 So good afternoon. My name is Jonathan Zittrin. I teach cyber law at Harvard Law School. And I want to welcome you to the session on Does an Algorithm Run Your Life? As a threshold question, what's an algorithm? And an algorithm, I think in computer science terms, is a process of some kind, a set of instructions, a recipe, steps that if you follow in a certain predetermined order, lead you somewhere else. Usually we would hope somewhere better. So if you write a piece of software, that embodies a bunch of algorithms. An algorithm, more generally, can be any form of process that we abide by, a convention of some kind. So for example, we might have in our own lives an algorithm that says when a device like this emits a sound, we hold it to our ear, no matter where we are or what else is going on. We might ask in this session that we change that algorithm so that any telephone is set to the vibrate or silent position and should it demand your attention, you ignore it. That would represent, then, a change in algorithm. So two people, myself and Sir Tim Berners-Lee, were asked to speak to the question of algorithms in life. And Tim is going to go first. Let me introduce him quickly. Tim Berners-Lee invented the World Wide Web. 1989, 1990, he came up with a set of algorithms that would say, if somebody wanted to serve up information in a particular format, this might be a good way to do so. And if somebody else wanted to make sense of that information, here's what a browser would look like. And to help the process, he wrote the world's first web server and the world's first web browser. He then gave it away. The protocols that comprise the World Wide Web are known by some as the patent that never was, something given to humanity to use. And we all are quite aware of the results. He since has founded the World Wide Web consortium, which continues to develop the various protocols that underlie the World Wide Web and most recently broke ground last autumn on the World Wide Web Foundation, which is meant to take the web and see how it might be deployed for the benefit of humanity. So with that, I turn the floor over to Sir Tim Berners-Lee. Algorithms run my life. I have a horrible feeling that the World Economic Forum organizers expected us to come along and say, yes. My life is being run by algorithms to help. And in fact, the background which was going to put behind me was going with lots of zeros and ones reminiscent of a black and white version of the matrix. But I'm afraid I've come to tell you that actually on a good day, no algorithms do not run my life. My life may run algorithms. But life should stay on top. OK, life for me has got a lot to do with things like this. Things like this has got to do, there are algorithms going on. In fact, there are algorithms going on. Some of the algorithms which give me a lot of a great kick when you're coming down that ski trail and just as you come to a corner and suddenly this ground drops away and it opens out on your mind immediately thinks about how to edge those skis so that you can take the turn and end up in the right place to take the next one. In my head, there's lots of algorithms going on. And I get a kick out of doing those algorithms. That's why it goes quickly down a ski slope because it makes your mind work harder. So I think there are algorithms working in my head which I actually find really fun to run. There are other algorithms that I run in my head which I find really rather painful to run when I do my taxes. When I have to fill in forms, yeah, it has to think. When I do arithmetic, I can do it. I can make my brain do it, but it seems to hurt. But it was just funny. Why the difference? Why don't I get a kick from the skiing type kick when I go down when I fill in my tax form? What would I be if we did? Anyway, that's the way I'm built. So the algorithms running inside my brain, but most of the time, the algorithms we think of are the ones run by computers because they really are very stepwise. And the point about HTTP, for example, is that it gets out of the way. The point about the web is, unlike a lot of systems before it, when you look at the web, you don't actually look at the web browser. You don't see. When you click, you don't see the HTTP message going across over TCP in IP packets across the internet. You don't see it. The result coming back in HTML and being disassembled, you just see that you're moving from one piece of information to the other. And the idea is that, mentally, in my life, I'm interested in information. I'm interested in the thing it's about in people, in organizations, gadgets, in entries, and in the next ketra. So for me, it's important that when a computer is just an intermediate between me and the information, it should vanish. And really good computer software vanishes. It doesn't give the impression. When you ask somebody, they shouldn't say, when you say, what did you spend the last hour doing? You shouldn't say, oh, I was using Firefox. Ideally, somebody says, I was exploring Davos, or I was looking at what I was going to do next. I was working at the higher level. The algorithms are pushed down below. There are sometimes, I must say, though, that there are algorithms which need to be done. And there are some things which computers really have to do and should be able to do better in my life. You might occasionally have had to do this. Sometimes you might have had to arrange a meeting with a few people who are very busy. You might do this about 20 times a day in Davos. When you've arranged a meeting, that's the sort of thing, actually, that algorithm. Look at all the people. Look at all the times. Figure out for each person how difficult, how painful it would be for them to clear that space. And then add up the total pain. And then apportion it to the people to whom you think pain is really, you do not want to inflict pain. And move it to the people for this time you're not going to inflict pain. It's a very complicated algorithm. And what's interesting is I've never found a computer program which can do it very well. And I've found some which have got quite sophisticated. But they need to know, for example, how attached I am to each of my appointments. And none of my calendars allow me to say that. So if algorithms, I want to help be as well. So one of the things that I'm involved in is research to make the computers actually much more aware of what's going on. So for example, being aware just of things like which appointments are important to be, which could be moved, that's the step that suddenly allows a computer to run that algorithm. I spent 2009 getting governments, the world over, actually just because it's 2010 you can still do this, put your data onto the web. It's very important, very valuable for governments to put data on the web. Why is it important for governments to put data on the web? Because when somebody's trying to write some software, somebody's trying to create an algorithm, there's masses of data out there about where the hospitals are in Haiti, where the nearest post boxes or the post bus stop in Switzerland, where what the schedules are, how much money was spent on what, what the financial status of the different firms, all this stuff is data out there which is grist to the mill of an algorithm. If I want to write a program to help me make a decision, then it needs input. Some of the input I get from my calendar, some of it gets from things I share with my group, but there's a huge background information which for algorithms is this data. So no, algorithms should not run my life. My life should run the algorithms. The algorithms should be servants of my life and in order to do that, we need to move to a world in which we're much more connected. We're calling it the world of linked data. So one of the reasons that I want, so I'll put a plug in there for linked data because when we have a lot of linked data, the algorithms will be better, they'll be more slick and they will, and my life will run them and they will not run my life, but they will make my life smoother and easier and allow me to spend more time skiing. Thanks. Thank you so much, Tim. Tim talks about the way in which technology should serve humanity and the way in which ideally it would do so almost invisibly. It kind of fades into the background and you can just get to where you're going rather than worry about the window through which you see it. I worry a little bit about the way in which humanity can start to serve technology and the ways in which that very invisibility can mean we don't even know it. Now there are both good and bad aspects to this. Last year, a researcher at New York University named Casey Kinzer did a rather unusual project. She designed little cardboard robots, placed a smiley face on them, put a flag saying where the robot wanted to go in the first person, humanizing the robot, and then from an electronic store placed a simple motor in it that propelled it at a very slow but constant rate in only one direction. She then released these robots on the streets of Manhattan. The results were pretty interesting. No one called the police. Instead, here's an example, she put a robot up in the northeast corner of Washington Square Park and over 42 people intervened to steer it on its way and get it through the busy streets of New York to its destination in the southwest corner. Now there's no real algorithm for how to do this. This is depending on humans to come along and the humans are anonymous. We will never know who even they were in helping. This is a very interesting experiment. It's one of those experiments for which you're not sure what conclusion to draw but you find it very interesting. Here's now some other implementations that tend to mirror this. I tend to arrange them in a pyramid looking for ways that you might start to treat the human mind itself as a technology as if it's just another server you could add to your rack space if you have a problem you wanna have solved. And you might say there's some very difficult problems that require skill that only a few people can help with. That's at the narrow top of the pyramid and that might be very expensive. Then you move towards the bottom, tasks that almost anybody could do when you could draw a lot of people in for a little bit of money. Let me walk quickly through this pyramid and give you a sense of how it's starting to shape up. The very top, I might put something like the X Prize Foundation. They award a very big chunk of money, something like $10 million to the first entity that can come forward and say privately launch a sub-orderable cruiser that can go all the way into low earth orbit and back again, that will win a prize. Your prize for coming in second is $0. And it turns out that more money gets invested in winning the prize than the amount of the prize. A social benefit, not necessarily a benefit to the teams that don't come in first. Right under that, I might put something like Innocentive, an enterprise started by Eli Lilly, the pharmaceutical concern. This is a marketplace for difficult problems for which the person having the problem can simply assign a bounty to it, a miniature X Prize. So here's an example. There's a problem with juice. If you leave it not refrigerated and you leave it out like orange juice, it can sometimes turn brown. It's still edible, but people don't want to drink it. So $20,000 to the first person before October 1st of last year who can come up with some solution to browning in juice. And here's over 270 people who have stepped forward to engage and to try to solve the problem. Once they have solved it, all rights, intellectual property in the answer, revert to the person who commissioned the challenge and in exchange, $20,000 flow to the person who solved it. This is a way then of ramping up human skill without having to have your own employees or even a formal relationship with some contractor to do complicated work for you. Maybe one layer below that in the cloud, I'd put something like LiveOps. This is virtual call centers. One of the stories told by LiveOps is when Hurricane Katrina roared through New Orleans, the American Red Cross said, we have a hotline. Here's how people can find their missing relatives. The hotline was instantly swamped. So they went to LiveOps, which was able to spool up over 5,000 people staffing the hotline, identifying themselves as the Red Cross in under three hours. How did LiveOps do it? Not by having a big boiler room of people working in cubicles. Instead, people just in their own homes, mompreneurs, they call them. There might be a child in a bassinet just off the view here. And they can plug in after they've accredited themselves with LiveOps and they have their own menu of identities they can adopt. They might take pizza orders at a shop. Somebody calls what they think is the shop. It goes to this mompreneur who has a script and says, thank you, I'll take your order. Press Ascend, it goes back to the shop and the pizza is prepared. Maybe it's at a drive-through window of a fast food restaurant or maybe it's I'm a Red Cross person. Please tell me who you are and I'll help you through the hurricane. Very, very interesting and in some senses very freedom enhancing for somebody that might otherwise be simply at home caring for a child. Under that, I might put something like Somasource which takes very small tasks and actually parcels them out to refugees who in a camp through a mobile phone can do the tasks. What sorts of tasks might these be? For that we might look to a service called Mechanical Turk. I don't know how many people have heard of Mechanical Turk. It's named after this ancient device by which you could play a game of chess with a robot and amazingly the robot would play a decent game of chess. How did it work? You actually had a person hiding inside the robot manipulating the chess board thereby possibly creating the world's very first sweatshop. Mechanical Turk is now the name of a service from Amazon.com where people can put out what are called hits human intelligence tasks and you might pay a penny a hit. This is one penny in order to provide keywords for images so somebody can come up and just earn a penny for labeling this image. Now I know you cannot help yourself. Right now you're thinking box, box, right? You just, you see the boxes, you're like they're boxes and sure enough they're boxes, you label them and then you get a penny and you can't help yourself but go on to the next one and before you know it, the morning has gone by. Better than solitaire, right? Many, many people using Mechanical Turk are graduate students. I think it's because not just their low wages otherwise but because it's the only positive reinforcement they get during their careers. Here's another Mechanical Turk question. What is the difference between vanilla and French vanilla for three cents? You answer the question, you get three cents, you have no idea who asked it, what the answer will be used for but away you go and it's gotta be your own answer. Here's a meta task on Mechanical Turk. They asked people who are turkers for money to hold up a sign and take a picture of themselves saying why they turk, then they would get paid. So I turk for making money in my leisure time, I turk for Christmas, I turk to battle insomnia, I turk for drug money, just kidding. It's amazing to see some of the faces behind what otherwise is intentionally designed to become so algorithmic, it's just like assigning this out to a very smart computer. In fact, the subtitle of this by Amazon is artificial, artificial intelligence. What could be below the pyramid when you're paying people a penny a task? Why not pay them nothing? So for example, this is called the ESP game invented by a brilliant computer scientist named Louise Von Ann at Carnegie Mellon. He just plays a game, he offers a game to people where you see something and you're supposed to guess what it is. And when you match another player in a guess, you get points. What are the points good for? Absolutely nothing. But people love to accrue points. And in fact, what he found was that you could get many people playing over 20 hours a week labeling images in a reliable way since the only labels that work are the ones that match an unknown anonymous player. He actually says his advisor insisted that people coming from the .edu domain not be allowed to play more than 20 hours a week so that they will get back to work on their theses. So he just ran some numbers and said, 5,000 people playing simultaneously could label all images on Google in 30 days. And individual games in Yahoo and MSN average that all the time. Google took note of this and bought out the ESP game. It's now busily labeling images and doing other such tasks for Google. Here's another application called human computing for electronic design automation. The kind of tasks that involve cramming transistors as tightly as possible onto a chip. And you can have a computer do some of the work, but at some point the computer gets frustrated designing its successor and can't get any further. So what you do is you put it out and find a way to map this task to a game that people can play clicking on the periphery where when you click it changes the pattern of dots here. If you get all dots green, you win. What do you win? Nothing. What does it mean having one? You have found a way to design a better chip. And the chip then can be designed. Which means the makers of the game don't even know what sequence will light up all the green lights. Finally, another example, the United States Geological Survey has decided to use tweets as a basis for determining early pinpointing of earthquakes. They scan the entire Twitter corpus in real time looking for tweets that mention earthquake and thereby can triangulate where the earthquake might be happening. So let's just think for a second about is this a good thing? Is this a bad thing? How do we process the increasing use of humans in an algorithmic way? So let me just ask a hypothetical question. This is the United States Public Broadcasting Service. It's a site for kids, PBS Kids. Imagine your child has encountered this site. This is what it normally looks like. There are four games for the kid to play. Let's just swap out this game for that game. And you find that your kid has encountered the site and is happily clicking on yellow boxes for Saturday morning, designing new chips unbeknownst to your child. Any problems with that? If you are having, let's ask first, if you're totally fine with that. If this is like great, keeps them off the street, let me know. Interesting. We have one person, a few. If you're somehow worried about this in some way, you don't like the substitution, let me know. If you have no view on the matter and are delighted to have your kid doing anything, all right, very interesting. Is greedy? Yes, let's say more then. Fair enough. Here's another example. This is inoscentive. Suppose instead of browning of juice, they come up with a challenge. Here's a challenge. We want some pyrazolopyridinal diazenines. If you can get them, we can pay for them. You give away the pyrazolopyridinal diazenines, and you get your money. You have no idea why they want it, what they're going to do with it, what kind of chemical it is. This is the sort of thing that may give rise to some form of worry. Substitute your challenge here. And the anonymization, the mechanization, can make us a little worried. Here's another human algorithmicization. The state of Texas in the United States decided to set up webcams along the border and ask people on the internet at large to stare at them all day long. And if they see anything happening, they should click a button that says tell the police. And if enough people at the same frame click the tell the police button, they'll send the police. This site was overwhelmed by people wanting to stare at the border all day long. And in fact, now we see services coming up. So you know how on some vehicles, there's a telephone number on the back, how's my driving, which turns out to very much improve the driving of commercial vehicles. So here's something where they'll just put an accelerometer into it and a camera that films the vehicle as it goes. And if a curve happens too sharply, they'll take that segment of film, put it out to the masses for a fee, a small fee, and say, do you see something bad happening here? And then that can filter up to a supervisor if people see a problem. Much more careful scrutiny of drivers. Much more pressing in of algorithms on doing of a task. In Britain, they've come up with something called internet eyes. If you have a CCTV cam, which in Britain, chances are good you do, you can have not just your own employees watch it, but random people on the internet who can earn reward money by relaxing on their couch staring at security webcams all day long, have a chance at reducing crime, be a hero. This is their latest site. She's now off the couch, staring more intently and happily at the machine. And here is somebody able to enhance their security through this mediated system. The University of Colorado, once the students gathered for what they call a great smoke out on April 20th. They smoked something not entirely legal in the United States. They said, you can't arrest us all. The police said, you're right, but we can take pictures of you all. And then put them on the web for a bounty. The first person to say who this person is gets $50. Yeah, very interesting. Turns out the state of Iran, several years later, has done something nearly identical. This is, in fact, a real snap of protesters in recent elections, protests following the elections there, have been photographed. They don't know the authorities who they are, so they put out a call for families or anybody to identify them and send backward to the government. Now, there's a ceiling on public cooperation here, because the people most likely to know who these people are might not want to turn them in. But now, and here, let me be clear, I'm getting hypothetical. Suppose we just run some numbers and we put this out to mechanical Turk. We take the 72 million ID card photographs on file in Iran. We put that on the left and on the right, we put a photograph of a person we want to identify. And one at a time, we just ask, are the people the same? We ask this of anyone in the world who wants to earn a penny to process these. And by my calculations, approximately $17,906 later, you've managed to identify any arbitrary person within your borders using labor completely far away. So now, let me re-ask this question I asked before. Suppose your child spends Saturday morning and this is what she's doing, just comparing photographs. How many people are perfectly fine with that? How many people have anxieties about this? How many people are indifferent or waiting to hear more? Yeah. So I'm curious about the ways in which not just the participants themselves might end up doing things they would rather not be doing, but society itself can start to be altered. Because the cues we read day in and day out of what people like and don't like can be manipulated through these algorithms. Here's a mechanical turk task asking somebody to write a 5 out of 5 review for a product on a website. Write as if you own the product and are using it. Tell a story of why you bought it and how you are using it and thank the website for making you a great deal. This turned out to be for a piece of equipment, a router. And sure enough, a bunch of five star reviews appeared on amazon.com for the router. Word of this actually leaked to the tech website slash dot. And then a bunch of people ran over to the site and left one star reviews for the router, no one ever having bought the router. Take that example and now imagine deploying it to the real world. Not just rating systems online, but whenever you want to lobby a member of parliament or something, you could put out a call on something like live ops or mechanical turk and say, we'll pay you $0.05 for lobbying your member of Congress or parliament or showing up in person where you are and in fact, a constituent. Go to that town hall meeting. And if you say something approximating the following talking points we give you, come back, have it filmed, send us the film, we'll give you $5. Now, this was speculative. Last autumn, it is no longer speculative. In the American health care debate, one of the industry funded grassroots slash astroturf groups get healthreformright.org ended up using virtual currency in Farmville, the wildly popular Facebook game, to take Farmville players and ask them to lobby for the health care against the health care bill in exchange for this virtual Farmville currency. So why work all day on a virtual farm when you can have your own farm currency by just calling someone in that real world you don't worry about and trying to get them to do something? So when I think about the magic of this example and make no mistake, I look at this and I find this is positive. I'm still not sure I understand it, but it's positive. I also worry about the ways in which it turns out we can manipulate things. Now, everything has two sides to it. That Iran example I was talking to somebody about from an organization called The Extraordinaries, which invites people to do micro tasks for nonprofit work. And he took that example and turned it into a good thing where you could submit a photo of a missing person in the recent earthquake in Haiti and then others online could sort through disaster images and then do the kind of match I was talking about in my hypothetical. He deployed this last week and made approximately 20 matches that previously had not been made of people who had been missing during the earthquake. Last example, this is a rather odd Amazon mechanical Turk task. Do something kind and take a photo of it. If you do, you'll be paid 50 cents. And the commissioner of the task envisions this as what he calls a kind machine, which appears to be somebody being placed into this processor. The algorithms run in the gears turn and hearts fall out the bottom. I look at this and I say, that's great. There'll be more kindness in the world. You can write a check and get kindness. I also look at it and I think, the next time somebody does something nice for me, have I had an authentic connection with them? Or are they going back to collect their Farmville currency? Thank you. So we have 15 minutes, a little over 15 minutes left in the session. Tim, maybe we can both come up and we should just open the floor for questions, reactions, or everybody. Yes, sir? You talked a lot about humans being part of the algorithm. Can I turn it around with the algorithm running the life? What about people who use Facebook, Google? They make decisions on who their friends are, what they buy, how they live their lives, where they go on holiday. And it may not be their decision. It may be the decision of recommendation engines of algorithms that are written by very clever people at Facebook, at Google. How do you think that is changing how we live our lives? Well, I think that the interesting thing that when somebody tells Facebook who their friends is, well, it separates out the recommendation engine. In general, that sort of thing is great because it allows the system, as I said, there are lots of algorithms which can then happen. Then they can ask for all the photographs taken by their friends or taken of their friends or things like that. So the Facebook site gets more valuable to them because they give information like that. And so long as that information isn't abused, in principle, the whole system works. There are a number of snags. One snag is that each of the social networking sites is a silo. So in fact, the typical frustration, there was a great cartoon economist about this, typical frustration is you go into one site and you say who your friends are, and then you go into another site where you've uploaded your photos. And then you want to tell that site that you want to see photographs of your friends. But they already told Facebook who your friends are, and this site doesn't know who your friends are, so you either have to tell it again, or then you think that's great. Now I have this beautiful photograph of all my friends. I want to make a t-shirt of it and a t-shirt site. Similarly, they're all silos. So we've got to, so there is an issue there, that issue about who owns that data. You've given, it's data that you've put in, it's about you, it's capable of making your life better and it's owned by the social networking site. So I think there's an issue there. So I suppose your call for what you might call data portability, you put it into one gated community or something you'd like to be able to take it out so that you can have any number of recommendation engines or something operating on your data. Is that kind of a leaf on the tree of what you might call the semantic web? Where, I mean, tell us a little bit about that. Okay, so I'm talking about linked data. Linked data is really the simple view of the semantic web. The semantic web is a web of information. If you think about your social network, okay, it is, oh, it's a graph, it's a web, it's a net. The semantic web is about exposing that. So supposing there's some things in your LinkedIn profile that you'd be happy to everybody to know. You're happy to everybody know. Your CV and your colleagues and your history, well, now supposing that's published as linked data then rather than just people who sign up to LinkedIn, being able to get at it, then that information would be reused in all sorts of different ways. And in fact, lots of different algorithms on different sites would be able to run on it. So yeah, the link, so the technology for exposing this is the semantic web. And in fact, there are some social networking sites. There are things like LiveJournal, which do do that. So Advocato, My Opera, for example, if I can actually use a semantic web browser and browse across from friend to friend to friend across this social network graph. In fact, crossing over, unaware of the boundaries between different social networks. And that is very powerful, of course. Other questions, maybe where the microphones already are? Have they been deployed? Wherever they might be? Feel free to tell us who you are if you'd like to. Could we have the house slide so we can all see each other? Hi, my name is Sandra Schmunis and I have a general question on the privacy issue. If anyone has Googled themselves... If no one has, I'm sure. If anybody not able to. There is data there, dating back to some random events, some things that maybe you want to have an erase button on the internet. Now you were talking about the issue of I'm in a public place and a picture of me is taken and then somebody goes without my knowledge through licensed database and that's some edge up. So what rights as citizens do we have to public data that has basically been out there? What rights do we have as citizens to prevent or to the erase button of the internet? Boy, that's a great question. Were I a lawyer, which I technically am but let's presume I'm not, I would give you some answer about here are the rights that governments sometimes usually ineffectually try to vindicate for citizens in circumstances like this and generally they're most successful when it's regulating companies that might be collecting data for people or governments exercising self restraint under the rule of law about how long they might retain data or for what uses they might put it if they've collected it. But you raise a larger question of data that's just out in the wild and the fact that you might technically have some right if it casts you in a false light or infringes your copyright or something, it's not like that really helps you very much because it's so hard to find out who's behind the transmission of your photo and how to get them. So my hope is to be able to enjoy the fruits of exactly the kind of technologies and deployments that Tim talks about where you can put your stuff out there and maybe have a layer that you might think of as a humanizing layer on the network so that you could attach what we would call a piece of metadata to a photograph that you didn't put there but that implicates you in some way and use that metadata to simply ask others something about it to say I would prefer that this not be forwarded on again and again and again. It cast me in a very embarrassing light, here's the context, I didn't put it up myself, would you mind? And that at least raises with the people who so easily copy and paste a link forward something to their friends, it reminds them of the ethical dimension a very personal way of what they do and my hope would be that enough people would respect that, not all of course but enough that it would no longer be viral, it would not take off and they would forward cat pictures instead and it might just at least be worth it to ask yourself if you got that kind of plea about a photo you were about to send that link onward to a friend, would you respect it or not? And at least you'd have the opportunity to decide. Tim, I don't know if you have thoughts on this question. I think it's the culture of what people are prepared to do. Can we move it completely? There's a feeling at the moment in all the places that if I found it on the internet and it's a bit of you, you lose. So suppose we change that, suppose for example I get as a teenager, go on a crazy trip across the US and take lots of photos and take a GPS trail with me and put them all out there and so that basically you can reconstruct my trip completely, you can figure out where I was and if you're my friends, suppose I can put it out there and say if you're a friend of mine and you really want to understand what it was like for me driving across America, then go ahead. But if you're a prospective employer, I do not license you to use that information. And imagine that if we employ somebody at MIT, we have to be absolutely sure I can demonstrate to the MIT HR department that there was nobody, that I didn't discriminate on grounds of gender, I didn't discriminate on grounds of race or marital status and various things and maybe we could also put a check box in when you apply for a job at this university. This university respects your feelings about how you use information. So if you check this box, then we will look, we will pour, then you're inviting us to look into your social networking history. If you uncheck this box, we won't, just because we respect you, just because that's the sort of employer that we want to be. Some of us may have posted some crazy things. So yes, imagine moving to that world, but it really is a cultural move where people have to start promising to act in that way. And I think it's a very much better place in where we have a concept of acceptable use of information which is tagged onto that information. And it may sound naive, but the use of law to just try to force what you want has its own drawbacks and costs. And if you can get this cultural shift to take place, it turns out to be itself extremely powerful. Think that very few men at this conference or in this room are wearing skirts or a dress, right? It's not because there's a law against it, it's because there's a very strong social convention unless it's part of a national identity by which you wouldn't do it. Other questions? Microphone, yes, there sir. My name is Alfred Berkeley and I'm chairman of a group called XBRLUS. And we've been working on the data transparency issues for the SEC's reporting system in the United States. I wanted you to attempt to follow up on the state of the art in your semantic browser. Does such a thing, where are we towards getting there? Well, it's funny that we've developed the semantic web for a long time before we realized, before I realized that actually the semantic web browser was a really important part of the loop. On the web, everybody does it because they look at somebody's page, they copy it, they put their own photo in their own name and then they put stick it on the web and they say, look, mom, I got a web page and mom looks at it immediately with a web browser. And the data projects had been using all kinds of complicated stuff to analyze data and we didn't have a data browser. So what, three years ago, I started coding some stuff actually myself over Christmas, gave it to some students and we've been developing it on and off, it hasn't really been something we've funded in particular but we've got it, but there are now a bunch of browsers, the original thing was a thing called Tabulator, it's a Firefox plugin and you can actually browse through the semantic web and you can actually also extract, when you look at from that more ass, you can extract tabular information and then do with the sorts of things you can do with tabular information and it's just a prototype so you can play with it but there are a bunch of other semantic web browsers now there must be, who probably a dozen out there and I think we really are the early stages of being able to understanding what's a really good tool. XPRL is financial information, one of the things that you could do with Tabulator is build in what we call a special pane so that when somebody's looking at something which is actually is XPRL data, you can write a little, add in some little functions, some views if you like, which are particularly appropriate to financial data and so that you could then browse through somebody, say through somebody's personal information and their CV to the company they worked for and then just slip into the financial information about that company and get a novices view that I'd need of the financial data and then be able to maybe slip through that into an expert's view of the financial data so that you can imagine when you've got access to all this data connected together, it's really powerful and I think it's going to be really exciting because user interface is fun, interface is instant gratification so maybe we should work, get together a little team working to put the user interface people together with the XPRL stuff. Next microphone, where may it be? Yes, in front? Yes ma'am. Maybe this is well known but I would like to ask you about policing information on the internet and the use of information malevolently. Just a small example, my very young grandchildren were targeted by gang who took photographs of them and posted the photographs on Facebook with information that after which there was nothing they could do about it and surely normal laws of libel must apply but perhaps you could speak about that. Yes, they do, yeah, but in a situation. Normal laws of libel apply. Normal laws of fraud apply. The internet is not a separate country. Everybody who's sitting there typing, these people who are taking photographs if they're saying mean, nasty and untrue things about members of your family then they are committing crimes. Or malevolent use of images, I mean unauthorized use of images and so on. I'm not the lawyer either. Do you want to? Well, I want to just- It's a general question about policing and also when you don't know who's put the information out there, how do you sue? The attribution problem is a very difficult one and it's mirrored in cyber attacks. Somebody attacks a website, brings it down, don't know who's behind it. That can be very difficult just as it is with photos and other things where there's a personal cost. It turns out that for extremely intense and negative examples with enough law enforcement the bad doers had better be really good hackers or you can find them. You can send them an email that actually has a little web bug in it and when they click on the link in the email you get a little insight into where they are and it's how in many cases it turns out with the right law enforcement or private investigatory help for extreme cases you can turn out to find out who the people are and it might even be people in the school or otherwise with means, motive, opportunity. It then becomes not simple to solve but like another crime that requires some gumshoe work. Yes. It's not well known, it's true and it's part of just how young the web is really. Yes. I have a question going back to your pyramid. Yes. The question is, is there a historic dimension where because of technology and social development that actually we sort of start at the top of the pyramid and then we work our way down because what's the difference between for instance code breakers during the Second World War and people being given bits of chip design? So the first part of the question is is there a historical development from top to bottom down that pyramid because technology becomes cheaper and cheaper and if that's true then the second part is what happens next? What's the next layer down? If you go forward then that, if that's a process then we get to do even more micro tasks that are economically viable. So what does that look like? Do we have humans looking for extra, instead of computers looking for extraterrestrial life? It's funny, I thought your question was originally gonna point towards the fact that the pyramids themselves were built with a form of micro labor, it was very low paid. But, I thought you were going. And I think those were built from the bottom up, it would have been much harder to build them from the top down. But I think you're right that it may be easier to start at the top, to commission a prize and a bounty and there have been contests like that before, that doesn't require an internet to do. So it may be that much of the action, the things that the price, the transaction costs are made so much lower are towards the bottom of the pyramid where you can engage so cheaply somebody that it's worth it to pay a penny to do. What's under that? Doing it for free. And what's under that? I suppose just existing. Not even realizing that you're doing something but as you go about the world you leave a particular digital trail and that trail becomes fodder for all sorts of things. In each instance we might ask to what extent do we want to have autonomy over the mosaics that can get created over the trails we generate? And that's a very complicated question because our autonomy to control that then turns into limits for others who want to make very interesting hay out of that. I'm told we are at time so please join me in thanking Tim Berners-Lee and ourselves for a terrific panel. Thanks very much. And if you do need that to delete something I've got the button here, okay.