 Thanks so much. Thanks to the alumni for coming out for this event. Greetings to those in webcast land who are watching this. That means that, from a privacy point of view, you should be alerted that, should you participate tonight? And I hope you will. It may leave the confines of this room. So it's an inverse Las Vegas situation for any contributions that are made. So I thought what I would do is share a few thoughts on what I think is an increasing and soon to be inevitable phenomenon that will be touching our lives. Tell you what I think is cool about it. Then tell you what I think worries me about it. And then tell you a couple of things about what we ought to do about it. And then we can talk about it. So that's the basic layout for tonight. And maybe as a way of getting into it, it's worth dwelling on this experiment. Last year, a graduate student at NYU named Casey Kinzer came up with the idea of taking little cardboard robots that she made, painted an all-important smiley face on them, went to Radio Shack and got one of those little electric motors that propel them forward at a constant but slow rate unable to turn, attached a flag to them written in the first person that says where they want to go, and then released them on the streets of New York City. That's what you get with sociology experiments. And to her delight and surprise, she found out that people intervened to help to get the robots where they wanted to go, rather than, say, calling DHS. And here's one such chart of a robot's path starting in the northeast corner of Washington Square Park, over 40 people coming to help the robot to help it get to the southwest corner where it means to go. Now, this is one of those classic sociology experiments where it's like it's really interesting, and then it's like, I don't know what to make of it. It feels good, but now what? And in my own sense of it, it reminded me of internet routing. The way the packets get from one place to another often passed along from one entity to the next without necessarily having contractual agreements through and through from one end to the other. But maybe more deeply, it shows that if you ask people for a small favor nicely and with a smile, they will often deliver. And it's that aspect of turning to strangers for help that forms a spine of a lot of the work that I do. What are the environments that we can build and cultivate online that, in turn, will bring out the humanity in people, rather than the usual lawyer or law professor's question of, who are the bad people among us and how can we punish them? Or the more recent, who are the good people among us and how can we incent them with cash rewards and tax breaks? This is a more holistic way of trying to tap into a certain force. So count me among those cheering this sort of phenomenon on. I'm not the only one. There's been a slew of books recently from people saying how powerful the crowds can be, whether as brains coming up with answers that even the experts aren't accurate about, or as sources of work, of progress, of labor in the world. My own book, The Future of the Internet and How to Stop It, covers some of these topics. I devoted an entire chapter to Wikipedia. And this was the original cover suggested by Yale University Press for the book. I was not that fond of this cover, not least of which because I had another book on which I was a co-editor coming out in the same month called Access Denied about internet filtering by governance around the world. And it too featured a large hand saying stop on the cover, despite the fact that these are actually left and right hands, it seemed like too much in a period of time. So I asked Yale if they wouldn't mind a different cover. So they came back a few weeks later with this proposal, which has a nice kind of 8-bit character to it. But I confess that while the word stop is in the title of the book, I didn't understand why a large traffic light was the focus of the cover of the book. I then got the idea of making the medium the message in a book that relies on the wisdom and power of the crowds, why not crowdsource the cover? So of course there's a website for that. I went to worth1000.com, as in a picture is worth 1,000 words. It's a website where people who are really good at Photoshop and photo manipulation gather, and it turns out the number of people really good at Photoshop and with nothing to do is very large. So they have contests, free logo design contest. That's right, it's a free logo. You don't have to pay to enter, and you could win $300. 28 entries later, and I had a number of ideas ranging from the sublime to the less sublime about what the cover of this book should look like. This one I thought was particularly evocative, a guy named Ivo van der Int in Holland is the one who drew it. And without ever meeting him, I selected that as the winner, and then had a brief conversation with him that said this had a bit of a kind of Calvinist flavor to it. Was there anything that might indicate hope? And he went back to the drawing board, and he came up with this. No, this. A little bit of choice in our future. And I was extremely happy. Some people took great pains to try to figure out from the cover which way the switch was pointing as to my ultimate opinion. Yes, it is facing forward. But who knows, maybe this is the wrong track and this is the right one. And so I was pleased. And what, it cost me 300 bucks. Not bad. But that's when the academic in me started saying, if it feels good, there's surely a problem with it. And I sought to figure out my own thinking around why I would be at all disquieted by the processes I celebrate in the book and the benefit of its literal production. So to help my thinking around it, I basically tried to organize the different ways in which we might harness faraway minds to work for us in this rough pyramid. The idea is that towards the top of the pyramid might be tasks and challenges that require very specific expertise, very rare. And the number of people you might draw to help you is very low. And that's why it's very narrow. And correspondingly, the amount of money you might have to pay to engage those minds might be high. As you go down the pyramid, you see a broader base here, which are tasks that may be much more accessible to whoever might want to do them. The number of people, in turn, is much larger who can do them. And the price you might have to pay could drop to zero or below. So I want to give you now a tour of the pyramid, starting at the top. Just to enumerate some of the services, then I'll tell you some of my worries. Then I'll tell you a couple ideas for what we might do about it. At the tippy top of the pyramid, we might put something like the XPRIZ Foundation. This is a private foundation founded with a lot of money designed to inspire private industry, motivated, perhaps solely by cash, to solve very difficult problems, such as to produce the first reusable vehicle that can go into space and come back fully under private auspices, $10 million prize for the first entity or company that can do that. What they found, and there was a winner for that XPRIZ, was that the amount of money invested overall by those vying to win exceeded the amount of the prize, which meant it was kind of economically efficient. Everybody's happy, except the company that comes in second wins nothing and invested a lot. But maybe they can patent something out of it. That's the XPRIZ. You really have to know something and have a team to get involved with. One layer under that, we might put something like InnoCentive, a project by Eli Lilly, the big pharma giant, which has created a marketplace that they describe in the following way. There are companies represented here by armless people. And they are called seekers. And they will send awards to scientists represented by this silhouette cutout in It's a Small World After All format, who will then solve the challenges mediated by a laptop produced in 1992. So the scientists register as solvers. They submit solutions. The Seeker company selects the best solution. In Ascentive issues, the award amount to the winning scientist slash solver. It's worth a visit to InnoCentive. It's an incredible array of challenges and bounties out there. Here's one of them, Browning in juice. It turns out that for orange juice that is not the tropic cannot totally fresh, always refrigerated kind, that you keep in a bottle for refreshments in an event like this, it can turn brown. And it's still drinkable. It's just not that great looking. The state of the art in how to deal with this is to create a label from top to bottom that fully covers the bottle. Bad surprise when you pour it out. So OK, $20,000 to the first person that can come forward with a way of ameliorating the Browning in juice problem in bottles. 270 solvers have stepped forward with their individual project rooms to interact with what may be an anonymous company. And they stand to gain the $20,000 one winner should that be solved and selected by the deadline. OK, that's pretty interesting. And of course, by the way, let's be clear, it's IP transfer. In exchange for the $20,000, you give up not only all rights you might have in any innovation contained within what you did, but maybe even the fact that you did it at all. It's meant to be as if the company had invented it itself. All right, one layer more down in the cloud. We might put something down in the pyramid, still in the clouds, like live ops. The story live ops tells about itself, which so far as I can tell is accurate, is that when Hurricane Katrina roared through New Orleans, the American Red Cross advertised a toll-free number for people who were in New Orleans and needed help to call and relatives outside of New Orleans to call to try to get paired up with people who were missing. Thousands of calls poured into this hotline swamping the standard people who staff it. The Red Cross got in touch with live ops and within three hours, there were thousands of additional people answering the phone, hello, Red Cross, with the script, with everything they need to handle the call, all through live ops. Now, where were those people? They were not in some boiler room on call, ready to go. Instead, they were live ops contractors. Anybody can apply to B1. You face a battery of tests administered by a computer, approximately 3,000 people start these tests each week. About 30 fall out the other end, having survived the winnowing, which includes how good is your internet connection, how good is your headset, how good is what's inside your head, kind of like a little SAT problem, that kind of stuff. And if you pass all the tests, you can get selected as a contractor. And as a contractor, you get a little menu of, where do you want to go today? When you jack in, you say, I'd like some work. They say, OK, you've just started out. Your first and only option is to take pizza orders. So somebody calls what they believe to be their local pizzeria to ask for something. It gets routed to a live ops person. Who knows where in the world? Hello, Joe's pizzeria. May I take your order? Type it in, hit Send. It goes to the pizzeria near the person. Everybody's happy. You do well at that for a while, and maybe the menu options start expanding. You can start working a drive-through window at the squawk box. People pull up. They ask for the french fries. You upsell them on the apple pie. Everybody's happy. And the friar later isn't overflowing, making you lose your cool. And the person never even need to know that the transaction was handled nowhere near the actual restaurant outside which they are idling. You do well enough at that, and you start leveling up to be, hey, you want to be a Red Cross helper. You want to make some political calls on behalf of a candidate at the 11th hour. Any of it can be ought to bear at any instant through a platform like this. And I think to the workers, contractors, you can't call them workers, to the contractors, there are a lot of benefits. Here's one of the live ops ads, house ads, calling all mompreneurs. There's your own house, you know, because it's got your Coke bottle empty because you've been being so productive. And just off here is the bassinet with the kid inside. Kid starts crying, finish up your call, you can jack out, take care of the kid, and whenever you're ready, go right back in again. A lot of flexibility not found on the typical shop floor. Of course, every interaction that takes place when you are in is monitored to the nanosecond by live ops. They will know how long you took on each call, how that compares to the average, how much upselling you're doing versus others, all the metrics. Six way from Sunday, which is part of what makes it so appealing to a contractor. What might put it one layer below on the pyramid? Well, how about something like Sama Source, for which the idea is to take people living in refugee camps and give them work that they can do on a mobile phone. Maybe there's some task they can be asked to do by looking at the screen and pushing the buttons that is worth it to someone to give to them. And in turn, for which they might pay a penny. What kind of task might that look like? Well, for that, we can turn to a platform intended more domestically, and that's Amazon's Mechanical Turk. Are there turkers among us? Anybody who uses Mechanical Turk? Well, perhaps that will change after the presentation. And there may well be some out in webcast land. Amazon builds Mechanical Turk as useful for those tasks that by all rights, if there had been sufficient funding for artificial intelligence, computers could do by now. But because there wasn't, they can't. But why not have people do them in a way in which we can pretend that it's a computer doing it? Which is why we call it artificial, artificial intelligence. At some point, someone will be able to find a computer that can do these tasks, at which point we will have artificial, artificial, artificial intelligence. But until then, they're called, well, first they're named after the famed Mechanical Turk, kind of like Zoltar on the boardwalk that would tell your fortune, this particular creature mechanically would play a mean game of chess. Decades ago. How did it work? You've probably already guessed. There was a person crammed into the box actually manipulating the Mechanical Turk. Perhaps one of the world's first actual sweatshops. So how does Mechanical Turk work today? You have hits, human intelligence tasks. Put up at, say, a penny a hit. Here's one. Provide related keywords for these images. And you look at this image. And you can't help yourself, right? It's like box, box. Like, yes, it's a box. Have a penny. Like, I think I'll do this again. Trucks, trucks, yes, they're trucks. There's another penny. And before you know it, you get into this rhythm and you're earning money and you're doing something. Many graduate students use Mechanical Turk. It might be for the money. It's not like they're paid that well, but I think it's because it's like the only form of positive feedback they get. So here's another Mechanical Turk question. What is the difference between vanilla and French vanilla? 50 or 60 words has to be entirely your own answer. No links. Just tell us what the difference is. Who wants to know? I don't know. What are they gonna do with the answer? I don't know. What's the answer? I don't know. But if you do know, there's three cents in it for you. And you can then move on to the next hit in the series. It's an incredible range of stuff on Mechanical Turk. In the category of what you might call a meta task, an academic did an exercise in which he asked Mechanical Turkers to write on a piece of paper why they do it and then hold it up for the camera and take a picture and then get paid. I think at one penny he didn't get any takers. He had to go it up to, I think, as high as 50 cents before he got a number of responses. I Turk for Christmas. I Turk to battle insomnia. I Turk for drug money. Just kidding. I'm not sure he's kidding, but it's hard to see his photo. This is such an evocative grid to me because it shows how if you zoom in enough, kind of like your suspicion with things in the world, like you keep waiting to see the little creatures that actually are making stuff happen, like everything turns out to be a microbe. Here's the task. You zoom in enough and yes, there's a bunch of people doing it, but the whole point of Mechanical Turk is to keep things zoomed out. So it's just like a computer. In fact, recently somebody wrote a so-called command line interface to Mechanical Turk. What does that mean? With a single line of text to your computer, you can say, here's a folder of images. Please provide keywords for each one at three cents a pop, deadline two hours, enter. And then boom, answers just start flowing in from the on-call 24-7 computer that is a collection, a hive of human minds waiting to be put to whatever task you have. So what could be lower in the pyramid than paying people a penny? Well, I already hinted at it, it's paying them nothing. How do you pay them nothing to do work? Yes, games. A brilliant computer scientist named Louise von Anne at Carnegie Mellon came up with the idea of designing a game that people would play that in turn would do useful work, such as labeling images. So here's his version from 2006 of the ESP game. Player one and player two unbeknownst to each other, but they know there's another player out there, maybe or maybe not at the same time, are shown an image. And you guess what you see in the image and when your guesses match, you win points. What are the points good for? Anybody? Nothing, absolutely nothing. But it does keep score and it has the all-important thermometer, although it does appear to be filling backwards. It's academics, what could you do? You accrue points. People love earning points. They play the game a long time. In fact, as Louise says, there are many people that play over 20 hours a week. His thesis advisor asked him to check and see if somebody were playing and coming from an address ending in .edu to cap them at 40 hours. Tell them to get back to work on their thesis or go to Mechanical Turk or something. Amazingly attractive. He then runs the numbers. He says 5,000 people playing simultaneously could label all images in Google in 30 days. And there are games on services most people haven't heard of like MSN and Yahoo, where 5,000 is no problem. Google took note of this. Google bought the ESP game. Today it's called GWAP. I wonder what committee and branding enterprise came up with that. And here it is, the thermometer is looking more three-dimensional. And there's the label images game. There's also the tag-o-tune game for things like services like Pandora that have genetic algorithms for associating music with other music. Here's a game in which you hear a tune. There's somebody out there that's also hearing a tune. You type what you're hearing. They type what they're hearing. And at some point, you race to see if you can answer the question whether you are listening to the same tune on the basis of the respective keywords. And if you are right in your guess, you win points, which are good for nothing. You can see here the top scores. Today, JC has 36,000 points. This particular correct answer worth 60 points. I would do the math, but why not put it out to Mechanical Turk for someone else to do. Here it is now taken to its most logical extreme. Problems in what we call electronic design automation. The set of problems that includes how to cram transistors on a chip ever more densely so that they can be smaller and faster and more efficient, cheaper. At some point, you have computers doing that, designing their next generation, Terminator style. But even they run into trouble. So these researchers found a way to take a very difficult problem for which they themselves do not know the answer of the sort of cramming chips and convert it into a game in which you click around the periphery here and the boxes are either gray or yellow as you click. That in turn changes the circles in the middle between green and gray. If you can manage to figure out how to click the boxes to turn all the circles green, you win. And if you win, it will turn out epiphenominally that you will have solved a problem in electronic design automation and they're happy too. Pretty amazing. What could be below that? Well, either games in which people pay you for the privilege of playing, like Ultima or World of Warcraft, where people pay an annual fee so they can mine virtual gold and then sell it on the market, or this is a new project from the US Geological Survey where it's not a game at all, it just follows people's tweets. The USGS now scans Twitter constantly looking for words like house just shook. When they find them, they triangulate to location where available and with it they are trying to estimate when, where, and how severe an earthquake has occurred. Greatly complements the data they already get from the seismographs, which you might think is a little more direct evidence of the earthquakes. All right, so that's a tour of the pyramid. Let me now just put some worries on the table. Here's worry number one and I'll introduce it since we're at the law school, hypothetical style. So here's a question for you. This is the PBS Kids site. So imagine your child or grandchild or little sibling going to this site on a Saturday morning, staying out of trouble, playing one of the four games they offer here to play. All right, that's the site as it actually looks. Let's just make a tweet. All right, now your child or grandchild is playing game number four in the bottom right corner, the EDA game. How many of you have a problem with that? Too handful. How many of you are completely fine with it? A handful. How many people are confused and tentative? Everyone else. This, of course, is a matter of public policy, is the state we expect and hope the populace to be in so we can go off in the corner and do whatever we want and by the time people have their wits about them, it's too late. In fact, often in technology, I think the phases are too early to tell. That's what we're in right now and then too late to do anything about it. So okay, I'll come back to this hypothetical. We'll do a re-vote in a little bit. This is utterly hypothetical and by that it means the PBS site looks like this. This is for kids. My question was, suppose PBS, which occasionally needs sponsorship money. It happens. Suppose they want some money and they kind of deal with the EDA people to put the kids to work. Who in it? You've just, did you change your vote or were you already against it? Oh, well you're still against it. I didn't win someone over. You tried to make it a little more specific. Yes. Or part one. Yes, I would say the age of people playing this probably, you know, 14 to 25. That's the demographic. Have you seen what they're watching today? All right, here's another concern. This is a different problem on inoscentive, the people that are trying to figure out about the browning and the juice. Here's the one where they're seeking pyrazolopyridinodiasonines. I don't know what they are either. Suppose you know how to fashion some for a price. That you can then anonymously deliver in exchange for your bounty and your promise never to disclose that you did it. Can you think of stuff that pyrazolopyridinodiasonines could be used for? For which you would be unsettled to know that this transaction was taking place. Yes, you don't have to be Joe Surnau, the famed maker of 24, to come up with some scenario by which you're like, huh, if the person doing the job is sufficiently pulled away from the overall context, you can get people doing stuff that they might otherwise have no interest in doing. Now, I don't know, how often do we actually have a chance to think about the moral valence of the work we do? For many people, it's never. For others, it's at the time you take an employer. It's like, am I gonna work for this person or company or not? RJ Reynolds gave me an offer. Okay, once you start working for RJ Reynolds or what's it called now, Flavia? Those are the coffee people. Altria, yes. Nothing could be wrong with that. That's Philip Morris. Is RJ Reynolds still called RJR? It's called Nabisco now, isn't it? It's the Crocker people. But I digress. You make a choice at the moment you wanna work for them but then you kinda know what's under the roof when you go and work. I once went to a wedding in Winston-Salem and asked for a non-smoking room which meant there were not free cigarettes in the room. There were still astrays. They just didn't give you free tobacco, which I was appreciative. All right, let's start shifting the worries now maybe away from the participant's point of view and towards greater social implications. This is a website from 2006, version one, in which the state of Texas set up webcams along its border with Mexico and invited people at large around the internet to watch the webcams looking for trouble. Millions of people tuned in to stare at the Texas border. They had an opportunity to click a button that said, I see something untoward here. And if enough people did on the same view, that would be a cue to send the police. And over those millions of views, a couple incidents, turns out not involving illegal immigration but other stuff came to the attention of the police when they intervened. From the point of view of Texas, I don't think it's just PR, an amazingly successful trial. All right, that's 2006. Now it's nearly four years later. And we're starting to see business models built around exactly this. So this is Smart Drive, where you know that kind of thing you sometimes see in like a taxi or a livery vehicle that films the road ahead. You put these into cars like this, mediated through this service. They have little accelerometers in them. If there's a swerve detected, immediately, the last 60 seconds of footage and the next 60 seconds of footage are sent to someone over the internet who's already signed up to monitor. They review the footage and then they can click a box that says, aha, unsafe lane change. Then the actual owner of the vehicle is alerted and within a minute or two, Katie in touch with the driver of the vehicle to express displeasure, at which point the person monitoring says, and he's on his cell phone. Okay. We see in the UK, the world capital of closed circuit television cameras, not just run by governments, but primarily run by private vendors. A new service called Internet Eyes, for which the ribbon, I think, is just about to be cut. This is version one of their website. You can watch Internet Eyes. Here's somebody studiously on her sofa, staring at some random party's CCTV camera, looking for trouble as if it's the Texas border. What's in it for her? Well, you can earn reward money if you find something going on and click the button that you see it and therefore have a chance at reducing crime. Wonderful British understatement. When I was at Oxford, the Thames Valley police had little police cars and under the crest where you normally have to protect and to serve or we'll kick your butt or whatever it normally says in America, it says reducing crime, disorder and fear. I always wanted to go up to a Thames Valley police officer and say, I just want you to know, I'm afraid, but I'm less afraid, thanks to you. I'm glad I didn't. I think I might have had my visa revoked. This is how the website looks today. Better fonts. She's much more active now. She's having a great time and there they are responding to whatever she did. And look, you can register for free with no recurring fees. They won't charge you in a recurring way to sit and watch other people's CCTV and when you sign up, you have five alerts per month allocated to your account, which means you get to press the panic button five times any more than that and it's gonna cost you. You gotta load up with more alerts. Helps cut down on false positives. I'm waiting for the moment that she actually runs outside to a CCTV camera, does something terrible, runs back and reports herself and collects the reward money. It's going to happen. On the subject of rewards, in 2006 again, 2006, very banner year for the planting of the seeds here. In the University of Colorado, April 20th turned out to be a significant day. How many people can guess why April 20th has significance? Okay, clear generational split in the room and it's not a direct graph either. It like bulges in those that were baby boomers in the 60s and then goes down again and then up. So it turns out that on April 20th, for their own reason, the kids at the University of Colorado gather on for on-field and smoke what my dad used to call the funny stuff. Yeah, so their theory is you can't arrest us all. The police are like, that's true, but we can take photos of you all and then put them on the web later at $50 a pop for a bounty. Whoever can identify this woman gets $50. So they did this. This idea, fast forward nearly four years, has been taken up by the government of Iran. This is not hypothetical. This is a real site linked to the government that took pictures that had been taken after the post-election protests of protesters whom they could not identify. And they asked people at large to come forward and help identify who these people were. Now there's a natural ceiling to this just like there is even with the $50 on the great smoke out because the people most likely to know who these people are are also well aware of what will happen should they report them. So it's not clear how much you'll get. But that got me thinking, what if we put this out to a service like mechanical Turk, what would that look like? We know what it wouldn't look like. It wouldn't be three cents if you can identify this person, right? I'm sitting in Kansas, what are my odds? But let's be a little McKinsey about it. Let's figure that there are 72 million people in Iran and let's even do a version that isn't specific to the city of Tehran where we figure the protesters are actually living. We then just come up with a way of taking our database of national ID card photos and processing them through mechanical Turk. First, just label man, woman or child, man, woman or child, penny, penny, penny, lather, rinse, repeat. And it turns out $14,000 later, you have all the photos sorted, done. Now what you do is you take on one side a photo of the protester and on the other side, five photos that match the gender and age and ask the Turk, does this photo on the left match any of the photos on the right? Yes or no, you get a penny. Lather, rinse, repeat. And it turns out that for about $17,000, you can identify arbitrarily any one of 72 million unidentified people through this process and the Turk need never know what he or she is doing. I promised you I would come back with the hypothetical about PBS. Can you see where this is going? So suppose this is game number four at some kid's site. Can you pay concentration and identify the people? How many people would be upset and distraught to see their kid, grandkids, sibling playing this game? How many people would be perfectly fine with it? How many people remain confused and unwilling to commit to an answer? Two, almost for the question I think. Just confused and unwilling to provide an answer. Also confused. Yep, I count myself in the okay, I would be distraught at this camp, but I'm confused at the same time. Because supply creates demand. Once you create a market that facilitates the treatment of the human mind as an uncountable resource, it's like the two scoops of raisins. You can scoop brain power at any problem, limited only by the size of the check you're prepared to write. You start to see the number of applications that previously economic efficiencies precluded. But now you just ladle out the brain power to try to make progress on almost anything. All right, next concern. Also I think in the societal category. This is a real mechanical Turk task in which for $2 you are asked to review natmedtalk.com on your blog. The conditions include that your blog has to be real, it has to have been around for a while, basically it has to be authentic. You have to add an entry reviewing natmedtalk.com and you can't disclose that it's a paid review. Once you have done that, you'll get your $2. Now nowhere does it say you owe them a good review, but let's be clear, on all of these Turk tasks, payment is 100% at the discretion of the seeker. They don't like your work, they don't pay. And how many people are prepared to litigate upwards a one cent failure to pay? You know, it's the kind of thing that lawyers will like, class action. But in the absence of that, it's not like you've got anything. And so I think it's pretty clear. But let's suppose it's not clear, let's turn to another Turk task in which it is clearer. Write a positive five out of five review for a product on a website. Among the conditions, write as if you own the product and are using it. Tell a story about how you bought it. On your way out, mark, negative reviews is unhelpful. This was for a Belkin router. A number of five star reviews started appearing after this Turk task went up. And word of the task leaked to slash dot. People on slash dot don't take kindly to anything. Slash dotters started leaving one star reviews. You get huge piles of reviews, 200, 300, 400 reviews for this router. Not a single person had bought it. Now the lesson from that to me is not, well that showed them, guess there won't be any more of this on Amazon. No, it said okay, you have to be a little subtle about how you do it. But it's also the first time it's done, yes, the outrage level hits here. But then it turns out less and less each time. Oh yeah, of course you can't trust those reviews. Some kind of naive? Why would you think those reviews are real? And before you know it, the entire system has reached homeostasis about it. And that's the end of the story. But now I don't think this inspires the same kind of level of fear and worry that the elections example does. So I need to turn up the heat on this. The way I have done it when thinking about this earlier was with a hypothetical. The hypothetical starts with again something real which is you look at comments as a search term on mechanical Turk, tons of Turk tasks that ask you to just comment on blog entries. Why do they want you to do it? So that you will then leave the name of a spammy site in your wake. Oh yes, my name is Jonathan and I'm from theberkmancenter.org and here's my comment on your blog. And then Google notices that Berkman Center keeps popping up everywhere. And the comments get through because they're not spammy. And sure enough, it's really hard to figure out what is spam in this context. Here are three pending comments on my future of the internet blog. The entry was about the iPhone. I have this thing about the iPhone, I won't bother you with it tonight. Hey, thanks for the information but I don't think the phone is less secure if you get all right. So it's not the most grammatical entry but believe it or not, as a blog comment goes it's like 80th percentile. It comes from appointment software, timepix.com. iPhone is great to have but to use I think twice. I own one and I gave it away. Cute, cute, dog pet groomingsupplies.com. And you start to realize that this is not normal behavior. It's not the word salad we've come to expect of automated spam bots. We just jam words together. If one of these answers happens to actually be really incisive and you know we're talking about the iPhone but it's from dogpetgroomingsupplies.com and I am 99% sure it came from Mechanical Turk. Do I care? Should I approve that comment or not? Approve it? Anybody? A few. Deny it? A few. Confused and increasingly upset? All right, most people approve it but take out the dog pet grooming supplies. That'll show them. Thank you for your valuable comment whoever you may be. Yeah. Interesting. Even dig the venerable service. Are there diggers in the room? People who use dig? This is something where you can vote if you like something online and people just do it because they like voting on stuff. You want my opinion? Why yes, Publishers Clearinghouse. So okay, you see something online and you dig it and then you go to dig.com and you can see what people are digging which means you can have your finger on the pulse of people who dig which is supposed to be valuable. Not supported by dig? Subvert and profit.com where people can buy digs at two bucks a pop. One dollar goes to subvert and profit.com. One dollar goes like the blogger asked to review natmed.com. To a normal digger, an authentic digger who has a long track record who every so often deviates from digging what he likes to holding his nose and digging something stinky in exchange for a dollar and then goes back. Very hard to figure out how to handle something like this. But okay, I had promised you something more scary. And for that I say, wait till this phenomenon enters the real world. So what if dig gets subverted or Amazon reviews? Come on, we've got consumer reports. They're trustworthy. So this is classic, right? How many times are there websites or other ways in which you are impotuned to lobby some representative or member of parliament about something perfectly legitimate lobbying and grassroots organizing efforts? So what if instead we started paying people to do this? Not buying their vote but buying a phone call with their particular member of Congress saying how much they hate the bill according to the talking points they've just been handed. Now that was hypothetical. This was another one of my, it could happen, it will. So it turns out it has since happened. Get healthreformright.org is one of these, not so much grassroots but more like AstroTurf groups that is sponsored by the insurance companies. Together we can get health reform right. Here is the required text that will be included in your message if you sign on. As the Senate considers healthcare reform I'm writing to express my strong opposition to a government-run health plan. I'm concerned, blah, blah, blah. That's the required text. What will make you send it? Well it turns out that in an unrelated phenomenon there's a game card Farmville on Facebook. I won't even try to ask if anybody plays Farmville because I'd be remiss if I didn't mirandize you first. In Farmville you go into Facebook then you further log into Farmville and you're like you have some carrot seeds what would you like to do? You're like box, I mean plant them, plant them. You plant the seeds and like a day later it's like you have carrot tops, you're like yes. And then at some point with enough care like those old Tamaguchi things that the kids had that little eggs they had to take care of it's time for harvest and you go to market and you get money and what's the money good for? Nothing. Maybe there's a buy new seeds, yes, cycle of life. So there might be a larger lesson about real money in this but in the meantime it has happened. The health insurer lobby gethealthreformright.org started paying Facebook Farmville players in Farmville currency good for carrot seeds if they would lobby not their Farmville representatives but their real representatives against healthcare reform. How do they get the currency? There is an open market like frequent flyer points. Suppose we wanted to give you all 500 points for your visit like you run a car, you can buy them wholesale, Zango will sell the points so that people can then in turn dole them out in exchange for stuff like this. And you're just like wow. Like a Facebook Farmville about Farmville I could live with it but you now see that the blood brain barrier has been broken and once that happens I don't know what next. All right, last concern and this now just a little more broad less meant to be the kind of sharp terror that I've hoped to inspire with the past two examples more the vague disquiet that you'll be able to dismiss in a minute or two. And for that I think about a little tweak that YouTube made. Recall that YouTube aggregated tons of videos together that users submitted it was a community and Google bought them for a lot of money after which the YouTube founders filmed themselves on YouTube celebrating outside a restaurant jumping up and down with excitement. Good press all around. So YouTube was not the favorite of a number of content publishers because their stuff would get online and content publishers want to take it down and they get into big fights about this big suit against them from Viacom still pending. In the midst of this YouTube introduces a revenue sharing program. So that if you are Viacom and you see a video of yours up and it's getting tons of views you can either press a button and boom it's gone or you can press a button and Google will start sharing revenue with you from any ad words that get displayed next to the video in which people click and they'll share that with you. Totally up to you Viacom. If you want it down it's down if you want to get some money we're in. Okay they have now taken that program and extended it to everybody else. You don't have to be Viacom now anybody uploading a video can click and enable revenue sharing button and as the video gets hits you start making money if they click on ads. Now my reaction to that is like great power to the people, funding source for amateur video wonderful. Am I worried about it? You start to see a crowding out of those video community sites that previously had no currency at all for the sharing that was done. So let's think about it by analogy. And CARTA is one of the few projects ever actually abandoned by Microsoft. Microsoft was like we had in CARTA we are going to stop it. It's like whoa who replaced you with a different company? What killed in CARTA was Wikipedia. That is widely acknowledged. Wikipedia was an encyclopedia is an encyclopedia that anybody can edit and is free to the world. But suppose in 2001 just before Wikipedia was founded in CARTA with the benefit of hindsight could go back in time and make a tweak. Here's the tweak. In CARTA users of 2001 we welcome your suggestions on how to redraft and improve and keep timely our articles. For which we have thousands already we just want to keep making them better. If you make any change or suggestion to the article and we accept it we'll pay you five cents. Now I don't know if that would have worked it's a total thought experiment but you could see that could work. There could be a loyal core of people who do that and earn money that way. Now try to start Wikipedia for which somebody says let me get this straight. You want me to edit do what I'm doing on in CARTA except I don't get paid anything. What? I believe that whether or not we expect a given act or transaction to be what I call perhaps too valiantly a mercenary one a lot depends on just the status quo and as we investigate and experience new means of interacting with one another if there is no prior market okay we might actually do what Yochai Benkler calls a gift economy out of it. If there is a market it may be awfully hard to then later say wait a minute maybe this should just be freely shared. Suppose there's a company that like is a package courier company bins of packages would you like to help get these packages to where they're going? There's $20 in it for you upon successful delivery. At that point I see one of these robots on the ground you're like you just want a freebie don't you kick it over that kind of thing. So there's this worry about crowding out and I even see it think back to the French Vanilla versus Vanilla question 1998 the news group alt.coffee news groups were message boards publicly on the internet that people participated in for fun and somebody says does anyone know the difference between Vanilla and French Vanilla and somebody comes back with the well intentioned but utterly inaccurate guess that it's probably beans that have been French roasted. So okay that's wrong but they're just helping one another there's no money flow in any of that. Then you see the way in which mechanical Turk can handle it where the disembodied question appears and you answer for your three cents or there are hybrid models we actually see them the more you look for them Yelp, TripAdvisor, you're kind of helping but there's a promise of reputation and there's some dot com over it believe it or not Amazon is also in this space. They've created not Farmville but Askville. Ask a question, get real answers from real people. Think about it compare that to the tagline for mechanical Turk right? These are real people acting like people. Mechanical Turk are real people acting like computers take your pick. So how does it work? They say it's a place where you can share and discuss knowledge with other people. Epistemic conversations are for favorite kind. I really love talking about knowledge and you can learn something new every day or help and meet others using your knowledge. Askville even helps you learn by giving you cool tools while you're answering questions it's all about sharing what you know and what you wanna know. So go ahead and meet someone new today and askville. Oh I was gonna say it's not grammatical. Yes of course this is just an example. Believe me if I tried to be comprehensive on the examples we would be here a while. But yeah of course it's just like a buzz or ardvarkvark.com instant message version of this. You sign up for Ardvark for the opportunity to be interrupted while you're working by a robot that says hi do you know the difference between vanilla and french vanilla? You're like go away it's like okay I'll come back with another question later. You're like why am I doing this? I don't know I've signed up I haven't turned it off yet. So Askville is all about sharing and caring and knowledge but then you start reading the frequently asked question document here's one of them. What are experience points? I'm like wait a minute experience points are from Dungeons and Dragons. Like no no we use experience points to determine how knowledgeable a user is. They're broken up into various levels then you go to experience points quest and quest gold in the facts. So quest gold what are you talking about? It turns out there's this hugely complicated scheme by which you are awarded. I am not making this up quest gold. Can I redeem my quest gold for anything such as oh I don't know money? They're not redeemable for anything at this moment but hoard your quest gold. And in the coming bad years when the rest have guns and blankets you'll be sitting pretty. So here's Askville what's the difference between vanilla and French vanilla? For which I can't help but say can we just answer the damn question once and for all? I guess that's what Wikipedia is for and yes egg yolks I guess the quest gold is more motivating. So fascinating though right? One of the real world's analogies to these different hybrids and models and spectrums is blood donation. You can give blood at the blood mobile and get a sticker or you can go to a different location and sell your blood to the singing people for cash. Is the cash good for anything? Yes, cash is good for everything. The blood that you get for the sticker if you're the blood bank is of higher quality than the blood you get when you buy it which is kind of like whoa the economists are still trying to figure out that one. Well I mean there are theories it's true but it's like huh the people that don't need the money don't ask for the money. It's a little counter standard classical economic theory. But of course economists are the ones worrying about like the problem of blackmail and by that they don't mean that blackmail happens. They're like why is blackmail a problem? Somebody is offering you a fair price not to disclose a true fact hello. So that's economists for you. But we digress. So what do we do about this? Well just a couple of thoughts and then let's open it up for a little discussion. One thought is just to treat at least the participant point of you worries as basically a labor law issue. And all the stuff we know from labor we just put over as best we can. Classic cyber law 1990 maneuver take the old law and just make it apply to the new space and hopefully whatever your instincts already are about labor law you can just incorporate by reference. If you like to bust those unions well bust them here too. If you think that the unions deserve protection and the employers can be exploitative great they can be here too would be the theory. This is a message board called Turker Nation not run by Amazon where the Turkers gather to talk about turkey. And so this seasoned Turker is complaining about not getting paid and about the quality of the work no more slave wages. Which another seasoned Turker replies the best way to govern the work you do is to learn to say no. Nobody makes you do one cent or less hits. No union on earth can give you self discipline. There's some truth to that. I noted with interest last fall when the Federal Trade Commission published new guidelines they're not allowed to make rules but they can say guidelines about paid blogging and tweeting. Which were meant generally by derision in the blogging and tweeting worlds. I'm just some guy and I tweet at 140 characters and I'm like I like Hostess Cupcakes and then I'm supposed to also fit in this paid for by Hostess Cupcakes like what? I think this is a great idea. Not like Silver Bullet great but kind of help establish the norms before they gel great. Because this is saying it's not fair. It's deceptive and you bet there's money involved whether it's a penny or not there's big money to the entity that is the seeker paying for these tasks. And come on the FTC has like eight lawyers they're not going after it like tweeters. They don't go after like companies that spill huge amounts of pig sewage in North Carolina. Anyway, we digress. So what they will do Mayee is go after intermediaries like Mechanical Turk who sponsor tasks that say things like you cannot disclose this as a paid review and say uh-uh, come on Mechanical Turk you gotta clean that up. And what's so interesting to me from kind of a scholarly point of view is that in nearly every other area of my work I don't like intermediary liability. I don't wanna see intermediaries like internet service providers conscripted to become the internet police and stop people from copying this or that. I worry about all sorts of things having to do with getting them into that habit. I don't like when Wikipedia has to worry about getting it hit itself from a defamation suit because somebody randomly came on and changed James Seganthaller's entry to say that he did something he didn't do. So it's weird to me to feel my instincts completely flipped here. And suddenly I'm like go get him FTC make those intermediaries sweat. And I'm still trying to puzzle through why my view has changed so much and whether it reflects something about the views I have on these other topics that I thought were well settled. At the very least I'm interested in trying to find the good stuff amidst the worries. And I feel like this act still stands out as such a good one. And in fact we can take even the worst examples that I might have and try to come up with the good counterpart just like to this good act I came up with the bad counterpart. I was talking with somebody about the Iranian election hypothetical putting it out to Mechanical Turk. His name is Ben Rigby. He works for a website called The Extraordinaries. The Extraordinaries is a website that's a clearing house for micro labor of the mechanical torque sort for nonprofits where they don't pay anything. But on your lunch break you can do some mindless stuff instead of solitaire. Maybe not on your lunch break. Just do it while you're working. That's usually when you play solitaire. And do some good out in the world. So Ben Rigby at The Extraordinaries after the earthquake in Haiti thinking about the Iran hypothetical we had discussed came up with an idea for exactly that where people would submit photos of people who were missing. They put up all the images that were coming off the wires from people milling around after the disaster had struck and then you ask anyone in the world to help match the images of those missing with the images of people who survived. He got about 20 leads out of this process from thousands of people who spent time looking through the photos. The template is exactly the same as the Iranian election hypothetical with one crucial difference. Here the people on the site doing the work knew what they were doing and why. They weren't just asked to do it as a generic game. And to me that's an important piece of what I hope can become best practices for this nascent but I believe someday extremely powerful industry. I end with the most ambiguous mechanical Turk task I've found yet. For 50 cents you are to do something kind and then take a photo of it. And the author conceives of it as a kind machine. People get poured into a funnel processed by gears and then hearts pour out the other end. It's somewhat unsettling. So do something kind and take a photo of it and with this then for whatever arbitrarily large check you're prepared to write you can get that much amount of random acts of kindness in the world. My first reaction to this was it's amazing. That's great. Wow. Somebody putting money to good use just like my monities would want where the person helped doesn't know the helper doesn't know who she's helping. Wow this is terrific. And then I thought the next time somebody helps me tips the hat, smiles and walks away. As soon as they turn the corner are they submitting the video for the 50 cents? And if so do I feel differently about that transaction? And in fact about the way in which the authenticity I've come to expect in the down moments when I'm not watching television or television advertisements certainly not when I'm watching cable news. When I'm not otherwise interacting with the world of commerce these moments themselves now have just become one and the same. Thank you very much. So I have just after 6.30 would it make sense to do maybe like 15 minutes of questions and then call it an evening? Yeah. Any questions or reactions or thoughts? Yes sir. Tell us who you are. And if you, I don't know if people can hear but there's actually this whole U.N. conference. Look at that. Oh wow that works. So all this reminds me of a 15 year old named Marcus Arnold actually who's a pithy activity on a site called askme.com for self appointed experts on things actually made him askme's their top related legal advisor all without ever setting foot in actual law school. And I suppose that's a long way of asking. He clearly must be stopped. Yes, yes. He threatens our business model. Yeah the difference between meritocracy and aristocracy in a digital age and how we're gonna measure that. And I suppose while we're at it it's worth noting that we're sitting on the campus of arguably the greatest and most valuable mind share in the world. So how does the ivory tower endure? Too late for flattery. So great question. I think the zero with principle the ground level principle has gotta be any task that a 15 year old can do for fun equally well to what someone else is charging $500 an hour for more power to the 15 year old. That's your meritocracy versus aristocracy point. Of course the question it immediately raises is the certification merely that like the universal life church says that you could perform weddings or is a 15 year old just knowledgeable enough to be dangerous? How many people can say whether his advice was helpful? How much tracking do we do for any of these services where we interact with people to see whether what they said bore fruit? And to the extent that the efficiencies of networking everything for which this is basically just one more example can help us create structures of fairly painless and automatic accountability. May not have to be personal accountability. I don't need to know his name but if I know that this is the guy that has been answering the other 10 questions and these are how the answers turned out that might tell me a little bit something about whether I can trust his next answer. So I would be interested in harnessing the energy of 15 year olds that can act like lawyers so long as the lawyers they can act like are real lawyers rather than bad lawyers that look good. And occasionally you can tell the difference. Other questions or even reaction to Andrews? Yes. My name is Justin. Good evening. I had a separate question actually. You gave some pretty frightening examples of what can happen when this is taken in the wrong context. Now, how do you exactly offer a solution to that? I mean you could police buttons and in many cases over policing can be just as frightening as under policing. So I guess my question is where would you draw the line? Yeah. Would you draw a line somewhere along that pyramid that you gave? The more I think about it, the more I feel like the pyramid is both too precise and not precise enough to help us map out solutions. Cause it's true that over the course of the talk I mapped out so many different phenomena. Who knows if a solution for one would even apply for another and if the right way to separate them is by level of skill or number of participants. It might be, but it also might not be. The things I'm looking to, and again I wanna be clear, I don't feel like I've solved this, but the things I'm looking to include in no particular order interventions with the intermediaries, exactly the approach that I believe descriptively to be very powerful in other areas of cyber law, but normatively to be, I think we should use that power lightly. I promised I wouldn't rant about the iPhone. One reason I don't like it, this isn't a rant, it's just an explanation, is that it is tethered to its maker in a privileged way. At any point if Apple or anybody who can serve a warrant or subpoena upon Apple wants to monitor or change the way the phone works, it's done like that. And we do not have that much experience with devices of this sort. This is an artifact of the 21st century and of saturated internet and cellular connectivity so that devices can exist if and only if they're in touch constantly with the mothership. So that kind of intermediary control worries me. And for more about that, chapter five of my free online book, ad-free, supported by you, our listeners. And yet here, no recurring fees, your first five pages are utterly free. But here, knowing descriptively the power of the intermediary, the clearinghouse for these platforms, that might be a natural place to look to push a little bit to create a certain floor on behaviors which could include such things as identifying seekers. So you can't just be an anonymous seeker, at least anonymous to the entire world or to the platform vendor. Being able to backtrace certain transactions so that if they have this character of the Iranian elections hypo, we can figure out who did what and blame them if we wanna blame them. Perhaps the intermediaries are the right locus through which to have a rule about disclosure of certain things. And maybe even, there's a site called Elance, which is another one of these outsourcing sites if you're a freelancer. Elance just so happens as a business decision to implement a minimum wage. You can only go so low in who you wanna pay. If you wanna pay lower, go to another site. And Elance also has an escrow system. If you're a seeker and you're gonna set up a task that involves graphic design over a period of a couple of weeks and you don't like the end product, well, too bad, you will have escrowed at least half the price and if they put in the hours, they're gonna get the upfront deposit. So these are examples of businesses that believe it to be good business. In that case, maybe the market will just solve this. I don't know about that, but you can see at least the power of the intermediary to set standards and tell those who don't wanna meet them that they're not allowed in. Force it into a gray market if it's going to be that. And I think in the game space, many of these games are themselves structured by nonprofits. I didn't talk at all about Louise Von Ann's other work, which has to do with cap chas, the squishy green letters or other colors that you have to type to prove you're a human and not a computer. You still might be a spammer given mechanical dirt, but you're at least a human spammer. And he started something called recapcha where you type two words instead of one. One is the test word. The other is a word drawn from a scan of an old book that a computer was trying to understand and it was too ambiguous. So when you answer the double captcha, you're helping to scan old books, but you know that or have an opportunity to click through and be told that. I would love to see an opt out. You know what? I'd rather not help scan old books, whatever reason, I'm anti-book. I know that's now going to be taken out of context. It's like, see, I knew it. I knew it all along. But these are some solutions. So let me quickly enumerate another category of stuff which has to do with, I hate to use the phrase consciousness raising, but it's new enough that our own compasses are really uncertain. But I believe the essence of what you might call basic philosophical tools or in law school applied philosophical tools, which are intuition and reconsideration. You come up with a hypothetical. You ask when you'd be uncomfortable with what your grand kid is doing. You try to figure out why. Then you sharpen the hypothetical and eventually you come, maybe it'll be right or wrong, but at least you can start sorting out your own crazed compasses to why you like or don't like what you do. You write a book, you persuade people and before you know it, pbs.org has to think twice before doing this, but they can be perfectly comfortable doing that. And so I'm up for that. And the libertarians among us should be totally happy because this is not calling on the government to regulate anything. It's just saying let's be part of the marketplace of ideas that help shape what the floor will be. I certainly don't want to just be resigned and say water always flies, it's absolute lowest level, let's just give up, it's gonna be whatever it is. And in the political realm, I would love to see ways that we can be aware of the way that last minute boiler rooms, right? I can go on to whether it's mechanical turk or live ops or some hybrid of the two, who knows what it will be. If I'm ready to write a check for $20,000, I can get 200 protesters in front of Austin Hall with signs that I have defined what they will say in their own handwriting, shouting slogans that I have pre-specified. And when the news covers it, there is no easy way to tell whether it's real or fake. That we need to deal with. I really believe we have to deal with that. I don't have the solution to it yet, but it's coming right at us. Yes, sir. How would the ability to bring pressure on the intermediary help solve the Iranian problem? Well, at least it would mean that the task of sorting through those images, if submitted to mechanical turk, if you had to say who you were, all right, maybe they'd have to go to the trouble of getting a third party to launder it, you have to say what it's being used for, I guess you would have to put them to the trouble of coming up with a fake cover story. I mean, they'd really have to go to the trouble to lie. And at the point they start lying, this actually gets back to another possible approach, which is to out-turk the turkers, set up a bounty for anybody who discovers someone cheating on the rules of commissioning mechanical turk tasks. You find somebody who's paying less than five bucks an hour, we give you 50 bucks as a reward for outing them and ditto for this. And then somebody within the ministry comes and collects the award. So that actually is in a way trying to use crowd sourcing to police crowd sourcing. Maybe that's how you would do it. But again, I make no pretendin' that this is like we're done onto the next. It's a really deep and hard problem, part of what makes it good. Yes, sir? I'm Joshua Gay. I work for CK12 Foundation, free and open educational resource project. My question has to do, I guess, with degrees of indirection in particular. I look at the Iranian problem as a computer scientist, as a person with a sort of machine learning background, AI background, and I say to myself, oh, well, if I set it up the way that Jonathan did, let's do it for good. Let's find missing people, let's match to maybe refugees or people who are asking for asylum and get them back or something, right? And we use all of that as training data to then have a better classification system, a machine learning algorithm, to then match, you know, arbitrary faces for whatever purposes we want. And I know this is possible. I mean, I did it quite well for Plankton, where we had hundreds of training images on Plankton, and then we could characterize what type of Plankton it was, and I've done similar work with Plankton. It's not a service, it's you're actually talking about yeah, see Plankton, images of Plankton, and then being able to categorize them, and similarly, it's- Do you Plankton? Yeah, I mean, these are all the same thing. Plankton, you know, I've done it for mammograms. It's the same technique called ensemble classification that the Netflix Prize just used. I mean, is there any feedback you've gotten on these degrees of indirection? This is what Google does. I mean, you look at 1-800-GOOG-411, a free 411 service that they offer. It's great. It works really well. A free 411 service that normally costs a lot. But, you know, it's training data for their voice. Yeah. You know, it's why they are able to I'm sure- A lot of things. Is there something, I mean, that seems to be the important thing to me is that, you know, can you do whatever you want with it later? Well, there are- And a big matrix of numbers. Yes. So one question is, you could come up with a perfectly and authentically innocuous task which the data is used later for some bad purpose. And there, we'd have to actually craft the disclosure rules so as to, at least if you go into it with this evil motive, you still have to disclose, darn it. Now, the larger question is artificial, artificial, artificial intelligence. What happens when computers get good enough that we don't need people to mechanically turk anymore? The real question is just, where will it be cheaper? You know, it's gonna cost you X to do it by computer with this level of action. It's gonna cost you Y to do it with people. People may turn out to be cheaper than the machines even once you know how to do it with the machines. Depends if the software to do it is patented, if the databases of the sort you're talking about are proprietary. You know, Google might know how to do voice recognition. It doesn't mean everybody else does. Google will license it then for some fee. But I can also have, there was a service called SpinVox. Would take your voicemails and turn them into text. And SpinVox was like, how do we do it? We call it the queue system. And they had like this picture of a computer. And of course it turned out there were people in centers just typing in the voicemails they heard. There's other privacy issues and all sorts of things. But the key thing I don't wanna lose here. I guess I have it as a question, although I think I've got my own answer. It's when you ask a human to do an act, there's a moral valence to the act that they ought to be in touch with. When you ask a computer to do the same act, I can't attach to the computer that moral valence. Which makes me say somehow, when a human is the instrument of some bad thing happening in the world, unknowing to the human, it is a worse, less just, more tragic situation than one in which the same ill befalls a target. But it's mediated through tools rather than through people. But let's be clear, mechanical Turk and live ops. You remember the picture of live ops? It is these screens with the headsets. It's meant precisely, I won't say dehumanized. It's meant to unhumanize the process just like mechanical Turk is. That's the piece that I worry about, even though you're right. If this AI makes it so you can identify an Iranian protester through automatic image processing in 10 seconds, all right, we lose, that's a problem, but now it's a standard Oppenheimer problem of technology can be used for good or ill. People can be used for good or ill, but we wrong them. God, who knew I was a Kantian by having them as means rather than ends? Kind of thing. That's to me the distinction which I know in computer science, the purpose is to blur it. It's functionally the same thing. Ma'am? But try the microphone, sorry. There's a button at the base, there you go. I create websites and platforms for nonprofits to help them work together and remain independent. My question has to do with behavioral economics. And a lot of what you talk about is people will do bad things for money. But a lot of times people do good things, not for money. And there's a fair amount of evidence of that in the body of behavioral economics. And certainly your blood example is a very good example of that. We're a bunch of healthy people are willing to give a blood for free because it's the right thing to do. So I have more of a sense of optimism in the human spirit than maybe was shared as part of your talk today. And I'm hoping that maybe there's an aspect here in behavioral economics that didn't get woven into this that you thought about. Well, I share your optimism. I offer as another reference, you can look for it online, a talk I gave called The Kindness of Strangers. It's all about this. So I totally share it. What I also believe, and I think this is totally consonant with the findings of behavioral economics, which in turn are indexed to decades old findings from behavioral psychologists. Tversky and Kahneman, these types. That say that so much depends on the environment around you as to whether the good you know you want to do gets elicited. We have environments that bring out road rage in the sweetest of people and you just don't recognize them. And we have other environments, one famously in the Netherlands where almost all road signs and rules are removed and people are left to their own devices to navigate the city. So the most critical prediction mayhem, actual outcome, I think a halving of the pedestrian accident rate. And what the engineer Hans Monderman found was that people had to take responsibility and of course their own safety or car was on the line as they drove. And it meant they had to be much more alert when they drove. They had to make eye contact with pedestrians and it gave them the chance because there was not a preexisting rule to say no, why don't you go ahead. They felt it was freely given. They felt better letting the person, no, no, no, why don't you go ahead. That kind of thing. Whereas if you have to let them in, it's like damn it. So I'm with you. I just see in these environments when you're creating Farmville, it's like the matrix, right? You just have these studs along the wall ready to build the set. What that set looks like. That environment will determine how much of the goodness both of us believe are within people will be brought out. And I would love for the CEOs of the Silicon Valley firms designing these environments to bear that in mind. I think it's in their economic interest to do it but I also think it's in humanity's interest to do it. Maybe a couple more questions. Yes, sir. Yeah, my name is Tom. In Picasso, I can put all my family pictures up and I can label one of them. It'll go about identifying that same person in all the pictures. Yeah, it's kind of creepy and then you're like, great. And I bring this up because it would be a simple matter for the Iranian government to take a couple of shots of all their citizens and use the same technology to identify who's who. So is the problem the internet or is the problem the Iranian government? Well. And are we trying to fix the wrong problem? That's point number one. Point number two. We do point number one first just so we don't lose it. So point number one, I say first as a lawyer, sue them all. We don't have to worry about the hierarchy of who's wrong here. You are right that the Iranian government in this example is the wrongdoer of first interest. They are the direct wrongdoer if that's how your equities line up. The intermediary sometimes bears the blame but this gets back to most of the time I don't blame the intermediary. I'm like, don't blame the internet, the same wing but the same air upon, the air that chokes is also the air upon which wings beat says John Perry Barlow. And it's like, yes you go John Perry but because we can shape it in different ways I would love to see it shaped in ways as intermediaries that try to minimize the bad and bring out the good. And at least respect to the people whose brains are being harvested. Boy that really sounded more, whose mind share is being tapped. Yes, sorry, part two. LinkedIn has a feature called the answers where an individual, where LinkedIn itself is a consortium of professional people and there are fairly high restrictions in terms of membership and credentials. And those answers are questions that are put out there to one's connections and in those connections can come back and sometimes and for free come back with some very valuable answers. So I think that's an example of some. Well and what LinkedIn is tapping into and Ardvark does the same thing is that if you mash up a social network with the project of answering questions the theory is you get more success because you will more likely do a favor for someone you know or a friend of a friend. And Ardvark or LinkedIn will say this friend of a friend is looking for X, can you help or refer? And you always want to do something, you don't want to help so you might as well refer. At which point now your credibility is on the line to some random fourth party to help this random first party whose friends with the second, I mean. You know, now is it a good idea? It might be. One worry is that all of the answers gathered that way tend to be limited to the social circle or to the person who receives the answer. I'd much prefer that archive from 1998 of alt.coffee where the answers are available to everybody and I take it that actually Askville the answers are available to everybody. So there's something to be said about trying to share it broadly but maybe invoke the answers to the questions through social networks. Could be a certain feature in that. Amar, do we have anything coming from the webcast people that's worthy of note? If that's the case, then let's take one more in room question. Lara. With the Berkman Center. I wanna know what role do you think social norms could play in tackling this and if there are incentives that could be provided to increase those social norms like making it public that a Tucker has just done a hit that was a five star Amazon rating. And did so without having bought the product kind of thing. Well, as we know from our colleague Larry Lessig and others, Professor Ellickson, norms are very powerful things. And they have the benefit of being so suffused that once they get set, they really can control behavior. Larry's classic example is he never wears a dress to school. If he wanted to wear a dress to school, he wouldn't because of just what he would get back for daring to do so. That's a very powerful rule that confines his life as surely as law does as surely as other things do. And of course, an absence of a norm can make all hell break loose, which is what publishers of intellectual property feel is the story of the past 15 years. And kids, they don't respect stuff. And even when facing the sternest of penalties, they don't appear to alter their behavior because they're crazy kids. So yes, I would love to see norms fashion. I think in a less pithy way, that's what I was saying when I was talking about before things are settled, doing this intuition and reconsideration and making the case out there for how it should be. Now, should that be embodied in a particular system of naming and shaming of turkers who take on tasks that fall outside the norms I think are important? Maybe, maybe not. But even just having turkers feeling free enough to act under their real names and having their reputations be portable so that good work on Turk could mean a job at LiveOps and good work on LiveOps could be ported elsewhere. You actually care about your digital trail. Right now, you're not allowed to take your LiveOps record with you. So you could work for them for three years, level up tons of times, and if they reserve at any time, they can just cut the cord, you can't really prove your worth to ShmiveOps that's ready to hire you. So when I talk about structures for accountability, I think what I'm talking about mostly are norms enabling things, at least for people to come forward and proudly beat their chest and say, I just did X. I just caught this person who has this screen name. I don't need to know who they really are engaging in this behavior. Let's all shame them together kind of thing. On that happy note, why don't I say first to thank you to the Harvard Alumni Association for chartering this and thank you to the Berkman Center for the webcast and for participation in the event. I understand there's an event this Thursday at the Berkman Center. Amar, do you wanna tell us about that? Yep, we are actually hosting a wire side chat with Professor Lawrence Lessig this Thursday. He's gonna talk about fair use remix and politics. So that should be really exciting. You can see all the information on the Berkman Center's website just Google Berkman Center and go to the event section. Got it. In the Berkman Center homepage, cyber.law.harvard.edu, you can sign up for events and stuff like that. And with that, I say to you all, thank you so much for coming out tonight and we stand adjourned. Thank you.