Gary Flake: Not just the discussion of some of the things we're going to talk about today, but perhaps even a long-term discussion. I have basically two things that I want to talk to you about.


One is fundamentally I want to put forth an idea, a vision, if you will, for that we think the internet means, and what it can mean to society. And so this is by necessity going to be longer-term and broader than what we're used to talking about here in the Search Champs setting, but I think it's something that's going to resonate with a lot of you. In fact, there's many people in this room where lots of these ideas that I'm going to synthesize into one big idea, is something that people in this room have actually been components of or advocates of for a while. Some of this should be familiar.


The second thing I'm going to talk to you about is that we've announced just today that we're forming something called Live Labs. Live Labs, I'll give you some details a little bit later, but in general, it's a really neat partnership between MSN and MSR (Microsoft Research). With the idea that the two working in combination and in a very deep and meaningful way could do things that no other such an entity could possibly do, and also do it in a way that's to the benefit of internal partners and also external partners.


One little meta-comment about my presentation today. I have a tendency to make slides with lots of bullets, and Randy told me that one thing that people said lat year was, "No more slides with lots of bullets!" So let me just say that think about what's on the screen as additional details if you so choose. It is not required that you read every bullet in order to grok what we're talking about. Focus on my voice if that makes you happy knowing, knowing you can use that as sort of additional material.


So anyway the first thing I want to talk about is what I call the "Internet Singularity" and you know, the sub-title to this is "why right now in the whole history of the universe this the best time to be in internet technologies", and how this is actually a very critical moment in the history of the world. The second thing will be about Live Labs.


So, the back story to the Internet Singularity is that some of you may have heard the notion of the singularity, which Ray Kurzweil has been an evangelist and there's other people like Vernor Vinge who actually sort of introduced it, and the general idea of a singularity as it pertains to the emergence of the technology is that as described by this little diagram right there. Ray Kurzweil and others have said that as soon as you start making breakthroughs in nanotechnology, those permit things like hacking human intelligence, or artificial intelligence. Making headway in that because you can manufacture on the molecular level new types of computing devices. Or, if you make progress on artificial intelligence, that would enable you to have new incentive to build new nanotechnology, or to have human intelligence, or if you can hack human intelligence through the genome or manipulating neurons, that would give you insights into making progress on these things. So the general claim that proponents of the singularity say is that we're at the cusp of a phase transition where progress, a little bit of progress in all these things is going to snowball, and it's going to innovation and new things are going to happen more and more, and this is pretty exciting. I'm not going to talk about this thing, I'm going to use it as a metaphor.


For the people that just walked in I want you to hear that what I'm talking about today is not under NDA, you can blog, you can talk, you can do whatever you want. The internet singularity, loosely defined here, is that basically, a tighter coupling between the online world and the offline world, ubiquitous computing and how all these things come together creates a similar sort of positive reinforcement between creating a number of different things. And this will basically fuel a greater increase in the rate of discovery on a number of different projects. It's a kinder, gentler singularity in the sense that some of you maybe familiar with the fact that Bill Joy and others have actually talked about the original singularity as perhaps even a doomsday scenario; I think this is all good. So, I'm going to make the case for why we're approaching an internet singularity in four steps. In broad strokes that those the steps will be, I'll talk about democratization, anachronization of how things are done, how content is created, how connections are made, and what all that actually means. We'll talk a little bit about parallel distributions and the long tails, the internet ecosystems and network effects, and finally, the innovator's dilemma. I believe that these four things feed into each other in a very compelling way that sort of completes the story.


Now let me also say that I'm going to shamelessly use what some people term as 'buzzwords.' At the Last Web 2.0 conference, every time someone was saying 'ecosystem' or 'long tail' you'd hear people shouting 'Bingo', thank you Will, appreciate that. I think that despite the hype on some of these things, there's also a great deal of substance, and I'll try to justify this in the talk. So lets talk about first about democratization Part One of Four. We've seen a number of different examples of this where basically there's an increasing amount of powerful available to smaller users. Today's desktop is yesterday's supercomputer; tomorrow's cell phone is today's desktop. The rate of increase in terms of the ubiquity and power of the devices that we use is just driven by Moore's Law and some things that are actually super Moore's Law.


We're also seeing greater availability of powerful software tools which I'll highlight in more detail in a moment and integration among these tool-sets that creates entirely new opportunities for creating things. The biggest changes are around content, commerce, and community, and I think that it's really compelling to think about it this way, because we see positive reinforcement between those things, that I'll explain a little bit later.


Democratization of content. Over the past decade or so, we've seen a number of technologies that have really lowered the barriers in terms of what it takes to do new things. With Office, LaTeX, HTML, OpenOffice as well, now desktop publishing actually is not only going to the masses, it's going across the whole world, in some capacity. And if you consider HTML or other web authoring tools as a first-class member of the publishing world, that it's clear that the masses are actually producing more publications than the traditional media world. Images, digital cameras, Photoshop, iPhoto, are turning the amateur into the pro-sumer, and you can work your way down the list. The story here is basically that again to reinforce the power of the tools that people have accessibility is just increasing, and I think yesterday was a really good example of this, because when we started talking about the developers things that we're working on with respect to web search, we will this year have the capability to empower anyone in this room to make their own custom web search engine. It used to be that the barriers of entry were making a search engine that had any sort of personalization that you design was several years, perhaps a couple million dollars, in that order of magnitude. Now you have to have access to Web APIs.


Democratization of commerce. I think there's a number of different things that have fed into this story: Yahoo Stores, Office Live, Amazon, eBay, PayPal, Overture, AdWords, AdCenter, AdSense, RSS, Web API's, Mechanical Turk, the ESP game, web analytics. What's really fascinating here is not only are we figuring out how to make the economic ecommerce system out there more liquid, it's actually - there's a couple other things that are really subtle here. All the exchanges that are happening in terms of click-throughs and visitors and all this other stuff is instrumented in a way that the physical world simply can't be instrumented in. What's happening here is that with respect to business intelligence, we're finding that when you have a system that by its very nature is instrumented for revealing data bout it, we have the ability to usher a whole new generation of tools, Whereas where 10 to 15 years ago only Wal-Mart could do the store shopping cart analysis to optimize sites this is going to become a tool-set that everyone is using.


Democratization of community, is a very easy thing to talk about in terms of IM, email, VOIP. Again, all these are really simple examples. This isn't a hard case to make. We're basically making it so that its possible for lots of different people to connect in ways they never have before. A partial anecdote here -- I actually met the woman who would become my wife about seven years back on match.com, back in the day when that was a little bit, not something that you would actually say, it was like a shameful secret, in fact we had a cover story for a couple years. [laughter] But now we now proudly tell that story. [laughter]


So, the major thing for part one making the case is that today's amateur has more resources than yesterday's professional. This is really exciting. The difference between the amateurs and the professionals is diminishing over time. We have two different populations that are gradually blending into one. And the number of producers is increasing by the absolute and relative terms. The implications here are that the creators are dramatically growing and radically changing in structure.


This feeds into a nice segue into power laws and long tails. A lot of you are very familiar with this. Long tails is one of the buzzwords that people are frequently talking about. But the way that I like to best characterize it this for those you aren't familiar with it, I think the canonical example comes from the biological ecosystem. If you take a look at the distribution of the biomass of the physical world we find that there are a small number of whales and elements, but an enormous number of bacteria. If you weighed the biomass in absolute terms you will find that the bacteria weigh more than all of the whales. Long tails have a similar property. Chris Anderson of Wired is someone who's been a real evangelist for this generalized data and has really made the point that the weight of the tail is oftentimes greater than the head. The canonical example that they use with regards to the digital domain is that while there are few Britney Spears there could be vast number of unsigned performers that have yet to be discovered. One sort of requirement for true long tails to actually develop. It requires that tail producers be able to survive without necessarily being discovered. With the decrease in costs in the entry fees for becoming a participant, this is something we've seen. It doesn't matter if anyone reads your blog. You can survive quite fine with no one reading your blog, and maybe someone later on will discover you. It doesn't matter if no one discovers your garage-band based music that you're pushing out. You can do that for a while as a hobby and then eventually go pro later. Long tails are often characterized as having this sort of popularity here. The number of producers that realize those numbers versus the sale and popularity. So, integrating over this part of the tail, where this is going up practically forever, it accounts for a very significant piece.


So some really interesting things about long tails. We really find them in content, and commerce, and communities with respect to the online world because the physics of the online world differs from the offline world. The size of the warehouse, the shelf spaces is irrelevant. Distance and the medium are largely irrelevant. Aggregation, remixing, tagging all contribute because these become something that are first-class versions of content. My playlist, pushing it out there, actually is a new form of content. I don't have to be a musician now, I can merely articulate taste about music.


And the scale can increase independently of the human action in a very real way. So, the implications here is that consumers, people that are normally only taking things off the web, or off an ecosystem, are now actually becoming producers. When your pushing out your wish list, or playlist, or any of these things, you are now becoming a producer in the ecosystem as well. And in combination these small producers may outweigh the large.


Rael, it's right that Rael's in the front row, Rael has been an evangelist for some of these things, on the value of mashups and mixes, and things like that. And this is just more to make the point that there are lots of different ways that people are creating new content on the web, everything from tags to metadata to annotations and reviews. A subtle point is that mere activity on the web is also contributing to this ecosystem. So, when we start doing things like looking at everything from AdSense performance and that influencing the order of ads or looking at how different sorts of behavioral things where near participation is actually influencing the efficiency and liquidity of the ecosystem.


I mentioned before that with the long tail we get this gradual blend of the consumer population and the producer population. Instead of being two distinct things they are starting to blur, blend and become one. What that does is it sets the stage for having network effects. A network effect just to define it. It's the phenomena whereby the value of the network increases as a function of the number of participants. So the easy canonical example of that instant messaging or phone. If you were the only person with a phone it's kind of useless. Its utility to you is actually increased the greater the number of people that also have phones.


And there's also indirect network effects, like with respect to OS development. Users use an operating system cause it has the most applications. Developers write for an OS because it has the most users. That becomes something of a virtual cycle. We know that direct network effects in many ways are much more compelling and sticky than indirect ones.


So, putting all this together on this third major theme, we really do see something akin to the whole being greater than the sum of the parts and how all these things come together. In fact, there's something of a positive feedback loop on how content, commerce, and community all interact with one another. Content becomes the language of preference: "I like this type of music, this type of television, I hate these types of ads, I buy these types of products". That becomes something of a mechanism by which we define a part of who we are.


Commerce is a way of promoting information flow and efficiency so that the market actually becomes a means of propagating reputation through marking signals and other things. And it becomes a mechanism for sifting bad stuff from good stuff to a certain extent.


Community in this, when we now have a ecosystem with a vast collection of new information. The community now becomes the collective filtering mechanism. So when I talked before about the role of playlists, or wish-lists, and tagging, and those sorts of things. That becomes the community lens by which you find new information. It helps bring order to the chaos that would actually ensue.


The implications are that each is the chicken to the others' egg. As all three become much more mature, they set the stage for further progress on the others. In combination the networks mutually reinforce one another in a very healthy way.


Now, the fourth thing I'm going to talk about, I suspect is the piece that is least familiar to this audience only because it's not a current internet popular meme, or buzzword. It is in fact a popular meme or buzzword in business circles. There's this thing called the Innovator's Dilemma which some of you may have heard, but I'm going to go ahead and give a simple definition of it, just to align us on the same page. The Innovator's Dilemma proposes that there is this ongoing pattern that is repeated over and over again in a lot of different domains. The pattern is that the first in the industry focuses on a small number of high margin customers. That's where the money is. Late arrivals in the industry have to focus on the lower margin customers. The late arrivals learn efficiency because they have to compensate for the lack of margin for big scale. Meanwhile as competition increases, margins across the board also shrink. The established companies rarely learn the efficiency that the younger companies grew up with. And so the late arrivals win, because they can take optimizations and apply them to the head.


So let me give you a very concrete example where this happened. This is a first person story about a company that was involved with in my past, that suffered in some sense from the Innovators Dilemma. I was the chief science officer of Overture before I joined Yahoo, and before Microsoft. Many of you might know that Overture actually invented the business model, the paid search business. It was actually Google that came a little bit later.


The thing about starting a business like paid search, is that initially, it was universally held as the dumbest idea anybody ever came up with, that it couldn't possibly work. In order to make it work, Overture had to architect the system in certain ways that gave it a lot of transparency so people can trust it. so we had things like a transparent bid landscape. We ordered by bid because people wanted to know precisely what they were going to get for their money. We put editors in the loop in terms of improving the listings that would be displayed. We allowed for the independent optimization of individual keywords. We did not have the strong emphasis on having partnerships with a large self-serve audience. In fact on both of our relationships with advertisers, in our relationships with the syndication partners, the publishers that we had, we had a tendency to go for big partners like AOL and MSN and Yahoo!


Then along came this company called Google, came up with a similar product, fundamentally had some real similarities, but what did they do differently? They said, we're not going to use editors, we're going to semi-automate the whole editorial process. We're not going to have a transparent bid landscape, we're going to make it a little bit fuzzy because that might have some good market dynamics. So we're going to throw away transparency there. We're going to have self-serve to the extreme so that anyone can sign up and in five minutes have their ads up and we're not going to worry about the human editors becoming the first pass filter. We're even going to have self-serve in respect to syndication and we're even going to go with approximate match instead of exact match on the queries, as Overture did, because when you go with approximate match, you can get to the tail of the query distribution.


So when in every single case Google made a decision that applied to the tail, it was the tail of the advertisers, the tail of publishers, the tail of the query distribution, the tail even in terms of the efficiency in some sense of the different economic patterns. And basically Google went from not being in the game to being the dominant player in that game as is evident by their economics today.


So I got a little bit ahead of myself. So the first companies in the industry, in order to succeed in the long term, either have to pick one thing: either have someone else destroy your business, or destroy it yourself and do the next thing. We've seen this happen in a number of different ways with respect to the evolution of the hardware industry as well. I'm claiming that it's actually much more interesting, much more dynamic, and much more alive in the online world because of how all these pieces of round data fit into the story. In the offline world, we have huge startup costs for new business, in the online world, diminishing startup costs. You can get a store up almost for zero costs. The aggregate size of the tail is limited by the physics of the physical world. The aggregate size of the tail is potentially unlimited. More business usually implies more employees so you have to work harder. More business in the online world implies you may not actually require that you hire more employees, you have to work smarter, maybe a smarter algorithm that makes the difference . Quality product usually implies high touch, again is simply a better algorithm. And innovation iterations follow product business cycle. Version 1, version 2, comes out three years later. Here, innovation iteration follows a data flow cycle. So you're launching, you're instrumenting, you're getting feedback, you're new product release just went out five minutes later, as it's behavior adapts in almost real time.


So the major thing that a bigger tail makes for more potential disruptions of the non-physical aspects of the internet speed up the natural clock cycle for all these things worked out. The implication here is that that societal evolution is itself changing.


So to recap on these four major points. Democratization of anachronization, massive parallelization, parallel distributions of long tails, decentralization is bigger than centralization, internet ecosystems and network effects, the whole is greater than the sum of its parts; and the Innovators' Dilemma as high bandwidth and lower latencies on information flow, more frequent improvement.


So remember this slide. We started off with as singularity. Hacking human intelligence, nanotechnology, artificial intelligence, this sounds a little bit science fiction-y or in some cases almost religious about how people feel about this.


And instead of hacking human intelligence we're going to talk hacking about human knowledge, and instead of the ability to arbitrarily create physical artifacts for through nano-technology, we're going to talk about the ability to create digital artifacts. And instead of having the strong super AI , we're going to talk about the ability to analyze the online world. The claim here, is that the virtuous loop, the positive feedback loop that arises from all these things is a snowball effect as well. And that as we have greater ability to create digital artifacts whether it's a photo gallery or the tags in a photo gallery, or a wishlist. All of these things when added up make for an incredibly rich ecosystem where the data and the aggregate actually gives us more insight into how the world works. This in turn feeds our ability to have greater human knowledge, to create better things, to have more information at our fingertips. This in turn feeds our ability to create new artifacts, and the loop continues. I think the past five years have actually been a pretty strong example of a lot of these things. As we've seen blogging mature from something that has little reputation to something that is threatening traditional journalism.


The internet singularity, is as I mentioned, is really powered by all of these things. So ubiquitous computing allows anybody to be a creator of new things. The relationship between objects and information that form our understanding of our world. Search now becomes a means by which humans to discover new things. But keep in mind search is something that has to leverage the community in order to realize that full potential.


So, one general prediction, or one general claim I would make about all this: as time goes on, the internet is gradually becoming a more accurate mirror or reflection of what happens in the physical world. Because its effectively instrumenting things that were previously uninstrumented. Your social network is now sort of encoded in your emails, and your instant messaging network. In my case you meet your wife on match.com and there's the trail from that.


The limit of the physical world effectively becomes instrumented by the virtual world in such a way that it lends itself to a form of analysis that was never possible before. This also has implications to how science can actually be carried out in a revolutionary new way. The theories are tested relative to internet data empiricists we'll have almost unlimited data, and simulations actually allow us to experiment in a universe in theory, which is a radical change to how science is performed.


Some existing evidence around that. A lot of this is old news but I would claim to you that the future of the search engine in what will it take for us to realize all the promises that we've been told about what search engines can be, will only happen if we get something akin to true AI. We need to the search engine to become effectively a close approximation to the human mind in order to do the right thing for us. Other examples of how this is happening. Natural language processing has been completely revolutionized in the past decade in terms of how large corpora are being incorporated for analyzing the statistical properties of words . So this is one example of how science is changing. Virtual worlds are now used to study emerging economies, world of Warcraft and EverQuest quest. Viruses and Trojan horses are now being studied by people who work in traditional epidemiology. Yada yada yada.


This last piece I think is particularly interesting. I like to use this example. The National Institutes of Health helped push a lot of human genome data on the eh web, they then made a document repository that then allowed medical researchers to make references to gene sequences as they were discovered. They are now opening up APIs to allow data miners to mine the medical literature to make new medical discoveries. This is radical if you think about it. Data mining is now used for everything from drug design, to discover the origins of disease or a get better understanding of how genes relate to one another or regulate one another.


A corollary to all of this is that I would claim we have something like a computer science singularity. The more we know about the universe the more we can embed online that improves our ability to just engineer things in the general case. The online world lends itself more to a more thorough analysis. The more we can analyze the internet, the more knowledge we can extract. We have this virtuous cycle that is blending the disciplines of science, engineering, and mathematics in a way that's never been done before.


So that's the preface. That was a big preface. But that's the preface I wanted to give you about what we're doing with this new institute that we're building out called Live Labs. I'm going to ask you guys to pass this out. This is a one page manifesto that we put together explaining some of our reasoning behind this.


In a nutshell, we're making a new type of research lab. It's a joint partnership between MSN and MSR, it's not a company reorg, it is in fact just the articulation of a mission around what the internet means to us. So I tried to share with you in this first part what our vision is for the internet, how all this comes together. Why we feel that the internet in some sense is bigger than any company and bigger than all of us, it is itself the thing we need to promote and nurture. And Live Labs is one way we hope to do that.


It's a partnership between MSR and MSN, we're creating over a hundred new positions, that are dedicated new resources to Live Labs, people that will reporting to it. It's also virtual organization in the sense we have folks like Sue Dumas in the back of the room who's going to be leading some information retrieval and user experience efforts at Live Labs. In general philosophy behind this, is we've already hit the sweet spot between science and engineering, top down and bottom up, short term and long term, problems and solution, so we can better incubate projects, ideas and people as well. To do things that are actually somewhat untraditional, non-traditional in Microsoft given its historic emphasis on having a lot of effort into pure research, and a lot of effort in the product engineering world. We're recognizing that there's a continuum between pure engineering and pure research. There's a sweet spot between the marriage of them. We've had tremendous success with the web search team at MSR already. The biggest progress we've ever made has been as a result of those relationships. This is doubling down, or tripling down of those relationships that make it alive. We're basically building out these engineering incubation teams. Live Labs present at MSN and MSR, We're also applied research team within MSN.


A high level goal of Live Labs is to create a virtuous cycle like the ones we've been talking about. The simplest way of articulating this is that we want to build great products, great things for our users, great things that really resonate with people, things that realize the full potential of the web, what we've been talking about today. We want to rapidly expose these products to the outside world, get your feedback, find out what you hate and what you love. Implicit in this we're going to put out things that are maybe even less quality than a beta. We're going to swing for the fence but that means that sometimes we might strike out. We also want to have a high bandwidth channel to the outside world to the extent that we're people in this room would like to work with us on that mission, if this really resonates with you, we want your partnership in a number of different ways. Like I've said before, we want this in one sense to a be a prototypical examples, in the past has been a discreet way of doing business, a discreet way of doing science, a discreet way of building products. Our secondary goal hire more talent. This implies we're creating new career tracks for people. We want to invest more into partnerships. In fact we have multiple billion dollar budget, connecting with academic community and the internet community. Today we are announcing that we are giving out a million dollars in PhD scholarships that we're giving out right now. This is a joint effort that been done with MSR. We're also announcing today a half million dollars in research grants. These are no string attached research grants, you can get the money, the winners are going to in fact going to be selected from a large number of people who are submitting proposals. We're going to go for the proposals that seem the most interesting in how they fit in to this the larger mission. And they'll get research grant of various sizes, and there's now strings attached. Part of that we're also going to be releasing data. And too answer some questions that we had yesterday about data issues by we have to take a little bit of crawl, walk, run approach.




Cause it's really easy to mess up in a big way if you go open right away, too fast, with data. So the first thing we're going to do - is we're going to be very diligent for virtual and identifiable make sure we aren't doing anything to violated privacy. The first time that we release this data its going to be under some sort of restricted NDA license so that people can't give it away to third parties - but can do whatever research they want with it. Even a richer set of data will become available as part of this research program - than some of the things that have hit the news. The bottom line here is that we want to fully engage on how the internet operates and realize a mission that is consistent with that vision.


You know, to wrap all this up, we have pitched to you today an explanation of what the internet means to you. If this vision resonates with us, you should agree to us that this is transformational for society. Society is at a crossroads, Microsoft is even at a crossroads and how we go and build with the internet and work with the internet community is very important for how that progresses. Our mission relative to this vision is to create and nurture that virtuous cycle, from creation to publication to feedback. We've talked a lot about science and product development something that also works in blogging the instant feedback that you get. Our strategy is to create an institute hit the sweet spot.



So with that I think this is a good segue to talk about some of the details around the announcement of Live Labs, if you have any questions I'm not expecting anyone to read the one-pager and digest it right away, but that’s viable online and you can talk about and Quote it or not quote it. But I think this is a good time to take questions.


…The link to the manifesto was sent out. We'll put the slides up later.

Audience Member: [xx]

Gary Flake: So, there's a couple different things in a couple different ways. As part of our developer outreach effort we'll be opening up more APIs and through the RFPs we've set up for the research program. Be able to have access to APIs. We're ramping up some of our API limitations for people who are part of this. I don't know the exact numbers, instead of a limit of a thousand, ten thousand we're looking a limit of hundreds of thousands. There's actually xx ways people can contribute. We're sponsoring a number of sabbaticals for academia, also for the internet community as well. If you want to come do a three, six, or twelve month stint working with the team and trying to figure out where's the common interest that's always possible as well.

Audience Member: [xx]

Gary Flake: Yes.

Audience Member: [ mostly undecipherable, mentions Alexa web service ]

Gary Flake: So one of the first things that we're going to be building out, a sandbox, additional sandbox resources, internally and externally. The first customers of that will be Microsoft people. What Alexa's offering is a way for third party to write code on their servers which is a tricky thing. It's something that we thought about, if you've heard about the search macro effort. We're toying with the idea, something like a scripting language so people can get really sophisticated. Its something we are considering it we also want to balance it but also not be exploited. It's an ongoing discussion.

Librarian Audience Member: I want to ask you a more philosophical question, this is a great talk, and I really appreciate your perspective. One of the things that comes up from my perspective as a librarian, and as someone who speaks to librarian groups a lot. With this stuff and with the phenomenon of the long tail and how it's starting to wag its head so to speak. Is this metaphor controlled? And controlled as tagging versus controlled vocabulary. Experts versus the general public.

Gary Flake: taxonomies...

Librarian Audience Member: People who actually have PhDs and know. Its an interesting kind of growth, there's a been a steady decline of traditional control, in a traditional view of an expert, and even the idea that, and I would love to hear your thoughts on this, do you see this is a backlash against this, in the sense that there are other people places that are completely different metaphor. And secondly, what about the naval gazing approach that one could say we are creating here by having community have limited knowledge, driving what is tagged good or bad and potentially creating a smaller and more specialized view of the universe.

Gary Flake: two things longer and shorter much more involved and so. About seven years ago I did a research effort at [xx] where we studied what happens if you aggregate if you xx for trust at a xx level versus if you identified a niche community and localized biases to that. sort of like the difference of the results. Sort of like the difference of remember when "Who wants to be a Millionaire" and they used to do the lifeline where they would poll the audience. There were certain situations when polling the audience worked, like "Who is Britney Spears' boyfriend". Do you ask, "what is the current number of genomes that are identified in the human genome project?" I don't know that I'd want to take a poll that might want to hold a smaller audience And what we found in the study were that the results were completely different and it was a very justifiable conclusion that the results when you limited the vote to the expert community was vastly superior. So that's one piece.



Another piece helping to nurture and advance the relationships in MSN and we talked about the key challenge problems that we should all be thinking about. One of them is a generalized notion of reputation or authority or trust. It's clear to me that there's not one uber trust mantra[??]. I may be trustworthy in articulating an opinion in one domain and completely untrustworthy in another domain. So this is actually in a recent group gathering with MSR I highlighted three area that were absolutely critical to being able to make progress on this, and that was one of the three, these generalized notions of trust and reputation. I agree with you a hundred percent that it's not going to be a single variable, and it's something that even if you do something as simple as trying to of all the tail music content that's out there what's good and what's bad. Because a lot of people will ding the niche music within one domain. people who are familiar with it will really promote something.

Librarian Audience Member: Right, right. Well I have to just tell you one follow up which is interesting things about the newspaper which is very dramatic, people who read blogs, personalized the news they actually see, which again is a very dangerous precedent in my mind. But the interesting thing in my mind...

Audience Member: Why? Why?

Librarian Audience Member: Because people are choosing to limit what they know and what they see and the serendipity element is potentially missing.

Audience Member: I totally disagree.

Gary Flake: Give me another example.


[many people talk over each other]

Librarian Audience Member: I'm not saying [xx] this idea of choosing to personalize your world to the point. But I think that ecommerce interestingly enough is creating a lot of...


[many people talk over each other]

Robert Scoble: I wholly disagree. Because I read 840 feeds on a very regular basis...

Librarian Audience Member: You're an edge case.


[talking]

Robert Scoble: I'm not an edge case! Everybody says I'm an edge case come back and tell me. I've heard this 15 times over the last 15 times that I'm an edge case. When I was in school and I showed my Macintosh to my friends, all my friends said I was an edge case. Who needs Windows and a mouse? Stop calling me an edge case! The world is following me!


[Lots of clapping, bravos, and laughter]


[many people talking at once]

Audience Member: I agree that you're a leading indicator of a lot of things, but let me tell you.


[laughter]


I have a TiVo, I've been a passionate TiVo user for a number of years. But something has happened. All the shows that I used TiVo regularly are now gone. And because I've never watched a commercial in the last seven years because TiVo's recommendation system sucks ass. That's the technical term for it.


[laughter]

Audience Member: Is that an MSR term?

Robert Scoble: Here's a good example. I have 800 feeds, I know by reading those feeds, I don't watch much TV, I know that Lost and 24 are very popular in the geek community, because I watch a large enough group of experts, people I trust, people I've built a relationship with over the years, and people on both sides of the aisle, both Republican and Democrat.

Gary Flake: Who was using Usenet back in the 80's? Who remembers the sheer stunning power that Kibo seemed to have that you can just say anything about [unintelligible] ...more Celtic dancing or something like that.

[laughter]


And so if we were talking about Kibo we'd speak like "he whose names begins with a K and is followed by..."

And he was basically grepping all the Usenet feeds and that was his trick. My point is, you are the Kibo of the web in some sense. You have sophistication in terms of you use [xx] that 99.99% of the people don't have. Now what we need to do, and part of the mission about all this, is how do we democratize that and bring this to the masses. So we don't have to have the sophistication of a Scoble in order to do that. Where my mom can do it.

Audience Member: Whereas, now somebody like Robert...in the world, I mean its not like Blogger X down the long tail is going to matter as much as Robert [xx]...

Robert Scoble: There's a proof point in the room, and it's Michael Arrington. Where's Michael?

Michael Arrington: Yeah!

Robert Scoble: How long have you been publishing Michael? Seven months?

Michael Arrington: Yeah.


[talking over each other]

Gary Flake: You're all right, and if everyone here is actually right, and if everyone here is in the transition. The subtle point here is that while Robert actually has [xx] is one thing, there are other things in place.

And because of the quick rapid iteration about how things can become destructive on a much faster timescale. Someone can unseat Robert in a pretty rapid way...


[laughter]

Robert Scoble: And I'd argue Michael already has.


[talking]

Audience Member: She's really wrong.

Librarian Audience Member: You didn't let me finish my thoughts.

Audience Member: Well I hadn't finished my thought!


I've lived for thirty years Chicago Tribune Sentinel was a plantation mentality where they underpaid the writers, abused the community. The tribune sucked all the money [xx] and the newspaper deserved to die. Deservers to die because it was a bad thing and all these things that have happened are terrific and I want more of them, more competition, from all the sources so the best stuff will run out.

Audience Member: I want to say one more thing, then maybe we should have, then some other topic. A case in point, a friend of mine is Ed Castronova, who is the economist who was having a career crisis and decided since his academic career wasn't really taking off he would combine both his passions for EverQuest and economics and he's the guy who did the study. His seminal paper, which I think will be recognized as one of the seminal papers of the past decade in terms of influencing whole new fields was how EverQuest reflected an economic system ways compared in many ways to the major countries of the world, the top fifty countries in the world. That paper was never published in a journal. It was self-published on the web. He could not get it published. So there's an opportunity for new things to emerge, that the old school wouldn't touch, that wasn't reputable, but now has actually has created a whole institute, and is viable...


Robert Scoble?: I had dinner two weeks ago with Douglas Engelbart ...and he told me about how he was kicked out of the academic world in the 1970's because his ideas were too edge case and his ideas too weird, and it took the world 25 years to get the mouse, now there's a billion of them made.


[undecipherable]

Audience Member: Speak up please!


[undecipherable, something about a TiVo]

Gary Flake: Lets actually talk about , any questions related to Live Labs? We have three minutes. Anything related to Live Labs?

Audience Member: Do you have any products that are free right now or are you just hitting the ground running?

Gary Flake: So, a little bit of both. We're starting a new organizational structure with this, however there are some historic efforts that are feeding into this. As an example. Sue Dumais, in the back of the room, a leading person in information retrieval, to say she's just starting from scratch from information retrieval in her efforts here is to completely ignore the fact that she's a major force in that world for a long time.


We also though have other product groups that are already strongly interacting with these teams, so there are

works in progress that have been works in progress beforehand, but we're bringing them under the fold because it just makes sense. In addition... you'll see stuff this spring.

Audience Member: [undecipherable] What determines the success of [xx]... Is it eyeballs, is it buzz...

Gary Flake: I think any of those could be a viable method for success. I don't think anyone wants to turn away from profit. but in the end we want to create the healthiest ecosystem possible, sustainability. It also means that oftentimes we will be looking for the bulk of the newly equipped or created value to actually go to the community or the ecosystem instead of us. It has to be sort of setup as a win-win thing.

Audience Member: How do you see Live Labs impacting some of the global issues that affect society as a whole. I mean, we in this room are of huge privilege, and huge ability to people think on the [xx].

How do you see this fitting into the larger world and places where we don't have much privilege?

Gary Flake: Lets break it down into a couple more concrete examples. If we wanted to actually impact a big part of the world, things like healthcare, education, and other things are low-lying in terms of what we can do. And we have partnerships already in terms of how we actually create a new paradigm in academic publishing. We had someone from Wiley here who is part of discussions. The lessons that we learn about how you can change how academic publishing works, may in fact help us get closer to figuring out how do we change self-publishing in general. And obviously that's going to have a trickle-down effect to things like how do we make improvements to Wikipedia, how do we make improvements to the availability of information. We're suffering right now from the fact that we have lots of data, a little bit of information, very little knowledge and absolutely no wisdom on the web. ok. And that's what I would say the role of the librarian is still critical to this success. We need to figure out how we can increase the flow from data to information to knowledge to wisdom.


Ok, I think I'm being kicked out. I thank you for your time. I hope this was something that resonates with you, and I hope you join us in this mission.


[applause]