Microsoft Research Podcast - 001r - Snippets from the Revolution – An Interview with Dr. Jaime Teevan

Episode Date: July 10, 2019

This episode first aired in November, 2017 - Dr. Jaime Teevan has a lot to say about productivity in a fragmented culture, and some solutions that seem promising, if somewhat counterintuitive.Dr. Teevan is a Microsoft researcher, University of Washington Affiliate Professor, and the mother of 4 young boys. Today she talks about what she calls the productivity revolution, and explains how her research in micro-productivity – making use of short fragments of time to help us accomplish larger tasks – could help us be more productive, and experience a better quality of life at the same time.https://microsoft.com/research

When this podcast first aired, Dr. Jamie Teevan was a principal researcher in the productivity group at Microsoft Research. Now she's helping to define and create the future of productivity as the chief scientist of Microsoft's experiences and devices. Whether you heard this show back in November 2017, heard it again for the first time last summer, or are just now getting your productivity on, I know you'll enjoy episode one of the Microsoft Research Podcast, Notes from the Productivity Revolution.

You know, sometimes when people hear about the work that I'm doing, about sort of taking these tasks and fragmenting them and helping us make use of our mobile time, they're like, Jamie, you're going to ruin my life. I'm going to have to work all the time. It's like all of a sudden I can't like sit quietly in line at Starbucks. I have to be doing work then too. And that's not actually what I'm trying to do. You're listening to the

Microsoft Research Podcast, a show that brings you closer to the cutting edge of technology research and the scientists behind it. I'm your host, Gretchen Huizenga. Our guest today has a lot to say about productivity in a fragmented culture and some solutions that seem promising, if somewhat counterintuitive. Dr. Jamie Teevan is a Microsoft researcher, University of Washington affiliate professor, and the mother of four young boys. Today she talks about what she calls the productivity revolution and explains how her research in micro productivity, making use of short fragments of time to help us accomplish larger tasks, She calls the productivity revolution and explains how her research in micro-productivity,

making use of short fragments of time to help us accomplish larger tasks, could help us be more productive and experience a better quality of life at the same time. That and much more on this episode of the Microsoft Research Podcast. Hey, Jamie. Hi, Greg. Give the listeners a short description of the research you do. I do research thinking about how to use artificial intelligence to make people more productive. So I'm essentially thinking about how you can complete your tasks better by working with the computer. Your research addresses solutions to getting work done in an era of

fragmentation. Yes. At least in part. Why do you think things are so fragmented now? And how is it different from like, quote unquote, bad attention spans of previous eras? Well, so certainly things are very fragmented right now in that we get a lot of interruptions. We actually do a lot of self interruptions as well. So we might be sitting at our computer and you'll get a little toast notification telling you have new mail and you're like, oh, I want to check that. Or your phone might beep to tell you somebody mentioned you on Facebook and you'll go check that or your phone will ring or somebody will swing by your office and want to talk. So there's all sorts of interruptions that are available, not just from the people around us, but also from our electronic devices. Things are also fragmented partly because

we have mobility. So we have access to information and our work anywhere we are. So if we're at a meeting or if we're standing in line at Starbucks or if we're, you know, commuting home, we have access to information about our work available there. Which is why you see everybody with their head in their phone. Which is why you see everybody with their head in their phone. Is it different? Has technology affected the kind of fragmentation that we experience? So it's certainly different in the short term in that kind of traditional industrialized information work has been really focused on having these good

solid blocks of time for us to get work done. We push towards that. That's how our schools are developed. That's how our work is structured. And we try hard to kind of cling to that. We, you know, block large chunks of time on our calendar so that we can get us focused work time, or we take Facebook vacations or email vacations. We work really hard to get at a sort of unfragmented time. But in some ways, when you look way back, you know, sort of pre-industrial revolution, you'll actually see that we attended, this kind of fragmentation or lack of attention is a positive thing. If you think about like hunter-gatherers, it's a good thing to be attending to the bobcat in the woods. It might

be harmful for you, you know, so that kind of fragmentation is actually a good thing to be attending to the bobcat in the woods. It might be harmful for you. So that kind of fragmentation is actually a good thing. It's the way we work. And so that's what we're trying to do with our research. The thing that people are doing right now in the face of this fragmentation is essentially trying to force themselves to work in an unfragmented manner. And what we're trying to do instead is shatter the tasks, fragment the tasks, and then have them go and match the way that we're actually working. So we take these large complex tasks and break them down

into small little pieces so that we can start inserting them into the holes. Let me go back to something you said in a paper that I read. You observed that we seem to be in a constant battle with ourselves for our own attention. Yes. And your research is in part at least providing technological solutions to that. And you've alluded to that just now. But talk a little bit more about, I'm a writer with six open browser tabs. You have my full attention. Yeah. Actually, it's kind of interesting. Even when you think you're focused and paying attention, you have six open browser tabs. So we actually self-interrupt in that way as well, not just going to attend to email, but we only focus

on any window that's shown to us for like less than a minute, you know? So as you're working, you're switching between applications, you're switching across things as well. And actually, each of these little things that you do are essentially a micro task that works to make part of your larger task that you're doing. And what's important in order to be able to move between those micro tasks is to kind of model and understand the context that's in your head that's necessary to complete that task and make sure that's available to you when you move on to the next step. Okay. So people have complained that technology is making us distracted. We can't sustain attention.

We sometimes need to. However, I've read studies that show or say, suggest, I think you can't say show, and that we actually should schedule in breaks and should schedule. So does that, I was reading that, you know, aside from anything you'd done. And then I got to your stuff and I'm like, okay, wait, this sounds like it goes together. Yeah, it does actually. So, you know, sometimes when people hear about the work that I'm doing about sort of taking these tasks and fragmenting them and helping us make use of our mobile time, they're like, Jamie, you're going to ruin my life. I'm going to have to work all the time. It's like all of a sudden I can't like sit quietly in line

at Starbucks. I have to be doing work then too. And that's not actually what I'm trying to do. We want to make people efficient at doing tasks partly by helping them replenish and recover as well. And so we've done a lot of research that shows breaks are really important and that we can also help guide people, not only guide people into being very productive and getting tasks done, but also guide people's state of mind to help them be calm, to help them replenish, to help them get into where they need to be. There was one study we did where we turned off everybody's access to Facebook for the whole day. We had people kind of list their distracted applications and turn them off for the whole day. And we did this over an extended period of time and we looked at how it impacted them and it drove them crazy. And it made them so much less productive because it was just very stressful not to be able to take breaks or get away or relax. And you can be intentional about

those breaks. Sitting and playing Angry Birds is not necessarily the most, the best way to kind of replenish your cognitive resources. Things like taking a walk outside are a really good way to do that. Clifford Nass and Stanford did a study on multitasking. Yep. And it was using university students who claim to be, you know, Olympic caliber multitaskers. And it turned out they weren't. Yep. And he said, no, they're not multitasking, they're task switching.

And each time you switch a task, you have to come back to the other and it takes away time and it's not productive. How does that go together with what you're doing on micro productivity? We work at MSR with a bunch of the world experts on multitasking as well. Shamsi Iqbal and Mary Sherwinsky and Gloria Mark was just up visiting from UC Irvine. And absolutely 100% multitasking is ridiculous, and you should never do it. That's 100% true. Let's level set on what multitasking is. My daughter never doesn't listen to music when she's studying, and she's got texts going on, she gets a Snapchat. She's doing a thousand things at once. And she thinks she's doing them all in a quality way. She's not. Yeah, she's not. So multitasking is a bad plan. Doing tasks serially

is a good plan. The way that this fits with our micro productivity work is we're making the tasks so small that you actually can be doing them. It looks like multitasking sort of at the macro level and at the micro level, you're doing tasks serially. It's about including all of the context and all the information necessary to complete that task in that small little bit. So we actually just ran a study also on this. We built a tool that helped people do writing tasks in small little chunks. So you might copy edit a sentence or you might read a paragraph for flow or that sort of thing. And we had people do that while watching a video.

And then we quizzed them on the video after they were done with that. And so we wanted to see how well were they able to attend to both tasks, the primary task of watching the video and the secondary task of trying to edit a document through either these little micro tasks that we gave them or these using a traditional document editing tool. In both cases, they were actually able to answer the questions about the video just about as well, but they weren't able to make a lot more edits and in general just feel like the cognitive load was a lot lower by using these small little extracted tasks than they were by trying to do it as a large, because, you know, otherwise you're trying to find like, oh, where was I again?

And what was I thinking? Right. Rather than having, yeah, rather than just having what you needed to do up front. So I think this would be a good time to do a level set of the word productivity. And when are we actually ever going to say we're productive enough? Is there some productivity utopia that we're working toward? It is really interesting to think about up until now, though, you know, so productivity is essentially a measure of the output as a function of the input that you're that you put into a task. And up until now, we're really working to try to help people produce more output for the same input.

And it's like there's a sort of time value. You know, how do you do things the most efficiently? So like, can you screw in this bolt as efficiently as possible? Or can you do this action? And like, how do we minimize movement so that you're even more efficient? Those are kind of boring, repetitive tasks. And the good news is we're getting really good at being able to take over those boring repetitive tasks. So in the context of a factory, we can screw in that bolt. We don't need a person to become super efficient screwing that bolt in. We can have a robot doing it. In information work, as we start breaking down tasks into their kind of substructure, we can start identifying those bolt screwing tasks.

And we can watch people do them, just like the robots do in the factory, and learn how to do that for them. And so people become useless for those boring, repetitive tasks. And instead, we become really valuable for the interesting, thoughtful, creative aspects of the task that we can't figure out how to do automatically. And that's really cool. So in our future productivity utopia, we're all artists and creative thinkers. And we're all kind of thinking about, you know, making broad connections and thinking about how things work together and adding our unique human insight into the process. All right, let's switch over. You've used the word sourcing in terms of ways to get work done,

and there's crowdsourcing and there's friend sourcing, but the most interesting one you've said is self-sourcing. And so I'm like, how is that different from the way I usually sit down and make myself try to get work done? No, so it's true. Most of, you know, all of our personal tasks that we do ourselves are essentially self-sourcing, so we're sourcing it to ourselves. We use the term self-sourcing as a play on the word crowdsourcing in the context of microproductivity. So a lot of the early work in microproductivity or thinking about how you take a large task and break it down into these small tasks was done in the context of crowdsourcing. So these crowdsourcing platforms are ways to quickly connect with people online to perform the tasks

that you need. And the problem is because they're sort of short term relationships that you have, that you bring people in to say, help me with, you know, I need help photoshopping this one picture or copy editing this document I wrote. It's not a long term relationship. Generally, what people have found is it's very, you're much more successful if you provide a lot of structure. So instead of saying, please photoshop this picture, you might ask somebody to remove the background and somebody else to, you know, make the people in the picture look better and somebody else to look at the overall composition or something like that. And that structure was useful for bringing other people in. And as we looked at it, we were like, well, actually there's the opportunity to use that

same structure to make it useful for myself to sort of essentially collaborate with myself over time. So how do I provide some contributions now and some contributions later, you know, sourcing myself now and later to produce an output? So then you're breaking it down so that you can see smaller chunks of it and it's not maybe so overwhelming of a big task to do? We've done work looking at the difference between doing a task as sort of this large macro task versus the exact same task as a series of smaller micro tasks. And what we found was when you do a task in micro tasks, you feel like it's easier. You produce higher quality output, actually, because you're externalizing all the work that you're doing in the context of these tasks rather than holding it all in your head as you're doing the larger task. And it's more resilient to

interruptions. So because there's all these sort of task boundaries introduced, you're able to deal with all this sort of incoming other stuff, serializing the little small micro tasks rather than trying to do a large task. So self-sourcing was a funny phrase for me to encounter when I started looking into your research. Another funny phrase was slow search. And I know you've done talks on this. You've done a lot of research on it, basically going back to your beginnings on search and personal search. Can you unpack slow search and what that means and what it's about? Slow search is framed in contrast to fast search, which is actually how most of the search that we

do right now is framed. So we've done a lot of studies looking at the impact of time on how people find information. And one of the things that we found is if you introduce really small delays in the search results that people get, like even by 100 milliseconds, which is actually imperceptible. So people only notice things, they notice delays in their user interfaces if they're like 200 or 300 milliseconds.

So if you submit a query to a search engine and we just hold on to that query and do nothing with it for 100 milliseconds and then start and go and try and return results, people actually perceive those search results engine and we just hold on to that query and do nothing with it for 100 milliseconds and then start and go and try and return results, people actually perceive those search result quality as lower. So they think the results are worse. They interact with the search results less.

They even come back to the search engine less often. And that's not because the quality is lower. It's because we held on to those results for just 100 milliseconds. A delay of something you can't perceive makes you perceive that it's worse. Yeah. Once you figure that out, search engines invest all sorts of money and time and effort into giving you search results 100 milliseconds faster.

Because we invest a lot of money in trying to give you better results, but clearly giving you faster results also makes you happier with the results. So we at Bing, for example, when you run a query, instead of showing you 10 links, so typically the search is described as 10 blue links. If you go and count the number of links, it's actually only eight links. And it's because the page loads that much faster when we dump two links off the page. And that makes it seem better for you. We're smart with it. So this

whole focus on speed is kind of ironic because actually half of our search sessions are multi-query, multi-session. They're long. We spend a lot of time, you know. So these ones where you're like looking for the New York Times homepage, like that's in and out. J.Crew. Yeah, exactly.

But the ones where you're engaging in a topic, those are longest-ended queries. So like 100 milliseconds shouldn't really matter there especially. And so we can start detecting when you're engaged in a longer session. And then we might go and take a little bit more time, give you more results, be a little more thoughtful about what you're doing. You have a foot both in academia and Microsoft research. What do you think are the most exciting opportunities for this field that you're working in right now? I'm really excited. I think we don't yet know how to help people interact with computer systems that aren't always right.

So one of the things that's happening is we're able to do a lot more automatically, but we're getting it wrong a lot. And search is really interesting in this way. It's one of the really few places where we interact with a computer and there is ambiguity. So when you enter a query and you get 10 links or eight links, you know that that list isn't going to be 100% correct. And actually people do trust that list a fair amount. You'll see they click on the first result more and they tend to think that the first result is more relevant than the fifth result, even though it's not necessarily that much more relevant. But we're aware of that and we're aware it might be imperfect. And we are aware that we need to have a conversation and iterate with a

search engine. There's almost nowhere else in our interactions with computers where that's true, where that ambiguity, that fact that it might be wrong is present. And we need to fix that because we can't be providing doctors with support for diagnosing people if they're going to take that to be true. They need it to be, they need to understand the, you know, what the computer is wrong about

and doesn't know and be able to kind of work together that way. And what's more is we need people to help teach the computers so that they can be better next time. And so that each time there's an interaction with this ambiguity, the next time it responds in a little bit more appropriate way. Which leads me to an interesting, you had a big slide deck that I think you've done from a presentation. And you said computers are good at this.

Yep. People are good at this. Yep. So talk about that. So there are some things that computers are really good at this yep so talk about that so there are some things that computers are really good at you know we don't and and i find it very interesting actually just thinking about like ai and intelligence when a computer the things that we don't think of that as intelligence and it's not like a computer can do a really large numeric computation and like if you could

do that off the top of your head i'd be like wow that's amazing but like a computer can do it and you're like all right whatever right you know um a computer in the context of search can look at you know millions and billions of documents and you certainly couldn't do that even if you spent the rest of your life just trying to look at all the documents so that's something that computers are really good at computers are not so good at synthesizing and understanding things that's something that people are good at we're good at. Computers are not so good at synthesizing and understanding things. That's something that people are good at. We're good at seeing the big picture. We're good at understanding connections. And one of the things that, you know, and certainly we'd like computers to be good at that or better at that. And we've got a lot of research going into that space. But I think it's a really cool instead to think about how to bring humans and computers together so that you get this

creative synthesis and big picture insight from people. And you get this like really amazing ability to do large scale computation from computers. And I mean, you know, honestly, we kind of get this in our own world right now in that we're all smarter because of the internet. And so, you know, it's really a way, keyword search is a way that we're pulling together this large-scale computation of computers with our own ability. So we externalize a lot of our knowledge. We don't have to know, you know, where every country in the world is, or we don't have to know every medical term because we can go look that up and then we can use our insights to understand the world better. I like the things that I read about what you're doing, bringing people and computers together. And it's not so much computers replacing humans, but more augmenting and helping.

Tell me what your sort of big picture on that is as you do your work in Microsoft research. What's your vision for what computers and humans can do together? This productivity utopia that you were talking about a little bit earlier, you know, I think we can make the world a better place. You know, I don't know really how to answer this without sounding trite.

I think you can go for it. You know, it's just, yes, we can accomplish all the things we've been accomplishing more. We can do it easier. We can be rich in cognitive resources because we're not doing other things. And we can really be maximizing the way we think about the world and the way we interact with people and the way that we, you know, move society forward. What would you like people to know about your research that you think they might not know and just Microsoft research in general?

Microsoft is a company that touches a lot of different sectors and a lot of different areas. And so in Microsoft research, we're doing all sorts of stuff. So it's, you know, it's not like I do research related to search and task completion, but there's other people who are doing work, you know, related to quantum computing and other people doing research related to, you know, cryptography. There's all sorts of different, you know, we have economists, we have ethnographers and anthropologists, like there's just a broad range of people looking at, you know, looking at DNA and how we can, you know, encode information in DNA. There's all sorts of, it's a really broad

organization because we're thinking about the larger company, which really has a broad impact on the world. What is it that most excites you about micro productivity? I guess, well, you know, I have four little kids. I am really excited about being able to make productive use of my time so that I can hang out with them and also be able to use the small little bits where I'm with them that I'm not attending. Like, actually, I mean, this makes me sound like a terrible parent. I go to the playground with my kids and they're playing on the playground. And

I know some people think you should go engage with your children the whole time. But you know what? I'm kind of like, that's my time for my kids to be running around and for me to sit and be quiet. And you know what I do? Like I take out my phone, check my email, and that takes about five minutes. And then I'm done. And then I have nothing to do. So then I start playing Candy Crush. And I'm on level like 2,500 in Candy Crush. Overachiever. And I would much rather be using that time to be productive because you know what? I'm actually with my kids at the playground, which is nice.'m outside if I could get stuff done then and like take that out of my work day

that would be awesome right well back in the day I mean talk about bad parents it's like they sent us out to play with no helmets yeah well they stayed inside and had a cigarette and a martini right yeah so you know phone is not that bad bad. No. I mean, they need their own space. And it's just the point is, it allows me to put my work into sort of these dead times. And, you know, it's basically defragging my life where I have some dead time. And that dead time exists while I'm at work or while I'm at home. And if I can use that productively, then I can use my whole life the way that I want to use it. You know, earlier I was thinking we feel guilty when we aren't engaged with our work via a mobile device or a laptop or whatever. And then we feel guilty when we are

engaged with the device because we should be with people. And it's this whole, I think we need to figure out a way to get rid of the collective guilt. I agree. That's like my big piece of advice I give when people are like, do you have any advice for me? I'm like, don't feel guilty. And I just think, I mean, like, do you know, you can use guilt as a signal to be like, you need to change, but you should also be forgiving of yourself. And you should be like, I need breaks. I need time at work. I need time at home. Like, just respect all of that variation. And that way, you capitalize on it. Like, if it's time that you're spending at home, like recharge, then embrace that, use that to for what it is. And and when it's time you're spending at work or when, you know, or if it's time, you're like, I need to just like, lay in bed and watch Netflix. Like, that's a fine thing, if that's what you need to do. Yeah. And then this by the same token, stop judging people. You know, if they're on their screen, because you're at the playground looking at some other mom going, Oh, she's probably you know, now I'm on their screen, because you're at the playground looking at some other mom going, well, she's probably, you know. No, I know. Parents are the worst about

being judgmental of each other. But, you know, I'm doing something really productive on my phone, and you're just, you know, surfing. Oh, my God. We actually did a study that looked at how people perceived other people's use of phone compared to their own use of phone. So we looked at device uses in meetings, and we found that people are like, I use my devices for very productive things. Everybody else in the meeting is using their devices to goof off. Right. It's funny. We totally think everybody else is screwing around. It's actually biblical. The plank and the splinter. Yeah. But then again, I would actually also say, forgive yourself for being a little judgy sometimes too. I mean, we're all kind of just, you know, recognize that for what. So the overarching thing is grace.

Yes, exactly. That's awesome. Thank you so much for coming in. My pleasure. I really enjoyed talking to you. And as a matter of fact, we will try to get a little party together and maybe we can all sit around on our screens and ignore each other. Yeah, right. I can like it on Facebook. Yeah, yeah, yeah.

To learn more about Jamie Teevan's work, along with other research that can make your life more productive, visit Microsoft.com slash research.