you glance at the map we have today, like the old Google Maps, it was just a regular map with a red pin. You can read it and you can start to memorize it, but it’s a lot of effort that you’re spend to memorize it. The concept of this napkin map, the basic things you need, you can read much, much faster. You will get the right roadmaps. You get the right structure. There’s now more bandwidth for you to look at. If you decide at that point, that’s enough, you can just memorize that. But now, we know where you’re going. Let’s say, you go to [inaudible] conference. And then, we draw you the map on how to get there. But we also should know what kinds of things you will be doing before and after. That way we can show you places on the map that you haven’t discovered yet that might be useful. Now we have time to look at those and to take those in. You also know that these places are more relevant. They’re specific to your task. They weren’t just regular tourist things that you’ve been doing the last five times you went to Paris. It’s these new things. So we keep thinking about, if we only get people to experience 10% more of the world than they would usually do, just because the way they read the map points out things that they didn’t see before. Because it’s now more contextual, it will be more interesting to them. Then, they choose not to go to the same cafe all the time. Let’s try this other one. Or they go to a national park that they should go to that they haven’t visited yet. Or they go and experience this attraction with their kids that they didn’t think about – that even exists – like some kind of trampoline place they didn’t think about. How amazing is that they see just a few more things, just 10% more things in their life. The trickle-down effect of that is amazing. You get all these small businesses, suddenly new businesses. People get aware of social issues that they weren’t aware of. They see beauties in the world that they want to protect after seeing it. The after effects of just this small change, they’re massive.
Om Malik 26:42
I’m going to miss the soulful aspect of getting lost and finding my way. Unfortunately, we have to find our way back because we’re out of time. I didn’t even get to open the Q&A to the audience. So sorry, guys. But thank you for making time for us. Hopefully, we shall have you back and the next year’s maps will be more magical.
Bernhard Seefeld 27:05
Thank you.
Jonah Jones 27:05
Thank you.

[applause]

[music]
Chris Albrecht 27:13
Thank you. You should break out your map app right now and plug in 46 mina, because that is where the after party is this evening at 7:15 – just a reminder. And then, plug in Om’s address because that’s where the after, after party is. He doesn’t know it yet, but if you just show up, he’ll be totally cool about it. Up next, we have Brady Forrest of Highway1, and he’s going to talk about, From Design to Doorstep: Bridging the World of Ideas