Episode #215 Stephen Anderson - Displaying Data in Digestible Ways
Culling through massive amounts of data is a headache. A dense table of aggregated data points can be useful in theory, but the manner in which it’s displayed is often a hindrance. Even more than that, showing that data in a chart or graph is confusing if it’s not effectively labeled. Data is useless when you can’t make good decisions from it.
Stephen Anderson is a leading thinker on design and psychology. He spends lots of time thinking about how to visualize information and uses the travel site Hipmunk as an example of a good data visualization. All the flights are listed with clear visuals; length of the flights, layovers, and times. They even let you sort the results by “agony”. But ultimately, from a cognitive standpoint, displaying the flight data in this way creates less of a burden on your short term memory and makes for an easy comparison.
Many times it comes down to taking a step back and really thinking about how data is displayed. It’s very easy to fall into placing something into a grid, table, or other familiar pattern because of what’s been done in the past. By really analyzing what kind of data you have and what you’re trying to communicate, you can uncover better ways to present that data.
Jared Spool: Hello once again, everyone, and welcome to another episode of the SpoolCast. We have for you today a very special dude, Mr. Stephen Anderson, who's been doing a lot of stuff for UIE for many years, but this is the first time he's giving a workshop at one of our events. This is the User Interface 18 Conference, where he's speaking on Monday, October 21, on displaying data in digestible ways.
Stephen, welcome to the SpoolCast.
Stephen Anderson: I am glad to be here, Jared.
Jared: I'm so happy to have you. You're giving this great workshop. We were talking about it, and one of the examples that you brought up that you talk about is a website called Hipmunk. Now, for folks who don't travel frequently, or maybe even ones who do but don't know about Hipmunk -- it's like chipmunk, except without the C -- it's a website for finding flights, just like there are any number of websites for finding flights, but Hipmunk's a little different.
Why don't you talk about what struck you about its differences?
Stephen: There are actually a couple things that are noteworthy about Hipmunk. One is just the way they present your flight options back to you. Where the majority of airlines and travel sites will come back with this list of all your flight options, they chose to take something very similar to the Gantt model in project management where you show bars and lines over time.
They chose to use a model like that to display the flight results and thus, the first thing I would comment on, that I would note about Hipmunk...From a cognitive perspective, that's really powerful because they're taking this information that when we look through this list, we're having to hold in our short term memory and they're actually putting it on the page.
And so, those things being, Is there a layover? Yes/No. We're having to hold that in our memory when we're choosing a flight. If there's a layover, how long is it going to be? How many layovers are there going to be? Which airline is this?
All of that information is now in the page so you don't have to take up space in our short term working memory to think about that. We just look at the visual and make a more effective decision.
That's the first reason I bring up Hipmunk.
Jared: It's interesting because the display visualization is so concise. I actually use Hipmunk every week. I'm always traveling. I'm getting on an airplane every week so I'm always booking a flight. Hipmunk is the first place I go because I can see the whole layout of the flight and all the other flights that I'm choosing between right there.
They color code it by airline and since I fly on particular airlines, I've memorized their colors. My eyes now just go straight to those colors.
Little things like USAir has this nasty habit of thinking that people can actually get off a plane, go across the airport, and get on another plane in Philadelphia in less than 30 minutes. Unless you are superhuman and can do what Superman does and fly around the planet to reverse time, you probably can't do that because if the first flight is late by even 10 minutes, there's no way you're going to make boarding on that second flight because boarding starts 30 minutes ahead of time.
You'll get there just as they're closing the doors.
I can tell immediately just by looking at Hipmunk that those flights are not flights I want. It takes me just a millisecond.
Stephen: When you see it in text, it might say half an a hour, it might say two hours. It might say 15 if you're going through London Heathrow. That's a line of text, but when you actually see it as a line, we look at it and just by the link of the line we know, "Oh, that's a longer flight. That's a shorter flight."
It's so much easier to process from a cognitive perspective. That's the first thing, the first reason I talk about Hipmunk. The other is a really small detail that they rolled in that people always laugh at when I share it. I think we laugh at it because we like it. It's the label in the short features that they have. They have the usual suspects.
They have sort by price or by durations or by number of stops and these types of things. But then they have this sort by agony feature that they have built or rolled themselves. I don't know what data they've rolled into it but they've probably taken the on time status historical record. They've probably taken things like the number of layovers.
All this data they've rolled their own sort by agony sort option so that the flight that sort to the top are supposed to be the least agonizing flights.
That's the other thing I point out. I don't see a lot of sites going to extra stuff and say, "OK, what can we come up with that would be a really unique data point? Rather than just pull the data that everyone else is getting, so in this case all the same travel information and then focus solely on the rendering, what else can we do with the data with the API to create something new, and in this case, of value to our users?"
That's definitely the other feature that I point out.
Jared: I think it's cool. If you look at the FAQ on the Hipmunk site, it says that it's primarily price, flight, duration, and number of stopovers. That makes perfect sense. If you can get a shorter flight to the same destination, you might even pay a little bit more for it. It's always interesting to look down the list when it's sorted by agony, which is the default, and see a higher priced flight jump to the top and go, "Oh, yeah, of course." Because it's a direct flight or it's shorter.
What's interesting is it's just a little bit difference in price. If it's a lot difference in price it goes down. It really seems to be smart.
I think what they're doing with that is helping people make the default path be smarter decisions. I don't know if you knew this, but their hotels site...They just added hotels. Hotels let you sort by ecstasy.
Stephen: [laughs] That's fantastic. Hotels are a great way to create a data point like that because there's so much, I'd say more so than flights, there's so much that goes into picking a good hotel. A lot of it does have to do with personal preferences or convenience as things that are harder to sift through and find, but when you go through a hotel site, that's the stuff we're usually looking for.
The reviews and do the rooms smell like cigarette smoke even though it's non-smoking and those types of things. If you can figure out a way to surface that data, and I don't know if they do in that case, but that would be incredibly valuable.
Jared: Here's my question. When you hear about this, you think about it. You go, "Yeah, of course. Why would we do it any other way?" Yet, nobody else has done it any other way. Hipmunk wasn't the first travel website. In fact, travel websites have been around for 10 years, for more than 10 years. For 15 years.
The question then becomes, "Why is it that Hipmunk was different?" I guess my interest is if I'm going to want to think about changing the way my industry works, what can I learn from what Hipmunk has done?
Stephen: I think, one, you can always say things are building upon what's come before them. It's a progression and Hipmunk was made possible by the Orbitz's and Travelocity's and Expedias of the world. They were made possible by the Savers of the world and so on. You can definitely go back that way.
But I think if you look at any creative idea or any innovative idea, it's usually...There's a couple things that are happening. There's the intersection of things that shouldn't go the other...Or that people wouldn't expect to go the other. If you've grown up in the world of travel and you're looking at travel sites all the time and that's all you look at, you're less likely to come up with a Hipmunk or whatever's going to follow Hipmunk.
But if you start to look at all these other worlds, all these different ideas you're going to say, "Why not do a travel site that's sort of like this." And so, if you look at...I think I referenced earlier it's kind of like taking a Gantt chart and bringing it to travel. I think it's the intersection of those two ideas, a Gantt chart and travel information that very well could have been the origins of that.
I think it's that intersection of two different things that we find behind most creative ideas or most inventions or innovations. That would be one response.
The other...This is just common of teams and individuals everywhere. It's a much more fundamental idea, this notion of functional fixedness. It's the simple idea that we start to get used to seeing things a certain way and we say, "That's how we do things. Why would you do it any other way?"
We become very focused on just how things have been done in the past. This is something I talk about in the workshop and the talks that I give on this. We tend to fall back on a lot of design patterns, things like a data table or a list view, things like this and say, "If we need to push data to the page, here's four or five, six design patterns we can choose from."
We never really step back and say, "Wait a second. What was the content that we were using a grid view to push to the page? Could that content have been served in a better way?"
I think when you really dig into content, and we talk a lot about putting content first now, I think you'll actually find there are better ways to publish the content to the page than just using these conventional design patterns that we've grown very accustomed to.
There's some design patterns that are fine for what you're publishing but I would say, in many cases, you can step back and say, "Now that we have the time to take the extra effort and look at the content, is there a better way to serve this particular kind of content?"
There is an example that I use where I talk about shopping for digital cameras. The same experience happened to me when I was shopping for a TV last summer. The tools that we have out there on most ecommerce sites for sorting and filtering and narrowing down our electronics options are really quite terrible.
We have basically fact sets and things that can narrow our options down but there are much better ways that I think that we could design for sifting through 100 or 200 different electronics options where the differences are pretty minor, where we want to get to that best fit and the best match for our budget and our needs and the features that we want.
At the very fundamental level, that's what I try to challenge people to think about this. There are better ways to serve the content to folks.
Jared: Part of it, like you said, is having a pattern library that talks about what all of our alternatives are. Out of that, I think, there has to be some principles for which we can then choose which patterns are going to work best for us in this situation.
Also, I think that that helps us understand the patterns in terms of, "This pattern's useful if you're trying to do this but if you want to do this other thing, this other pattern is going to be more useful."
When you're doing your work, are you building up libraries of this stuff? Do you constantly collect what people do and catalog it?
Stephen: Yeah, absolutely. What I'm about to say is going to contradict everything I was just saying a moment ago. We have these conventional UI patterns that we've grown accustomed to, maybe 2,000 or so that we commonly use. The same thing goes for the visualization world where there are several dozen visualization patterns.
People who offer these frameworks, they'll say, "If you want to show relationships, here are the four or five ways to visualize relationships. If you want to show changes over time, here are the seven ways to do that." And so, you're starting to build up this pattern library for very specific types of information you need to show.
That's great. That definitely broadens the palette we're working from. But I actually like to not think about that structure and just...I throw it all in my inspiration, all the sites and the screenshots I find. I throw them in a file with no organization and then when it comes time to...When I'm working on a visualization or just a UI, whatever it may be, I'll open up this folder and I'll just flip through all these things.
It's this almost random, not structured approach where I say, "Here's the problem. Here's the content I'm working with. Here's the interaction I'm working with. Now let me flip through all of these patterns." Occasionally, you surface a pattern you would have never thought would have fit with the content or you'll surface multiple patterns and Frankenstein them together to create a new pattern.
That, to me, is where a lot of the creativity comes around. In the workshop, I talk a little bit about the common patterns like if you want to share relationships to this. But my bigger focus is, "Look. Think of creating a box for all of the inspiring ideas." Maybe it's a Pinterest folder or a board where you start pinning ideas. Whatever it may be, and every time you find an interesting way to represent information, add it to it.
Then on the next project you're on where you have some complex relationships to show, go to your inspiration board or your inspiration box and folder, flip through that and it's very much like going into your closet and trying on different clothes until you find the one that fits for that day.
In many ways, it's just saying, "Add to your pattern library." But I don't think so much about structuring and organizing the library as much.
Jared: I think that that's an interesting idea, that you are just keeping them uncategorized on purpose so that you have that...I guess it's a freedom, right, that you have from it.
Stephen: It is, and I guess what I've found is it's definitely a challenge for a lot of folks. I think in a lot of areas, we want to know, "If this is the scenario or these are the conditions, what should I do?"
I'm challenging people to say, "Well, I don't want to give you a prescriptive framework. I want to give you a more fundamental way to think about things so that you can come up with your own frameworks and your own responses."
That's definitely a hard idea. We're challenging people to come up with new patterns versus rely on previous ones but I think, at the end of the day, if we're going to really look at the interaction of the content and the experience that we want.
I think this is how we should approach things, starting with that content first approach and not the, "I have content. Let me fit it into one of these preexisting molds," which is, I think, what we have in many cases today.
Jared: I think that's interesting. I remember doing this exercise years ago. Leah Buley was doing a workshop at one of our conferences and I sat in on this exercise where she was having people sketch ideas for a product. She had pages that had six little phone sized screens, sheets of paper on them. Six ups.
She was saying, "I want you to spend just a minute or two on a version of the screen that you're working on and then I want you to just draw it over again and just keep drawing it over and over again." But then she would say, "This time try it as if everything was done only with list boxes. Now try it as if everything was done with drag and drop. Now, how would you do it if it was a touch interface? How would you do it if it was full body motion, what would the interactions be like, like Kinect?"
She kept changing the inspiration and I'm wondering if that's what you do with these. You pull out these things. You say, "What if I took this current thing I'm working on and I made it like this one thing that's in the folder that uses sort by agony." Does that make sense?
Stephen: That makes complete sense. In fact, I do the same thing with all of my regular website webop work where I think the frequent question I use is, "If I was a five-year-old, does this UI make sense?" That's a great one because five year olds are just starting to read but they don't have a high tolerance for lots of text.
Just throwing on this really ridiculous constraint for our need, which we were having trouble with the 1024X768 browser dimensions, so let's make it even worse and see if that forces us to abandon the ideas we have and come up with something new.
I think that's the whole goal of these exercises, is to get out of how you're seeing things currently and see it from an entirely different perspective, even seeing the same exercise done with very strong brands. You might look at the same web app for example and say, "If Apple was designing this, how would this look?" I think that's where most of us tend to gravitate these days.
But then change it to, "If Disney was designing this very same time tracking app, how would Disney design it? Or if Virgin America, Richard Branson was designing this, how would he design it? If the team behind World of Warcraft designed this time tracking app, how would they do it?"
What you're doing in the process, whether it's through a constraint or this role is you're being forced to change your perspective and see things in a different way. That's a fundamental idea, I think, that can apply to any creative challenge.
Jared: A lot of designs involve large amounts of data -- the airline and hotel stuff that we've talked about, choosing a TV. The projects we work on have sales data or teacher data or student data or whatever the data is. The knee-jerk reaction for all of that stuff was basically, "Let's just put everything into a big table."
Stephen: Tables are easy.
Jared: Tables are easy.
Jared: Yeah and there's any number of table widgets that you can get so it makes coding easy. It's like, "Let's put it in a table and we'll figure it out later."
Stephen: Yes, I see, because I work primarily with web applications, I tend to come in before any other designer has touched it, where it's been primarily an engineering team. No offense to engineers out there, but engineers love their tables. It seems to be the default way to publish information to the page.
Jared: It reflects the schema, right? Reach row is an element of the database, a piece of data and each column is a field for that element. And so, it just naturally works.
Stephen: It works. It's a mirror reflection of how the database is set up, which is the problem because a lot of people don't think like a database thinks. The people building this do, but not the people oftentimes who have to use these systems.
I've seen crazy things. I'm sure you've seen, as well, where it's not even a lot of data. Maybe by the time it's published to the page it's one or two rows with maybe 10 columns. You look at this and are like, "That's an odd way to publish this to a page," when it's a one row data table.
You see these things all the time in the early stages of these products. I was giving a talk last week in Bulgaria, giving a talk on micro moments, these tiny interactions around...I think we've talked about before. I was critiquing Dropbox and the way they showed my...I renewed my Dropbox subscription for the year and where the page that said, "Give me a simple receipt," said, "Yeah, you're paid up using this credit card and then these numbers. Here's what you paid for."
I got this six row data grid that...I had to read through it and sure, all the information was there but they could have said it in a much more succinct way. And so, I made an offhand comment about tables and not always the best way to serve information to the user.
At the end of it, I got a Q and A from someone saying, "Hey, basically what's your beef with tables?" My first question back to them was, "Do you happen to be an engineer?" He was like, "Well, yeah." I was like, "OK." But the theme out of that was make sure you know your audience so even in the visualization talks I'm really careful to say, "Look, if your audience is engineers, at the end of the day don't take away the table from them. Don't take away the .CSV file. This is what they like. This is power, in many ways."
But if, and this is more often the case, if your audience is not engineers or not sales jockeys that are...A term I use for these sales people who love their Excel spreadsheets and macros and all. If that's not your audience, then you probably need to think about other ways to show or surface the information. Grids and rows and columns are just not how we naturally think."
Jared: My theory, tell me if you agree with this or not, my theory is that a lot of times things end up in tables because no one's thinking about, "How will the data be used?" And so, the thought is, "It's cheap to code a table and it gets all the data visible. We'll put it out there and see how it's used. Then we can redo it." But then nobody ever goes back to see how it's used so it just stays in that table.
Stephen: You hit the nail on the head. That's exactly what I see over and over again. It is the easy way. It's the path of least resistance. It leaves you the freedom to come back later and tailor it. The question is, do you actually set aside or prioritize the time to do that? It's this tension between prioritizing for use and the experience and usability and prioritizing for new features that we see all the time.
Jared: I almost wonder if there's some way we can start programming teams such that when they see a table or when they have the urge to use a table...Maybe it's when they insert the code to use a table. Something pops up and says, "Are you sure there's not a better way to do this?"
Stephen: I would love that. Let's build that plugin and ship it to code editors everywhere.
Jared: Yeah. What are some alternatives to tables that you've seen recently that are actually pretty slick? The Hipmunk swim lanes are a great way to do flight data. What's interesting about that is that it has multiple dimensions. It has a start time, and end time for the entire flight so you get to see the duration of the entire flight.
It has the start and end of each leg. It has the layover, which is neat because when it's big enough they can insert the city in it. They're color coding for the airline. They show relative to time so you can see the early flights or the later flights.
One of the things, for me, is that I never like to take the last flight out because if...Like yesterday, I was flying out of Miami. There were thunderstorms and if I was on the last flight, I would have never gotten home because my connections would have been totally screwed up. I had managed to get an earlier flight and it still took me an extra three hours to get home but I made it.
I learned that the last flight is the one that is the most risky. I always take at least the second to last flight. I can look at Hipmunk and immediately tell what's the last flight and what's the second to last flight. I just know which one I want to go with, whereas in a table, that's really hard to sort through.
Stephen: Absolutely. I've done the same thing on these international flights where there's a layover. Knowing sometimes that the first flight might be delayed for whatever reason. For a 10 hour international flight after that, the second leg, you don't want to miss that. I'll actually look and say, "If there's a problem. If this first flight's delayed, does this airport...Are there other backups, other options or is that the one flight that's going out of..."
Sites like Hipmunk really help you see that right away.
Jared: What are some other examples you've come across recently that are alternatives to just tables of data?
Stephen: I don't know if you recall, but the earliest version of this conversation, this talk I gave, I actually pointed out the problems I was having with my long distance data plans. Whenever I travel internationally and turn on the data plans, I'd come home and I'd get double billed. I'd get billed for the amount I was paying, the 100, 200 megabytes I'd paid for but then they would bill me for all the data usage as if I'd not paid for that 200 megabyte plan.
I'd get on the phone and we'd have to go through, line by line, working it out. It was always in my favor, the errors. They'd say, "Oh, yeah. We owe you money." But in that talk, I just did a quick visualization, I think it took an hour in keynote, just a mockup. "Here's a great way for me to actually track and monitor my actual data usage to see if I'm going over or under."
I even accounted for things like prorating. If you're paying for two weeks because you're only traveling for a week but you know you need half of the 200 megs. I got into the details on this. Here's a simple visual way that you could look at it on your mobile phone to just get an idea of if you're going over or under or if you need to keep an eye on your data usage.
I was looking at a friend's Android phone last year. He said, "Hey, check this out." He showed me pretty much something in line with what I had sketched, some curves and a line showing your average usage. It was pretty much the visualization that communicated what dozens of lines of data in a grid in my cell phone bill never did.
And so, that's one example I've seen. That one I had to point out because it was very similar to what I had sketched in that presentation.
I'm trying to think. I'm going out to Mountain View a lot these days, and I fly into San Francisco and taking the Caltrain down to Mountain View. I don't live in San Francisco so I don't know all the trains like people who live there do. There's some, the bullet trains, whatever that will not stop at certain stops but they get to places much faster and others that stop at every stop. I think there are three different train lines.
I can't remember the exact URL for this but I found this visual to all of the Caltrain rails and all of the stops. You basically choose northbound or southbound. And then going across it has the time of day. Then you can kind of see all the stops. These little dots on the line.
It's hard to describe with words, but when you see it, it's like, "Oh, yeah. There are the trains and that's the one I need to take." It's interactive. You can hover over it and get the details for when you need to grab a particular train to arrive at...Back at SFO or back in San Francisco at the appropriate time.
That was really nice. I've seen other people deal with this in data, but being able to actually see it through these lines and dots on a graph, that was actually much more useful, to see it in that visual way.
Jared: Google Maps has a new interface for their public transit directions that show basically...It's Hipmunk similar but it shows you which piece you have to walk and which piece you ride on. You can see when the different trains leave. It's really cool looking. Again, it's a swim lane type approach so if there's...
You can take a bus or you can take the train or you can just walk it. They compare those all side by side on a timescale.
Stephen: Is that in the desktop version or is it on mobile?
Jared: It's on the desktop version.
Stephen: I'll have to check that out. Usually, I'm pulling up transit directions on my phone so I've probably missed out on that one.
Jared: If you go to Maps and then you try and find your way around New York or San Francisco and use the public transit options, it's actually pretty cool.
Stephen: Living in Dallas, we're pretty spread out, unless you have a car you're out of luck.
Jared: I don't know what you think of this public transit thing. That's my Dallas accent, by the way. It's sort of Dallas Russian.
Stephen: You need to throw in a "y'all" in there.
Jared: Y'all, I don't know what you talk of, this Russian y'all. [laughs]
Stephen: I'll tell you where I'd love a visualization of something. I have more and more books now on my Kindle app for the iPad. Also I have a mix, because my wife and I share the account. There's the books she downloads. There are the books I download. Then there's the overlap, the books we both enjoy reading.
But it would be really nice, like in the real world, just to be able to arrange my books on a bookshelf or even just folders or groups of some sort. I think all I have is sort by author and sort by title. Even things like when I go online and I push in a few new books or samples of new books, I want to go in and look at that. There's no place where the new books can show up. Instead, I have to look for this flag for new.
And so, if you have more than 100 books, you're scrolling through, looking for that little flag that says, "New" or "Sample," or whatever it may be. Something that's more like arranging books on the bookshelf like we do in real life would be really helpful about now, just to start to organize all the books we have in the Kindle app.
Stephen: Maybe if someone from Amazon's listening, maybe they're working on that right now.
Jared: Please. Now. Fix. Yes. I'm there.
Stephen: There's another one that's very small. It's not a rich data example but it's really helpful. I use a time tracking app called Freckle. I think the URL is LetsFreckle.com. But when you enter your data, it's actually represented back to you on this month view calendar and rather than put it in numbers and things, they actually have this pie chart. If you only have one project it's not really a pie chart. It's just a round circle.
The size of the circle is the relative time that you've put in that day. If you put in three hours, the circle's going to be smaller than if you put in eight hours. Then the other thing that's great about it is if you do have multiple projects then it's actually a pie wedge and you can see different colors for the different types of projects.
Again, the scale for the hours and the wedges for the different projects. It's a really nice visual way.
Jared: So the size of the dots still changes based on the hours?
Jared: That makes sense why it's called Freckle, because I guess they all look like little freckles.
Stephen: There you go.
Jared: That's very clever.
Stephen: It's a great visual way to see your time and also a very small thing that I point out. Let's say it's Tuesday afternoon. I'm entering my time for Tuesday. I entered it yesterday for Monday and as I hit submit, I see that Monday circle get twice as big but there's nothing on the Tuesday circle. It's an instant visual feedback loop.
I can say, "I have my date wrong. I just added seven hours to Monday when I meant to add it to Tuesday." That's happened probably three or four times over the past, and just because of that visual cue I was able to correct the error.
That's another thing I'd like to point out, is when you use visual things, we're more likely to be able to spot and then fix the errors versus if it just added a row or a line that said, "Yeah, you entered seven hours on Monday," I wouldn't have noticed, probably.
It's our sense of vision. We pick up these little details, these subtle changes. Of all our sense organs, it's the most highly developed one so little changes like that in size, scale. You'll notice games, like I'm playing this horribly addictive game that my wife got me onto called Candy Crush. I think it's popular right now. It's kind of like another form of Bejeweled, in many ways.
But as you're getting lower and you're running out of turns, it has a counter with all your turns. As soon as it gets to five or lower, it starts blinking. Even though it's really small, it's a really tiny dot. Your eye catches it and it adds to the anxiety and the tension but you see it, just that little bit of animation. Again, because we're attuned to pick up those subtle visual changes.
Jared: Stephen, this has been awesome. I can't wait until your workshop at UI18.
Stephen: I am looking forward to it myself.
Jared: It's going to be great. It will be fun to learn how to read these things and use them.
Everyone, you need to go to UIConf.com, U-I-C-O-N-F.com and look up Stephen's full day workshop on displaying data in digestible ways because it's going to be a fabulous way to take your applications and designs to the next level. It's just, as you can tell, Stephen is a joy to listen to and he's so smart.
Steve, thank you so much for being part of this.
Stephen: Thank you, Jared.
Jared: I want to, once again, thank our audience for listening in and taking part. We love to hear your thoughts. Send us your thoughts on our UIE blog -- we love to hear what you have to say -- at UIE.com and please let us know what we could do differently.
That brings us to the end of today. I want to thank you for listening and thank you for encouraging our behavior. We'll talk to you again next time. Take care.