Who Owns the Future?

each year Microsoft Research helps hundreds of influential speakers from around the world including leading scientists renowned experts in technology book authors and leading academics and makes videos of these lectures freely available hmm thank you for coming my name is Kevin Koontz and I'm here to welcome Jaron Lanier to the Microsoft Research visiting speaker series Jason Jaron Lanier is one of the most provocative and creative thinkers thinkers of our time drawing on his expertise and experience across computer science music and digital media to challenge conventional notions about how technology is transforming society he is the best-selling author of you are not a gadget and he is well known for popularizing the term virtual reality in fact as the founder of vpl research incorporated years ago he was the first to sell virtual reality goggles and gloves and today's with Microsoft Research Jaron is here today to discuss his latest book who owns the future which has received a great deal of attention earning reviews interviews and commentary and dozens of media outlets in the US and overseas please join me in giving him a warm welcome hey hey how's it going so this is sort of weird for me cuz uh I kind of live a schizophrenic life where and I and that's my contract I'm sort of I'm schizophrenic by contract right I have one life as a so-called public intellectual where I write these controversial books and run around and blab and whatnot and then I work here doing research you know where I you know I complain about staffing allocations and and all this stuff that we all you know all that stuff and although I can't tell you what they are but bad we have some really cool results at least I'm very happy with my research here but anyway I'm here in my other persona nonetheless still in the building 99 that's a very weird experience for me like an alternate universe I feel like I've just stepped through some sort of portal so one of the things I do love when I give talks is I play music and that even that's kind of weird cuz I'm talking about the serious stuff of the future of economics but I sort of still do it as this sort of hippie artists person I don't know how this all happened but somehow it seems to work so I'll play music for you unless - weird are you interested okay so the the this is instrument I play a lot for audiences because I it's it's just kind of kick ass it's from Laos it's called a can haha haha Oh all right so that was music actually there's a cool thing about this instrument which is it might be the earliest digital number so this is a one of a family of Southeast Asian mouth organs this is from Laos but they're all kinds of variants of it and I believe it's the oldest human design of a set of objects in fixed positions that are similar that can be turned off or on combinatorially so this is 16 bit number and it's about 15,000 years old so this is it it's where it all started so we all got in trouble actually um so I'll tell you one version of history that gets us from this to where we are now in the ancient world these were traded on the Silk Route and the ancient Greeks and Romans knew about them the Romans made a giant version of this to accompany the gore in the Coliseum so it was sort of like the feature soundtrack of its day and it was called the hydrolyse and because they were Rome they it was steam-powered it was gigantic and there are actually some wrecked hydrolysis that survived so we can actually see them today and they're so big you can't just use your fingers to open and close the holes you have to use these planks that you open and closing those evolved into keyboards so the the hydrolysis of all've din to the medieval pipe organ but it also evolved into keyed string instruments very early on as well and that turned into of course the harpsichord in the piano but from the very beginning there were attempts to automate so even of the hydrolyse there were attempts to open and close multiple planks at once and build a higher level mechanism macros and so this this idea of building a bit of higher-level control in two-player instruments continued through the centuries and there was a non deterministic player piano that could so-called improvise a little bit that actually inspired a fellow named jacquard to make the jacquard programmable loom which in turn in turn inside at the Babbage programmable calculator which in turn inspired Turing and Vannoy Minh to formalize this field that tortures us all to this day so this is it this is the start first mover advantage right here okay let's see of concern in this talk is the question of how digital network architecture relates to you cannot economic and political outcomes in a society and has a prologue I will describe my personal experience of decades of waiting with great anticipation for the benefits that the availability of digital networking would bring to people I had been in like I'm sure many of you here in this room I've been involved in this game for a really long time and starting in the 70s when I was a teenager I'd been infused with this bug that someday but being able to share information collaborate on networks there would be this wave of improvement in wellbeing for people it would be analogous to the wave of improvement in well-being that resulted from electricity in the walls or plumbing in homes cutting cold running water or vaccines or decent fertilizers or you know the interstate highways these basic capabilities that made life better for large numbers of people once so we were now years into a period in which networking has become available and I think we see mixed results I think we do see benefits but what we don't see are economic benefits now here's an or let's say we see a kind of economic benefit that I think isn't sustainable so I was personally shocked by two sequences of events both of which defied my expectations one of them was just in the musical field so I play music professionally at soundtracks whatnot and in the 90s I had a career as a recording musician and assigned to a major label and I did pretty well at it and but it was during that time that I was deeply upset and disenchanted with the befuddlement and corruptions of the music business as it was and I was absolutely certain that if we went to a different model of sort of open source open culture and so forth where musicians shared their music that the benefits they would get would open up possibilities and that a whole new generation of musicians would cleverly invent new ways to have careers and there would be this wave of well-being oh I didn't realize I was being interpreted please tell me if I'm talking too fast okay or somebody I don't know indicate because I know I can sometimes go fast all right right so um so I was just sure that would happen and I actually made up a lot of the rhetoric that's become just an orthodoxy today I mean what I found is that if you question the open culture orthodoxy and the idea that information would be free and all you're just pounded with these arguments and the weird thing for me is I made up some of that stuff I was there like I I if you go back to some of my writing from the 90s I was really articulating a lot of the stuff that I get from kids these days kids these days and so it's one thing to be complaining about kids these days but it's nothing when they're parroting back the stuff that you said and like it's a very weird echo chamber II kind of surreal experience but anyway um what I saw was around the turn of the century musicians just started to do badly plain and simple you know and but there was a particular pattern that bothered me especially I mean I understand that with technological change and with economical evolution sometimes so they're going to be groups that are disadvantaged and I'm not expecting that everybody has some entitlement to always do well under every circumstance I understand that I'm a big boy I get it however what we were seeing was a disturbing pattern which was reminiscent of what we what were called Horatio Alger stories in the United States which stayed back to the 19th century a ratio Alger story is when there's a widespread luge in that people are doing well when they're not and a lot of people live on false hopes where the statistics are so against them that no matter how well they perform them no matter how much merit they present they actually don't have a shot but they're a token number of people who can do okay and it creates this false impression and if you have an economy that's built too much on false hope it'll fail and so that's the pattern I was seeing I was seeing tiny token numbers of people who had found a way to make do with a new system we had created the post Napster system and yet there was an illusion of a massive number of people who were succeeding but it was totally false and I put a tremendous amount of effort into trying to uncover every single example of somebody who is making it in the new system in music and I continued that and there really is almost nobody I mean statistically it's a total failure but there are token examples there are the Amanda Palmer's or whatever these people exist but they're just incredibly tiny numbers of them there's this tall thin tower and then there's this emaciated long tail all right so that's one thing that really bothered me and the result of that was a specific human cost where I saw people who'd had successful careers in the sort of middle of the music business not the Madonna's or something you know superstars but people who were like well-known jazz musicians suddenly needing benefits to pay for their operation or some problem and it was getting to the point we were having benefits once a week at jazz clubs to try to deal with the most difficult cases and I realized we're killing our musical culture like something has gone desperately wrong that really got to me but then in around 2007 and 8 the next thing that got to me was the nature of the recession that hit now look there are a lot of explanations for the recession yeah we we had an unfunded couple of wars that'll do it yeah there's the rise of China and India there's more competition for resource base yeah there's more older people than ever and more ways to spend money to keep them healthy blah blah blah but the whole developed world at once went into these bizarre debt crises in a similar particularly stupid way around bundled phony securities and that was really strange and that one really got to me because earlier in the 90s I'd had a role as a consultant to people who are trying to figure out how to apply what we now call cloud computing in big data to finance the terminology was different in an earlier phase but and that was and that stuff had worked out terribly there were a few experiments that just flopped awfully long-term capital was one anybody remember that one so that was this that was this experiment in trying to use big computation to sort of make a perfect financial scheme it was fronted by a bunch of people who'd won Nobel prizes in economics and seemed very legit until there was this humongous collapse and a huge public bailout that one was bad but it was kind of entered into with some innocence because I think the people sincerely didn't realize they were screwing up and I knew some of them I saw it firsthand I'm pretty sure that that's true so there might have been a technical failure there but I don't think there was a mal intent Enron was doing the same thing more or less but with mal-intent and then there was this huge collapse huge public bailout and I also need the folks at Enron and it was funny cuz I had a startup in those days and Enron wanted to buy it and I was like oh no Enron sis horrible ugly evil thing no no we have to sell to somebody better so we sold it to Google and now I'm out now I'll get to that love you Google people but anyway then with a seven and oh eight we saw the same pattern again exactly the same thing and I'll explain to you how I think these are all similar of course they're differences but I think they're strong similarities and what I realized is that this is not something where people are able to learn lessons there's a kind of a temptation in the way you can use computing to create fake financial schemes that just seems to be unassailable and what I realized at a certain point is that the failure of the music business and the ascent of fake finance we're actually two sides of the same coin so that's a story I want to explain to you so first of all let me give you a few MA models with which to think about how big financial schemes have become fake the metaphor I'm going to use is Maxwell's demon it who knows about Maxwell's demon here see it's great to be in a lab environment so it's a it's a it's a teaching tool that's used in introductory thermodynamics so I'm going to talk about physics don't run screaming I know it's a computer science lab but it's all based on physics ultimately it's a good thing to talk about so Maxwell's demon izl is a little imaginary guy he's a 19th century guy so he's eloquent and he speaks in long long sentences but what he does is he operates this little door it's a little tiny door and it's it separates two chambers and the chambers are filled with a fluid could be air or water perhaps and what what he does is he's looking at the molecules that come close to the door on either side and if there's a door if there's a molecule on the right that seems all jumpy and perturbed and that's a hot molecule and he opens the door to give it a chance to get through and if there's a nice languorous cold molecule on the other side he opens the door to give it a chance to go through and gradually he's selectively opening the store just flipping one little bit to separate these two chambers into hot and cold now that's an awfully valuable thing to be able to do because then you can open a bigger door and let them mix again and run a turbine generate some electricity then repeat the whole process and you have perpetual motion endless free energy right ok so what's wrong with this why don't we get free energy from this guy why can't we just build this the reason why is the act of discrimination the act of computation the act of even the smallest action is still real work there's no such thing as non work there's no such thing as purely abstract information and so what happens is the operating the measurement takes energy operating the door takes energy computing whether to open the door takes energy all of these things also radiate waste heat they're entropic and cumulatively it always costs more than you gain that's the cost of mutation that's why your computer gets hot - all right and so the interesting thing about this is that every possible perpetual motion machine somehow can be equated to this same the same no free lunch system of Maxwell's demon okay now what happens when you have a big computer with a lot of connectivity and you can get a lot of data into it on a network is even if you don't intend to you're tempted to try to turn into Maxwell's demon yourself but in economic sense and I saw this happen firsthand as a consultant especially in the 90s I had a weird consulting career in those days because there weren't that many people who understood big networks and so I got I worked with the early high frequency trading type of type schemes I worked with Walmart which was a really important early big computing operation all kinds of other examples so I saw what happens for instance one of my consultants consultancies at that point was with a it was actually the largest American healthcare company at the time and I saw directly how it was transformed prior to the existence of digital networks with a sort of an endless amount of freely gathered data and this huge amount of computation insurance was limited was computationally limited the way the way the schemes worked was entirely computational in fact the term computer used to refer to humans and almost exclusively women who were employed in these giant long buildings in upstate New York who'd sit there calculating actuarial tables you know that that's what computer used to mean before before Turing and and so they would have human computers calculating these things and the statisticians who are called actuaries had a limited amount of data and they could come up with very broad brush approaches to setting rates for insurance policies and that's how the business work but as soon as there started to be lots of data and really big computers and because of Moore's law when that stuff got really cheap a whole new picture emerged started to become thinkable to model individual people and place odds on them and you could do that not only based on the scientific theoretical knowledge that had been published by medical researchers but you could create your own correlations because you could gather your own data and you didn't even have to understand them it might just be that people who have purple wallpaper are more likely to have a stroke or some bizarre thing like that maybe that isn't bizarre I don't know but but you you wouldn't have to understand the correlations you would just compute them and then what you do is you pretend to be Maxwell's demon you say I'm going to open the little door and all the people who are likely to need my insurance are going to be excluded and all the people who are likely to not need to use my insurance policy will be included so you attempt to make the perfect perpetual motion insurance plan where you take as little risk as possible so you're performing you've turned yourself into Maxwell's demon now there's a little story I tell which is true and because I guess they're one of my consulting things I was with the a bunch of people from this health insurance company and the CEO of it was sort of taken with this observation that this new world of that he could become Maxwell's demon although of course that wasn't the way he was talking about it and he said you know what I can do now is I can I can get rid of that guy who's going to have a heart attack years in advance I don't have to insure them anymore and right at that moment and I remember thinking with her oh my god that's not what computing supposed to be for something's gone terribly wrong here that's not what we've all worked so hard for and at that very moment there was this huge swooshing sound and then there was like this earthquake in this explosion room and it turns out there was a meteor strike and right by us and I won't tell you exactly where it was but it was on us on Pine Island so it wasn't far so you can figure it out if you really want to be diligent but so anyway so what this leads me to is if any of you are astronomy researchers and you're interested in meteors what you can do is you can use health industry executives as bait and because otherwise I mean who else gets to yeah I was amazed to be near a meteor strike and I just from them I didn't get to close to the guy you know so it was like yeah I hear you so so this this sort of Maxwell's demon fallacy really breaks for the very simple reason that the overall economy the overall world isn't big enough to absorb all the risk that you're avoiding by trying to have a perfect scheme you have to offload it into the world and there isn't some infinitely big economy that can keep on absorbing your risk and can keep on providing you with more and more benefits so the scheme has to break and that's when I realized that there was a unifying paradigm in all these different failures that what was going on with health care in America was driven by a certain model of computation that wasn't sustainable but that in a deep level on this Maxwell's demon level it was profoundly similar to what I was seeing in finance over and over again it was similar to long term capital it was similar to Enron it was similar to the recent recession and all the the bundle derivatives and it's similar to what's going to happen with student debt and high frequency trading and carbon credits and anything else for somebody's trying to compute a perfect position trying to compute perfection doesn't work in reality I mean doesn't even working like there's a funny thing where sometimes I talk to people there saying no with the computer you can make this perfect pristine thing because in computers are perfect when I'm thinking wait have you developed software do you know anything anyway there's a whole question about whether it's even realism from a computer science perspective but at least from a physics perspective it's profoundly not realistic and also from an economics perspective so the thing about these schemes is that they appear again and again and again with different surface colorations different terminology different semantics but this idea of trying to calculate the perfect position comes up again and again so I what I've realized is that schemes like Facebook and Google have strong similarities as do recent large elections that are highly computational as do the new face of national security organizations all around the globe as do new criminal organizations basically what's been happened is wherever you find the greatest centers of power and clout that have been strengthened and improved since networking you crack them open you'll find a big computer in the middle running a fake Maxwell's demon scheme so I call these siren servers and it comes from the ancient Greek from Homer the siren is this uh is this dangerous creature who doesn't directly attack you or try to eat you but just confuses you so you fall and drown of your own doing and that's how I look at these things sirens servers are only a problem if we allow ourselves to be idiotic they're not like some alien force or some intelligence that's screwing us up but it's a kind of a temptation it's almost like a drug because as soon as you can do as soon as you have this illusion that you can compete your way to a perfect financial scheme at first it works that's the problem it's like an addiction feels great at first then you pay in the long term right so it has this drug like quality to it in order for the scheme to work the information that feeds the algorithms has to be free otherwise it would cost money to try to be a sort of siren server so this whole information wants to be free stuff which I'd been so actively promoting in the 80s and 90s turns out to actually feed this beast it turns out to actually be the cocaine that the Maxwell's demon want to be runs on and that's where the idea fails so another way to put this is if you have a bunch of people in some sort of a an attempt to create a utopia let's say and they're all sharing information and they're all in a network the ones of them that have the most effective computers the biggest computers the most highly connected computers that have been able to hire the most clever recent PhDs from Cal Tech and Stanford or whatever you know and you dub of course and anyway whoever's got the most effective computer can make use of that same openly shared information to such greater benefit than other people that differential becomes so big that it's actually not sustainable so what we've seen since the advent of widely available cheap networking is not this sort of a strengthening of a broad range of people in the way that we saw with the availability of electricity and drinking you know water hot and cold potable water and all these things instead it's created benefits almost exclusively in the most concentrated elite people which includes many people in this room I certainly feel part of it you know with what the Occupy movement calls the 1% if you want to use the language of the left but you know you have this idea of a recovery after this recession that is almost exclusively benefiting a very tiny part of society and you have a loss of social mobility and a lessening of the middle class across the whole developed world at once which is just astounding all right so to talk about this I want to give it a historical framework and I'm going to go back to the 19th century and this has to do with how we think about people in a world of technological change so the 19th century was strongly characterized by nervous futurism in a way they worried more about the future than we do today we've kind of we don't really talk about the future as much now as people did in either the 20th or of the 19th century the 19th century was all about machine anxiety I'll give you some of the highlights of machine anxiety in the 19th century we can start with the Luddite riots early in the in the 19th century these were textile workers who were concerned that improved looms would put them out of work they rioted and they were executed in public in order for order to be restored it was a very ugly difficult scene so we use the term Luddite today to mean somebody doesn't have the latest phone or something but it started out really as the birth of the modern labor movement other sign posts in the 19th century our early Marx starting in 1840s I also like to tell this story owes I was driving in Silicon Valley one time and I heard somebody on the radio talking about how this new scheme they were promoting was going to allow productivity to cross international borders with extreme efficiency and I was thinking oh it's another one of these stupid startup companies I can't listen to more of this crap I hear all this I hear this all day long and just as I was turning it off it said I realized it was the lefty station KPFA and that it was a anniversary reading of dust copy towel and it just turns out there's passages and marks that read incredibly current you know and I loathe marks as a proposer of solutions I you know heat marks had this idea that he was smart enough to know in advance what the perfect society would be and how to get there and that's a very dangerous kind of anti scientific thinking because you can only do science empirically but he thought he could have perfect foreknowledge so I'm not advocating marks at all I think has been a disaster but as it as an observer of his times he was really extraordinary and as a tech writer he was really good he might be the best tech writer we've had actually he might be better than McLuhan who's just amazing anyway what are some of the songs do you know from the 19th century one of them is if you're American anyway one of them is probably the Ballad of John Henry and this was about a guy who is in a race to lay down railroad track with a robot that can do it and he wins but only to drop dead from exhaustion this is a really popular song and then another familiar element that's with us to this day is science fiction science fiction the genre was started to explore the anxiety that people could become obsolete because of our own creations so we can go back to Mary Shelley and Frankenstein if we want but in the in the late 19th century we have just sterling examples from HG Wells with the time machine for instance in the time machine humanity splits into two species the rich ones are the descendants of the people who owned social networking servers and whatnot the other ones are the people who use them and the rich ones farm and eat the poor ones and there they're all miserable science fiction is always about whether people are going to become obsolete there's two kinds of science fiction we're either made obsolete because of our own machines or because of aliens but the Machine our own machines are the much more common element of obsolescence and so some of the recent ones are the matrix and Terminator movies and inception and Battlestar Galactica and it's on on and on and on so and that was that was borne out of the labor movement that's a remarkable thing you can read a cross over like if you look especially at Mark Twain's early writing there's this amazing thing where theoretical ideas about machines putting people out of work turn into science fiction stories you can see the labor moving movement morphing into early science fiction so that's actually its origin and that's what guides so much or imagery about tech to this day so here's an interesting question in the twentieth century we did not see people put we did not speak see ultra widespread unemployment because of new machines instead we saw better jobs why'd that happen well I think it happened because the labor movement triumphed on the one hand and on the other hand industrialists realized that they have to think about their own interests and that there was actually a completely unacknowledged commonality between the two so on the industrialist side so Henry Ford was a racist bastard let's just be clear about that his own descendants will say it more clearly than anyone else and yet he was a successful entrepreneur and one of the things he said is that it's crucial that he be able to price his car so that his own factory workers could afford to buy them because you can't have a market without customers you know it's like so simple so if wealth is too concentrated you can't have a market so if you want to grow your business you have to grow the market tada all right it's I mean it's not this is not rocket science this is actually a pretty simple idea a basic basic entrepreneurship then from the labor movement side they faced they faced a really tough struggle now there there's been a lot written about the labor movement obviously I'm going to talk about it in a way that it's usually not talked about from a techy perspective so from a techie perspective here's an interesting question on an example of a technology that used to support is this huge industry that then went away with buggy whips right that's that's a cliche what was talked about oh you know whatever it is is going to go the way of the buggy whip all right so the transition from dealing with horses to dealing with motorised vehicles is really a big deal and I don't know how many of you have dealt with horses but I I have dealt with horses and horses our heart our work you know they're actually really hard to deal with and if you love horses and if you have some really interesting you know sympathetic horses that's one thing but to have to deal with them all day long even the ones that aren't so nice and you're dealing with feeding them and dealing with their hoofs and brushing them and then the poop the poop my God all that and then you move from that to a motorized vehicle and it's like yeah it's easier it's like way easier in fact motorized vehicles are fun to drive a lot of people in this room have probably bought a nicer car than they really need because driving is actually really cool we like our cars they're just great toys we enjoy them it's really fun to ride a well-engineered car so this brings up a really interesting question if we had we have to pay people to deal with the horses because who would do that if they're not getting paid it's miserable but why the hell are we paying somebody to drive a cab or a truck because driving is fun like why should those people be paid and so if you ever meet a teamster and you wonder like why is the Teamsters Union so tough and brash and kind it's because they had to fight like crazy for the idea that even if life gets less miserable less smelly and less dangerous you still ought to be paid so the idea so better technology can be associated with better jobs rather than fewer jobs so long as you decide that it's still ok to pay somebody even if they're not risking their life and if they're not miserable and covered in crap all day long alright that was this huge huge huge transition and it took decades to fight for it now one of the interesting features of that realization is that to answer to say that people really should be paid requires the creation of some somewhat artificial ratchet system to give people a little bit of a license or something to get paid for the job so that you don't have race to the bottom and it becomes unpaid again so for instance union membership taxi medallions academic tenure these are all mekin Zim's tenure actually goes back to the Middle Ages but it served as part of this movement in the 20th century to create ratchet systems where people could achieve a kind of a status where they were paid for something that wasn't that wasn't actually miserable and life-threatening okay now we come to the 21st century the 21st century we have rejected that old covenant and the rejection happened I think in a lot of different ways in different places at once always surround always in this connection with the the fake the fake perfect scheme I was talking about always in connection with a Maxwell's demon but I think the first person really articulated in public was Sergei who I really liked from Google but at any rate the way the idea went was okay maybe you can get paid to drive a truck but just to do stuff online I mean give me a break information you don't get paid for that that's too easy so you know whatever work you do online if it's like sharing your music yeah just put it out there for publicity and so now we enter into this new scheme where we're saying if things get if technology gets advanced enough that it can be delivered as a software service then we stop paying people you know then we start to say the benefits you get are going to be what we call informal benefits instead of formal benefits and so this is a key idea if you talk to people interested in development in the developing world one of the key well the key quest is to get people out of an informal economy into a formal one informal economies can give you bargains they give you barter they give you reputation they give you all these things but the problem with an informal economy is it's real time what that means is you have to sing for your supper for every single meal so for instance if we tell musicians you can't get royalties on your music anymore but you can still you can still play live gigs and you can call it in you the problem with that is that then you have to play a live gig constantly what if you get sick what if you want to raise kids what if you want to take care of aging parents you can't be a biological entity anymore you're always right on the edge of failure and that's exactly what's happened with people we're living that way a real time economic career based on informal benefits is a career of insecurity and all it takes is one little string of bad luck which will always come along just because of how randomness clumps you know it'll always come along and at that point you're knocked off so it works great if you're an immortal you know perfect robot not a human and especially if you're an immortal perfect robot who can live with rich parents who still want to support you then it works great which is of course that everybody wants to be but we none of us can be so so the if there were only going to be a limited number of people who would be disenfranchised by making information free that would be absorbable we could figure out a way to compensate for that so right now the kinds of people who tended to be forced into real time economic careers by the open culture idea are the journalists musicians photographers those kinds of people we could come up with institutions to compensate for instance their various attempts to create new institutions to support investigative reporting because we don't have nearly enough investigative reporting for our times I think that that statement shouldn't require justification but the problem is it doesn't stop there the problem is that it it covers everything in the economy except siren servers eventually so let's look at some of the upcoming waves that are going to become I call it software mediated that it's hard to come up with just the right terminology for this stuff 3d printers are great example if you want it if you're a member of MSR and you want a 3d print something just go talk to the guys in the hardware lab across the hmm and they'll print out something for you and it's fun it's great I love 3d printing it's still early for those of you who haven't used a 3d printer it's like this box that looks kind of like a microwave oven or something you download a file from the internet just as if you were downloading music from one of the end from a BitTorrent site or something you get your file you and then these little nozzles follow instructions in the file and deposit materials a little bit at a time until your object is printed out today we mostly print out objects in a limited number of materials and colors and you don't print out anything but you know I mean you don't print out everything you might want but like in 10 years and 20 years I imagine we'll be able to print out new phones and tablets and things like that all the components of them are sort of printed already to some degree I think I think we can do it so what that means is a complete transformation of manufacturing because now suddenly you can enjoy the efficiency of printing out things on an as-needed basis and on a where needed basis you stop transporting goods around you stop having factories instead you have this distributed system all you distributor the antecedent groups however recycling becomes vastly more efficient and precise than it ever was before because you have a precise record of how everything that was printed was printed so you can unravel it with great precision because the information is lost so instead of recycling being a gross process it becomes a fine process so you'll be able to recycle those antecedent groups so you suddenly have this amazing green effect this amazing efficiency screws China royally because you have to then tell them oh you know all that huge manufacturing infrastructure in southern China you know Foxconn oh yeah you don't need them Microsoft's making a big investment in that stuff you know all right so so but obviously you know as much as the manufacturing sector is declined in the US it's still a big part of our economy even and it's huge in China and other parts of the world and all of a sudden that goes away now it's actually not going to be all of a sudden you know it'll come on with some slowness but you know about how Moore's law works it accelerates you know so the you know if if one year suddenly you can print a new phone then a few years later you'll be printing new medical devices and a few years after that you'll be printing everything including the printers by the way so they spread virally at some point they're not it's not like there's some store where you go buy your printer so what happens then retail goes away manufacturing goes away I know I'm exaggerating it won't be that clean it's always messy there's always exceptions there's always gacho's all that stuff but just in the broad picture there's obviously a huge problem here because what's happening is then we're Napster rising the fabrication of physical stuff we're Napster izing material culture and then do I need to list many other examples in a lab like this there have already been effective demonstrations of automated pharmacists legal researchers bio bench researchers all kinds of educated middle level jobs can already be automated I'm pretty soon we'll be automating our CS interns and maybe we can automate our managers and but anyway so that so the thing is that this wave spreads it doesn't just stop with the creative type people as the 21st century progresses it hits every part of the economy those Teamsters who managed to survive the obsolescence of the buggy whip and drive trucks are going to then face the new challenge of the self-driving truck and that one will surely knock them out so let's look though at how automation really works now when I was a kid there was a guy who was the sweetest most generous mentor to me when I was a very uncompetitive named Marvin Minsky who was one of the founders of the artificial intelligence movement now in 1958 a couple years before I was born Marvin had given some of his grad students an assignment to over the summer write a translation system from one language to another now that might sound crazy to us today but who nobody knew at the time I mean it was a perfectly reasonable thing to hypothesize about why not so hypothetically in those days based on how people understood language then it should have been possible to take dictionaries for the languages and write some sort of parser translation scheme and come out with a translator right now of course as we all know it doesn't work that way the only way to translate between languages effectively is with a big data strategy so we have these huge corpora of that we get of previously translated of passages and the so this so and it works it's great that it works and we were in a race with our colleagues at Google and elsewhere to make better and better language translators what we're all doing basically the same thing which is gathering at huge antecedent examples and then performing statistics to create new examples now let's notice something critical about this which is that there was a there were a group of real humans who translated passages in order to generate the examples that we use in order to create the so-called automation so it's kind of stage magic what we're doing is we're mashing up the efforts of real humans in a new and useful way but it doesn't mean that those people don't exist they do exist nor can you say that you only have to gather data from them once and then never again because language is dynamic so all of us are constantly scraping the net for new examples of translations to keep our ability to translate current and dynamic right ok so this is this is a key point there's a kind of um a figure-ground flip or sort of a gestalt transformation that can come into application here and I know I see this differently than many of my colleagues but this is how I see it any time you show me something that's automated or something that's called AI there's a way to flip it and see exactly the same phenomenon in different terms where humans did all the work it always traces back to humans there's not some alien species that's sending down data to us so far as we know anyway I mean some of the stuff you found online I wonder but you know at least the useful data is all tracked back to real humans now that raises an extremely interesting point to me which is if we were to achieve that figure-ground flip and instead of thinking about AI instead of thinking about automation instead if we were thinking of the whole system is being run by real people from whom the data comes but just having the mediation become more and more useful if that's the way we think about technology which is absolutely as valid as the usual ways then there's a possibility of thinking about an economic solution that gets around the siren server problem that provides a way for people to lift themselves out of the idiocy of trying to become Maxwell's demon now to explain that alternative I have to go back to the very origin of the idea of networking so the first person to write about how people could use digital networks to communicate with one another to collaborate actually predates the ability to implement a network because it happened before packet switching was invented and that was Ted Nelson's work starting in 1960 so Ted Nelson is still with us he lives in Sausalito on a houseboat he's a buddy of mine he's in his 70s now and he's not the easiest figure to understand in some ways he's kind of a beatnik hippie sort of person and his early writing was infused with a kind of psychedelic glow her countercultural zest to it that might not be to everyone's liking and is not necessarily as clear for many people as it might be and and that has to be said nonetheless starting in 1960 Ted was the first person to describe people using digital networks to collaborate it was brand-new I'm not aware of anything earlier and he did so with extraordinary insight I think sometimes the first person on the scene can see more clearly than people who show up when it's already cluttered so what Ted rila and and what he called it was hypertext which is where the HTM HTML comes from so there's a direct descent of his original terminology to what we use today so Ted had Hollywood parents who benefited from the labor movements of creative people so we usually think of Hollywood as being you know populated by super overpaid actors who just grunt while they fire weapons or something and become the Governor of California or whatever it might be but actually the unions for actors and and whatnot benefit a middle class of people and his parents benefited from that so he understood that even if while you're doing is pure information you're vulnerable to erase to the bottom where you're demoted into an informal real time life unless there's some kind of a mechanism but what he realized is that instead of these artificial sort of ratcheting mechanisms like unions maybe something more organic could come about in a digital network and what he proposed is a universal micro payment system remember this is before this was invented Universal micro payments were invented before packet switching does a remarkable thing they're the actual origin point for networking so he proposed a universal micro payment system so that when people make use of information that exists because the other person exists that other person receives a micro payment for it so the people whose translations proved particularly useful to translation algorithm would keep on getting little dribs of pennies the people who if you write code whenever your particular line of code executes you might get a little driven drab of money out of that and it's a really interesting idea which hasn't been adequately explored for instance let's look at code we tend to think of the caught economics of code as being a war between two camps one of which is the open source world the Linux people and everything and the other one is us at Microsoft who are supposed to be the evil empire but the thing is there's this third way that has not really been tested that might be better than either of those if there's a micro payment system that's activated as your code runs and the more your code runs the better you do and the way I put it in the book is Sergey and Larry could have become really really rich just from a system like that without having to build a private spa Empire you know that you know that but the other thing is if you look at the Linux stack and if you look at the number of people who've contributed to it or the number of people who've contributed to something like the Wikipedia if that stuff was monetized you'd see a middle-class distribution coming out of it so the intriguing possibility here is that a universal micro payment system might actually generate a sustainable middle class even if technology gets really good and what we call automation becomes really advanced without the need for special systems that are inevitably very difficult and sometimes corrupt and awkward like unions and medallions and licenses and all this stuff now let me so this is a big idea I'm not certain it would work I'm not proposing to be like Marx and to know in advance what the perfect world would be and how it'll happen rather what I'm proposing is a line of research to see how it can work so now you know now I'm not sure whether I'm giving a book talk or research talk in MSR I actually I'm doing work to model this at down at SPC at our campus in this summer I'm trying to build agent based models of economies and trying to do monetize networks with them to see what kind of distribution of outcomes we get I'll give you a few basic ideas about how this kind of research works if you look at a spoken-- hub style network where everybody goes through a central arbiter and an example of that is YouTube or the Apple Store then the outcome of sort of winners and losers is a very stark power curve so that's where you get just a few big winners or Kickstarter's another one like that you get a few big winners and then you have this huge longtail of wannabes and the neck is pretty thin in between them and that's when you get the horatio alger effect where people think they have better chances and they really do and it's not sustainable now on the other hand if you look at a thickly connected network where people are interacting with each other and there's not as and there's not a central arbiter allowing only one person to get through at a time and I mentioned the Linux community's like that and the Wikipedia's like that or another example is Facebook where anybody can connect with anybody and people can get compound products out that have been contributed to you by many people then the variety of people who source who are the source of information that people see takes on a completely different character instead of this deep power law you start to see something that looks like a bell curve so the average person on Facebook actually is exposed to a wide variety of people not just a tiny number of stars and the average piece of code in the Linux stack was you know involves contributions from a large number of people two stars that's not to say that there aren't stars it's just to say that there's a distribution that has a big hump in the middle there's still stars this is not a world in which there are no elites and everybody's the same this isn't some socialist utopia it's just a world where there's a bell curve instead of a power curve all right so why do we care about bell curves okay so I already mentioned before that if what if what you like is market dynamics if you think capitalism has any value you have to realize it won't work if there aren't customers right so that's what Henry Ford realized you have to have a strong middle class you can't have a market it's just really that simple you can't have a market if you have some sort of petrol monarchy or oligarchy or something that's fake all right but then if what you care about instead is societal dynamics or democracy if you're sort of coming more from the left and you don't like market so much you still need a middle class because if income becomes too concentrated then politics becomes corrupt which i think is actually an issue in the u.s. right now so the point is you can abstract away whatever ideology you have it depends on a strong middle class I don't care if you're libertarian left or right you need it and so what we really should be asking is how can we design network structures so that economically we're generating middle-class distributions now the term middle-class can be problematic now maybe not in this audience I don't know but a lot of times I'm talking to the sort of literary crowd and if you say middle-class what they think is the visual it's our parents it's everything that's not cool and beautiful and hip fine a big middle-income block in the middle a bell curve it doesn't matter if you want to call it the middle class especially in Europe that's a really hot button let me tell you as I learned the hard way it's like you want to promote the middle class are you talking you want is it I'll leave it to beaver now is that the idea you know I think no no no anyway so what what I think the crucial thing we have to understand is how can we design a network that yields a middle-class outcome from information sharing that's sustainable because if we can get to that point then the 21st century can answer the fears of the 19th century but in a way that's even better than what the 20th century did if we keep on doing what we're doing of siren servers and fake fake Maxwell's demons we're just going to keep on having one collapse after another with one public bailout after another with more and more concentration of wealth and power less and less social mobility it's just the pattern will go on forever you know and obviously we can't keep on doing that I have this sense of how long we have which is twenty or thirty years I say that because I think that's about how long the intense technologies of automation will take to really get out there and get cheap you know so that's my sense of how long we have so if we do the research now if we approach it honestly if we're not ideological but simply trying to be problem solvers I think we have time to fix it I feel confident that we can I I want to address one other point that I often hear about just to preempt a question that I always get the question goes like this isn't it true that there's only a tiny number of people who are really doing any valuable thinking or really creative and what most people be useless in the system multi just recreate some sort of elite distribution and I just have to say maybe let's be empiricists but don't there's a kind of a weird stealth elitism that creeps in that that assumes a priority that that would be the case and empirically in those cases where we have data I don't think it is the case I mentioned Facebook as one example where we see a broad middle in terms of who's exposed to who rather than a star system that we see in hub-and-spoke networks so we've already seen that network topology changes that and if it were really true that most people were only interested in a few stars we wouldn't see that now another objection I often get is oh my god how can you be talking about Facebook that's such such fluff you can't monetize that don't encourage them you know and here's what I want to say about that our job is not to judge each other I'm not like some cultural critic who's good personally I'm not on Facebook I find it to be fluffy and useless but you know what that's just me like it doesn't matter what I think about it who cares the point is entertainments always like that you show me entertainment of any era in history in any location in the world and I'll show you some part of it that just seems stupid and pointless because there's always something like that people are different that's good that gives us that broad distribution that gives us those bell-curve outcomes right so if you want to get a sense of how much value is already being denied to people by siren servers you can start to in your own life keep a tally of the differential between what you'd spend if you agreed to join into somebody's computational scheme versus if you didn't so for instance if you have a shopping cart at Safeway or another store keep track of what the differential is for a year your Facebook activity on average is worth about a hundred bucks if we're to believe the valuation so that's maybe a hundred bucks it's not a lot but it adds up look at the difference between keeping track of your frequent flyer miles and not if you think you're really getting bargains from these things of course you're just that's a magic act there's no such thing as a bargain that doesn't exist it's just a price so if somebody says oh this is the bargain price it just means that they would otherwise be overcharging there's no so you have to get out from under stupid marketing tricks and especially if you work at Microsoft I mean we do them too when we sell stuff I mean get wise never be the sucker to always be the snooker in a market economy okay you know general principle of survival so if you start counting up all that stuff you'll find that for a lot of people it's already well up into the thousands and even the ten thousands and automation has barely begun so as this progresses the and people will be specialized like there might be one person who's a star on Facebook and another person who's valuable in some other way maybe as a 3d print object designer right the point they'll be special it'll be all over the place but on average I believe we already have empirical indications that there will be enough value there to create a persistent middle class not out of charity not out of entitlement not out of revolution not out of some kind of Proclamation not out of Luddite riots in the streets but simply out of honest accounting you show me AI and I'll show you accounting fraud if I want to put it really harshly and it's true you know it's like that's the flip I'm talking about so there's a great deal more that can be said about this of course wow I mean it goes on and on that's why there's a whole book about it but this is basically what I'm up to these days as far as the economics work the book is designed for popular audience and has all kinds of stories about other things it's I hope it's fun to read but that's the core that's the core idea I think the key question to ask about doing well in a market economy is are you succeeding through growing the market or through shrinking the market among Silicon Valley venture capital firms now it's very popular to say we like funding schemes that shrink markets so for instance a kodak is bankrupt by the way kodak guess what Kodak did kodak grew up in the same community and with the same workers or the descendants of the workers who'd have the biggest buggy whip manufacturer so that morphed into kodak and now kodox bankrupt and the company that's performing approximately the duties that kodak used to which is letting you take family pictures with interesting colors and share them with people is instagram instagram sold for a billion dollars with 13 people kodak supported hundreds of thousands of people with solid middle-class jobs with benefits and security so that difference is the difference that computation has wrought now the thing about it is that i don't begrudge those 13 people i love success I love Silicon Valley success I enjoy Silicon Valley I enjoy startups I've done a bunch of them so I don't have any problem with it the point is that when we find success we should find success by expanding the market expanding the economy and you expand the economy by monetizing more value that's what expansion is in economic terms when you monetize less value in order to concentrate it for yourself you're actually shrinking the economy to concentrate your own I'm absolutely convinced that if we got to a monetize scheme this is not some leftist project or an anti corporate project instead I'm convinced the Facebook's and they're well now it's a part of Facebook with the instagrams the the Microsoft's the Google I believe we'd all actually do better because we'd be part of an expanding economy as tech improves instead of a shrinking one because to shrink one under the the ideology of automation is to pretend that people aren't there which means that we're pretending that the value isn't there which means that the economy has to be smaller so it's the wrong height where I want us to grow rich I want us to be successful but we're doing it in a wrong way and the reason it's wrong is that it's not sustainable we're swallowing our own futures just for short-term gain all right that's it so we have about 10 minutes questions before we need to do book signing with in 15 minutes so okay questions and doing just business feature but some would argue that's the I'm trying to say it again you didn't talk about that support business Oh ad supported business models yeah well okay let's try didn't on whatever it has a question he's is he saying I didn't talk about the ad supported business model you told you mean like Google and Facebook ads so that's true I didn't talk about that so um the term advertising has been repurposed recently advertising used to be an active communication it used to be your oh man today ssin of a product I've acted in the ad I've been a professional in the advertising business because I did jingles for commercials for many years and I do a lot of work now actually supporting Microsoft advertising but that's another story so I have no problem with the advertising business as it's always been sometimes I have a problem in the book I describe how I found myself suddenly annoyed by this annoying radio jingle for in furniture store and realize it was actually my own jingle that was so sometimes of course I'm annoyed but but the thing is what happened with Google is a redefinition of the term advertising to mean micromanagement of the options in front of people so the problem is you can't search through a million links so you really can only look at the ones that are most immediately accessible and by manipulating which ones are accessible you manipulate people and if you have a behavioral model of those people based on big data then you can make those that manipulation be more successful now I know that the way we commonly put it is that that's when when because then you're getting the links that are most useful for you blah blah blah but then I ask why aren't you getting those links anyway likewise why like if if Google or Bing are doing their job there shouldn't be a lot of room for extra paid links because they should already be getting you these funds I mean it's that's sort of a basic idea right and so the problem with it is that very grad as well okay the problem from a consumer perspective is that you but you start gradually being manipulated by third parties who are paying to do so and inevitably that means that in the long term you're losing prospects in order for the same to work your information has to be free so for instance you get free music because your choices in music provide a profile of you that's then used to sell you I don't know antacids or whatever it is but the the long-term problem and the reason is not sustainable in the book I go through how they'll eventually be little artificial patches that can synthesize chemicals this is a long thing but anyway whatever technology is now making something that can be advertised as a link on Google or Bing or Facebook will eventually get automated away by free software so it'll no longer be there as a customer so Google's business model is gradually going to evaporate its own customer base so it's not sustainable is that clear and then another problem with it is it forces it's the only official business plan for consumer facing internet services in the world of free information so these totally different companies like Google and Facebook with different competencies and cultures are forced to compete for the same pool of customers which is ridiculous and creates this sort of claustrophobic bizarre competition that doesn't make any sense this would be saying like like bulb and horse-feed people should be competing with each other it doesn't make any sense like Google and Facebook should be different but they're not because there's only one business plan so yes I think it's a stupid business model it's the only legal one sort of you know if you really believe in free information the only model left is to micro model people and keep the model secret from them so you can manipulate them for pay and then furthermore another problem with it is that that the we've all grown used to the idea that there are these recommendation engines that tell us who to date and what music to listen to or whatever or what where to buy our plane tickets and that but the thing is we all know in our heart of hearts that it's a little scam II like we all know like any social scientist or psychologists that studies the dating sites comes to the conclusion that the algorithms don't work but we make them work because it's not actual science at social engineering and we allow those two to be confused and that then creates this atmosphere where big data becomes treated as a form of manipulation instead of science which and sort of makes us distrust it I think and I think it's a whole other topic but big data is really important I mean real big data that's not that's not part of fake business games is critical to our survival it's the only way we know about global climate change and big models are the only way we know about the human contribution to big climate change so this stuff is very serious and the public knows about it in this way that they really know in their heart of hearts as a confidence game is a scam and that's really really unhealthy so anyway there are a lot of reasons why I dislike the advertising model that's not to say I don't work on supporting it while here because hey you know we when one has to be part of the world and also looking ahead for how to make the world better so I don't think it's helpful to be like this perfect soul and say I am just going to boycott reality because I don't think it's good enough instead what you have to do is work well within reality as it is but then also try to think reasonably about how to gradually improve it okay any other questions um yeah I love how we get where we are to there mm-hmm a common example for me is the vigorous evening survey phone calls to which my my typical response is well how much you're going to pay me to take you sir right right right seems like the right model if they don't think it costs them that 50 hours a person easier to collect data sharing a little bit of that will get better data for less money you don't want work that's my model I don't get that right so the question is how do we get there from here it's a hard one because we've gone down pretty far on another path right so in the book I outline a little bit about that I don't want to be too prescriptive because I don't want to commit Marx's error of presuming perfect for knowledge but I um I think you're a couple things one is every time a new platform of hyper automation comes around like 3d printing lately what happens is the open source movement grabs it and said oh we're going to have this open source all the models have to be open source because that's the side of everything that's good and holy or whatever and like just for once just to be experimental let's make one of those things be paid just to see what happens like what if 3d models weren't open source what we just as an experiment said we're not strict Orthodox we're not absolutist we're just going to try to see what happens and if what came out of that is a lot of interesting people doing well and more and better models all that stuff that would mean it's a one way one way is to do isolated experiments where the isolation is created by technological change another way is to start theoretically which I'm which I'm approaching and then to try to sort of advertise it to politicians and captains of industry or whatever another way is if all the companies could just get like you know there's four or five companies that kind of run the consumer Internet at this point it's like hyper consolidated people talk about how media is wide open because of the net but the truth is in terms of what actually reaches people it's more consolidated than it's ever been and we can just sort of get together and sort of try a big experiment I realize it's hard you know just us and Apple and Google and Facebook we could just do it you know how hard would that be for godsakes anyway um we all get along right newspaper business we're gonna try to monetize all right well you know the thing about monetizing is that you can't do it in isolation the micro payment system genuinely has to be universal at least in a domain so like if it's in if it's 3d printing it has to be in a domain because if it's only like a local thing like if you're just trying to monetize one newspaper it's very hard because of course the open free thing will route around it so it does have to be Universal I mean I think part of it is ideological I mean and I'm partially at fault for this but we've raised a generation of idealistic young people who were absolutely convinced that free free information is the only way for things to be okay and they have to understand that systemically and empirically it's just not working like it sort of works in the immediate sense but it doesn't work micro macro economically and it doesn't work for your lifetime yeah I think you cut us a little bit but another book itself yes um you talk a little bit about what it means to be you in 20 years when you're talking about what you feel your life will be like as an author as a public intellectual as a teacher in 20 30 years I this model works oh well um I mean you know what I so the question is what would it be like to be a public intellectual or writer in 20 or 30 years in a sense I don't worry about that too much because so few people are that's like a very small part of society I'm much worried about the broader middle like I said I mean I'm kind of a weirdo like I if we design the future for me it wouldn't work for other people you know like I I have to accept the most going to be an outlier but you know like I mean the utopia for me would really be a weird one let me tell you if you like there'd be weird instruments at every corner and I don't know I get infinite resources in my lab if you like oh you might run Linear Accelerator sure yeah you need that that I don't know that sounds right yeah uh yeah how you like or how would get any insights as far as how you would deal with like defectors or something in the new system work by somebody you talk a little bit of your book about like two-way links so like you know if I wrote a paper or something they link to be then I get a little bit of that action I couldn't charge less than he charged but I mean if you're buying kind of unless you DRM ideas then like what's to stop somebody from like you know reinterpreting that and well you know no better it is just no worse than the system - right now I always get this I always get this question about how you didn't force it and the thing about society is it has to be mostly voluntary so of course there's like art so um there's a I once knew a criminal who was serving time and said to me about 1 in 20 people is going to be a criminal and that was his experience and I've kept watch on that as in life in many in many different sectors of the world and I think it's a reasonable estimate so we can say is 5% of people will not accept the system and I I don't want us to become a really hard ass society where there's like like the police from Brazil who swooped down on bungee cords to arrest the people because they copied a file or something and I especially by the way right now I really don't like enforcing a anti cut like enforcing copyright with a really iron fist right now because there's no reciprocity I mean like if some kid copies a music file but meanwhile their life is being examined by thousands of remote computers to model them and manipulate them honestly it's hard for me to say to that kid oh yeah you're you better respect those copyrights because they're being abused all the time I mean are taking advantage of so but eventually you know what has to happen is there has to be a categorical imperative there has to be a Golden Rule feeling look this is a lab with a lot of techie guys I bet a lot of us know how to pick locks you know or at least I'm just guessing a lot of you here could go out into this parking lot and steal a car right now and you wouldn't have any problem with it and I you know they're not that the reason you don't steal cars is in part because it's illegal it's in part because you might have like these ideas that it's the wrong thing to do but it's also in part just because you don't want to live in a world in which cars are being stolen all the time you know you like the idea of normalcy being the car doesn't get stolen you know and that feeling that that broad sense of categorical categorical imperative of acting in the world and the way you wish other people would act towards you is really what holds the whole thing together the police and enforcement can only do a little tiny bit and so you know this is another example of a Maxwell's demon fallacy if you think that some big computational scheme is going to keep people in line of course that's going to break I mean give me a break so so this scheme has to be a social process in which the broad majority of people feel it's in their own interest and it has to demonstrably be in their own interest you know or else it fails enforcement can play a role there can be a certain amount of it perhaps but it can't be the centerpiece it can't be the main question and it can't rely on I think there could be DRM but the RM DRM should serve is just a reminder of what social contract entered into it shouldn't serve as an iron fist you know so the questions online can we take at least one and then have that a wrap-up sure just about how the market counting works on Second Life so again how the market economy Oh second life yeah so second life is an interesting experiment it's I think it's a little less in the air than it was a few years ago but I was an advisor to it at the start and it's a it's a you sure you know what it is it's an online virtual world where you control an avatar with a very sort of low bandwidth method and it's got a slightly sort of Burning Man kind of a feeling to it overall and I think there are some successes and some failures in it it is monetized in the sense that people buy and sell and sell virtual tchotchkes on it it's not universally monetized and that a lot of things happen on it that aren't monetized so it's like a half way system it has pretty poor quality tools and it's a very rough implementation when the thing was going up a typical argument I had with them was that you can't possibly plan to ship it with only that it needs to be better and they say oh come on we need to ship it you know and it was like it's very much like the arguments we have in Microsoft all the time I think and it I think they're probably right because it did get it had its moment in the Sun I don't know you know I mean the distribution of outcomes is not quite a bell curve but it's not a it's not a stark power-law either it's kind of in between so I'd say it's an intermediate result in terms of the spread of outcomes and I I know I mean I think it's it's I think I mean it's encouraged me that something can work I don't think it was perfect and I don't think it I don't think it gives us like the proof that we understand everything but I think it was it was worth doing well thank you very much you

Channel: Microsoft Research

Views: 42,832

Rating: 4.876543 out of 5

Keywords: microsoft research

Id: -B1hOBOTMSs

Length: 77min 54sec (4674 seconds)

Published: Fri Sep 09 2016