 Good afternoon everybody. My name is B Dale for those of you who haven't met me. I was asked to briefly introduce our next speaker, which is an immense privilege for me as John and I are. Our paths have intersected a number of times over the years and he even to me has been an immense inspiration as one of the folks who has probably spent as much time as anyone I know of introducing and explaining the concept of free software and how it can be a good thing for businesses and individuals in all sorts of venues all over the world. One of the consequences of his sort of unique opportunity over the last few years to travel and speak on behalf of free software principles in lots of places around the world is that he's had the opportunity to see what he thinks are some interesting opportunities to sort of take a more holistic view of how IT happens in certain environments and maybe do some things that would help make the world a better place. So without further ado, John Paul, better known as Mad Dog, to many of us. Thank you. Here, hopefully this will be okay. Maybe not. We did? Okay. All right. This is about a project called Project Cawa. This is the agenda for the talk. We're going to talk about the goals of Project Cawa. Some of the techniques we'll be using in Project Cawa to make it viable. The current status of it, and a summary of it, and then finally open up for questions. And although I can't stay here for the whole of DevCom, I will be here for most of tomorrow. So if you have any ideas or things like that, please feel free to come up and talk to me about it, because this is very definitely a project which is in flux and growing. So here's the goals. First of all, I'd like to create millions of new high tech jobs. Yes, you heard that correctly, millions, not hundreds of thousands, millions. A lot of these jobs are going to be in Latin America. That includes Mexico, Central America, and South America. But in looking at it further, I realize that there could be lots of jobs created here in the United States. But the economics are slightly different. You'll see why in a moment. We want to make computers easier to use than they are today. We want to do this in a more environmentally friendly way. We want to be able to basically create a wireless bubble of the most urban areas that would be gratis, free of charge to people, and to also try and attack the contention problem that's happening currently with cellular systems. We want to be able to create a low cost, or perhaps even gratis, supercomputing facility. And finally, we want to do all of this without using government money. When I was in college, I was a Democrat. As soon as I paid my first income tax bill, I became a Republican. During George Bush, I was a Democrat again, and now I'm back to being a Republican. Okay, first take the first one, creating millions of new jobs. What we want to do is to enable a systems administrator of free software to actually become their own business person, to actually create their own business in providing computing services to other people. We want to provide the business plans for them, which is typically hard for technical people to do to get their mind around the business plan. We want to provide marketing materials for them to market their talents to people. You'll see how that's going to work in a moment. And we want to make training and certification and licensing incredibly easy. The training part's always hard. You have to learn. But we want to be able to make the certification and licensing for these jobs easy for them to do. The systems administrator is going to be not only a systems administrator, but an entrepreneur. They're going to go out and find customers that want these services. They're then going to be able to go to the bank with letters of intent from these customers and get a commercial loan from the bank to then buy the equipment, which they're going to lease to these customers over time, and then to be able to pay this back through the money that they get from these customers. The systems administrators and entrepreneurs are going to have to have training both in the technical aspects of computer science and the business aspects. We intend on providing this training both online and for the people that don't have a good access to the internet through the process of putting the electronic training materials on a DVD so they can put them into their computer and play them. We're also going to provide classroom training for those people to find it difficult to learn on their own. And also an apprenticeship program so that people that have had little experience with computers can become apprentice to somebody who's an established systems administrator and learn that way. And after they get their training, to be able to certify them as having been trained and get them licensed and bonded. Now, we started this project in Latin America. I go down to Brazil probably 40 times a year in different things I do. I saw my first Beowulf system and physically saw it at the University of São Paulo in 1996. When most people think of Brazil, they think of the rainforest or maybe carnival or piranha or stuff like that. But in reality, São Paulo, Brazil is the second largest city on the face of the earth. They have, according to how you count it, between 29 million people and 19 million people. And that picture in the upper right-hand corner is a picture of São Paulo. There are actually five large cities that grew together. I tell people, if I was the mayor of São Paulo, I would just kill myself. And 80% of the people who live in Latin America live in an urban environment. São Paulo is not the only city like that. You have Manaus, you have Curitiba, you have Porto Alegre. You have a large number of urban cities. And the percentage of people keeps increasing as people move from the countryside into the city looking for jobs. And typically what this causes are large slums called favelas. We'll see more about those later. Now in those favelas, just like with one laptop per child, there's lots of people who need training. And while Nicolas Negroponte kept showing pictures of the African plains and the little child sitting underneath the banyan tree and wanting to bring education to them. And I applaud Nicolas for that. I think the one laptop per child inspired a lot of people to start thinking about it. I really don't care about that little kid underneath the banyan tree. What I really care about is 2.5 million kids who are in São Paulo or the kids in New York City who are not 500 miles away from the internet, but often they're only 50 feet away from the internet. If we can bridge that last 50 feet in an economical way, we can bring the information and the training to them that they need so they can get jobs. We want to do this using Think Client Server Computing. Now I've been in the computer industry for 40 years. And when I started off, you had a mainframe computer that you took your deck of cards down to and you prayed to the card gods that they wouldn't be dropped on the floor and you get scattered and stuff like that. And every once in a while that actually happened that you didn't get them scattered. But really the best type of computing experience I had was in a time-sharing system. Where basically you had operators who took care of things like doing your backups for you and they took care of things like installing new programs and all you had to do was sit down at the terminal and type away and life was sweet. So I think the Think Client Server Computing actually comes back closest to that and still gives you the power of something sitting on your desktop to do the graphics and do the interactiveness you need as well as having the capability of expanding the server to be able to give you the computing power that some people need for certain projects and the IO power for that. So we want to create a high availability server and we want to put them in the basement of all those tall buildings you saw in Sao Paulo or perhaps in the community centers and then wire those to the Think Client's a gigabit or perhaps even 10 gigabit per second Ethernet. Now a lot of people say to me, why wire? Why not do things wirelessly? Well there's such a thing as bandwidth and there's such a thing as privacy and I think that you can really benefit a lot by trying to create as many back calls as humanly possible from the Think Client to the server and be able to maintain it. We'll see a little bit more about the bandwidth issues in a while. We also want to be able to support fact clients with this because there are people that do have Windows systems and macOS systems and they want to be also supported but we're going to have to charge them additional money because it's harder to maintain those type of systems. You have to worry about the licenses, you have to make sure that people are licensed. Now I'm a great fan of the BSA, the Business Software Alliance. I don't mean the Boy Scouts of America. I love them because they keep telling me all the time how many people are pirating software. In China it used to be 96% of the people pirated software but ever since the WTO moved in there they're now down to only 84%. Vietnam is still at the 96% level and even here in the United States the so-called richest country in the face of the earth we pirate about 34% of our software. So if you're caught doing that you could get all sorts of trouble and this is one of the reasons why we would have to charge a little bit more money for maintaining Windows and Mac OS systems. Now as far as making computers easy to use I'd like to introduce my mother and father. Dad is 89 and mom as well. We won't say how old she is because she wouldn't like it. But Dad's idea of doing a backup is what you do with a car to put into a parking place and mom's idea of a virus is something you cure with chicken soup. That doesn't mean they can't use a computer, they do. They surf the web, they send email and stuff like that. But doing the things like getting rid of viruses or spam or doing backups is just something that's beyond them because number one they don't do it that often and they forget how to do it. So if you could do that type of thing for them and install new software for them and make sure it's done that would be great and then all they could do is just sit down and use a computer. Now believe it or not they're not much different than a lot of the people who use computers today. Folks, I hate to tell you this but we're weird. We like computers at least most of us do. We like working with them, we like futzing with them and stuff but 99% of the people on the face of the planet actually hate computers. They hate them. And they would just as soon not deal with them but they're forced to. And because of this they actually lose a lot of time and money. How much money is lost every day? There's approximately 1.25 billion desktop computers on the surface of the earth. About 95% of those are this operating system we don't like to think about. Oh sorry I called it an operating system. From this company we don't like to think about. Now if each person that used that lost only 15 minutes a day, approximately $5 in worrying about viruses or worrying about spam or having to install new programs and having to then reinstall the whole system because they screwed something up. All those things are not doing a backup for the last five months and then needing a file to get back. If you said that they only lost $5 a day that means that as a world economy we're losing $6.25 billion a day. That's almost enough to pay for one of George Bush's force. Now if that type of a number is too big for you you can think about it in a smaller number. Let's say a company has 300 knowledge workers, people that use the computer a lot and they lose 15 minutes a day. That's just as if nine people never came into work. Now if you've ever been a manager nine people not showing up to work is pretty bad. You really pay attention to that. Particularly when they're not sick. They just didn't show up. Kind of like having a hangover after system administration day. This is not counting what we call the frustration factor. The people that don't want to use the computer because they know they're going to hate it. They can't just sit down and read their email and browse the web because the computer is going to do something to them. They know it and they fear it and they don't want to do it. What would happen if every time you sat down a computer you had a rewarding experience? It actually did what you wanted it to. Now I'm not so Pollyanna as I'm going to say that we can get this time, this time of error, this time of waste down to zero. But what if we can reduce it to only two minutes a day? That means that six of those people magically showed up for work. That means that we saved about four billion dollars a day as a world economy. That's a goal. So what is the systems administrator entrepreneur's job duties? They're going to maintain the software on the server. The thin clients aren't going to have any software on them. They're just going to upload that all the time. So this person is maintaining this software on the server in a virtualized space so that the thin client can actually reach down and look at a virtualized space. They can be running the version of the operating system that they want to because they're talking to multiple virtual machines. This person is going to monitor that software, make sure it's working properly, installed properly. They're going to do backups. They're going to eliminate spam. They're going to eliminate viruses. Do all the work that a good systems administrator is supposed to be doing. Now going back to Sao Paulo, you would see those tall office buildings. Well, a lot of times there's not just one company that is tall office buildings. There's many companies. Maybe companies that have two to four or five people in them. And those companies can't afford to have a systems administrator full-time for them. So what they do is they train some other person, the office manager, the systems secretary or something like that, how to take care of the computer. Or maybe they have a bar come in and take care of it once a week or once a month. And that just isn't enough to do that. So what we're going to be doing is we're going to have this person basically at the bottom of every one of those buildings down in the basement taking care of these servers. Maybe it'll be one building, maybe it'll be two buildings, but they'll be within easy walking distance of any of those people in the building in case they have a problem. Now the other thing the systems administrator and entrepreneurs are going to be doing is giving classes, classes to the end users as to how they can use a computer system better. They're going to be doing things like giving support to end users. Now like I said 40 years ago when I had a problem with my computer, I turned to the person next to me because they also had a master's degree or doctorate in computer science and we figured out what was wrong. Or if we couldn't figure it out, we turned to the person on the other side of me because they also had a master's degree or PhD in computer science. Or if we couldn't figure it out, we went to the person on the floor below us because they had a master's degree. Why? Because the computer systems cost 2.5 million dollars and that was when a million dollars was a lot of money. And these things weighed 50 tons. And if you wanted to go for a course to learn how to use it better, it was simple to talk your manager into it. You just say, hey, this course will improve my efficiency with a computer by 10%. Imagine you go 2.5 million dollars, 10%, that's 250,000 dollars. Go for the course. But today the computer systems cost 200 dollars. You just say, hey, send me this 5,000 dollar course that'll make me 10% faster on the computer. And they laugh at you and say, what? You know, it's crazy. This is breeding of familiarity. And what's happening is our support keeps getting moved further and further away from us. And now, if you have a problem, you have to call up either India or West Texas. And neither one of them can you understand. Particularly the people from West Texas, okay? So this person is going to act as the support, the first line of customer support for these people. And they're going to negotiate and sell additional resources and hardware and things like that to their customers to make additional money. Now, besides the horizontal support that these people will be providing, horizontal being the operating systems, compilers, web browsers, things like that, we're also looking at vertical markets. And one of the vertical markets is small to medium business. Now, in this country, 86% of the people working in the private sector, that's not the military, not the government, 86% work for small and medium business. That might be surprising because you hear about GM and stuff like that. 90% of the gross domestic product comes from small business. And yet, these are the people that typically can't get the support that they need. That's a problem. So we're going to have, we're going to concentrate on small to medium business. Apartments and condominiums. When I was growing up, I was watching the Jetsons and stuff like that. We were going to have automated homes. We were going to take care of our heating and our light, save a lot of energy and stuff like that. It never happened. But you know something what we could do today is we place a lot of these little systems that are floating around our house. The wireless router that nobody has tuned. Everything says links us, you know. I mean, it's always set to links us. We could set that up. We could make sure that none of them interfered with each other. We could replace a lot of other systems. We'll get more into that a little bit later. And then hospitality. Small hotels who would like to have the same facilities that a large hotel has. Or point of sale terminals. A lot of point of sale terminals are extremely expensive closed source proprietary systems. We could make it open source and replace them. And make them better. Not just replace them, make them better. Apartments and condominiums. Let's take a look. First of all, you're in your house. You say, hey, I want to just connect to the internet. I want to use computer systems. Sure, we could do that. Over-the-air digital TV. With one tiny little chip, we could make your computer system a very nice high definition TV system. A lot of you know this already. You've already done Myth TV or Linux MCE or something like that. You may already have created a world-class media center out of your PC. But mom and pop find that difficult to do. We could do that for all of them at one time. IP TV, IP radio, radio from around the world. If you're in a new language, you can listen to radio from Brazil or radio from Germany or stuff like that. Voice over IP, inexpensive telephone calls or even an intercom system inside of your house. Controlling your lights, security system, inexpensive webcam, a little bit of motion detection software. You can have a security system. You store your pictures, store your music, have it as a calendar, alarm clock, store your recipes for your kitchen, all sorts of things as home automation. Now, when I first started to talk about this and say, hey, we could do this, we could offer this to people, I had a professor at one of the universities in South America come up to me and say, my students are already doing that. They get an internet line from the telephone company and they set up the little server and they sell services to all of these, all of the people that live around them and that's how they make a little bit of money for their courses. Well, what we're going to do is systematize that and make it legal. Hospitality functionality. Down in Brazil, there's a whole bunch of tiny little hotels we call Pusadas and they'd love to have some of the same facilities that the Marriots and Hilton's do, particularly X-rated movies in the room. They love to have that because that they can charge for extra and we can supply all of that stuff for them. We can give them a well-run wireless internet, wired internet with all the services of a large hotel and the system's administrator to keep it going all the time. A lot of times you go to a hotel and you try to use the wireless and half the wireless routers aren't working. They're not delivering, they're not working correctly. Well, you would have somebody that was constantly monitoring that and making sure they did work. Now, the next thing we want to talk about is the environmentally friendly aspect of this. In Brazil, there is the currently the world's largest hydroelectric plant. The one in China is still coming online. It's not quite the same as Itupu, which is here. Itupu generates 16 gigawatts of electricity every hour. Your desktop PC typically is anywhere from 250 to 300 watts of power. I've actually seen an IBM workstation that was 850 watts of power. That's about 150 watts of power more than the house that I was growing up in had all together. But even at 200 watts, that means that this power plant can only do 70 million computer systems. That's to run them. But there's a rule of thumb that says for every watt you put into a computer system to run it, you need at least one other watt to cool it. Air conditioning. Actually, because of the inefficiency of air conditioning, it's two other watts. And this is particularly true in Brazil where there's a lot of it that's around the equator and they're very warm. So if we can cut the amount of electricity, we also cut the amount of heat generated, and so we win two ways. The other thing we want to do is create a smaller desktop. A lot of times you go into a place, the first thing you see is the person's desk and one quarter of it's taken up with a PC sitting on it. Or they've got it underneath the desk where they kick it constantly and it gathers dust. Well, what we would like to do is make them very small and mount them on the back of an LCD panel so it frees up that disk space, that disk space. We also want to have them a longer lifetime to keep them out of the dumps and to make them ROHS compliant so if you ever do grind them up, it's easy to get rid of them and safe. So a solution here is to use some of the thin clients coming out based on something like the Atom processor and make sure they work off a 12 volts. Why 12 volts? Because 12 volts is the universal voltage because car alternators work with 12 volts. Car alternators, boats, campers, all those things. You can get a 12 volt refrigerator. You can get a 12 volt microwave oven. You can get 12 volts, lots of stuff. And if you had them working off with 12 volts, then when you take them out to some places far away, you can use an old alternator off of a broken car and a water wheel to generate the 12 volts or a solar panel to generate the 12 volts. The other reason for making them 12 volts is we're incredibly inefficient. We take 110 volts or 220 volts out of the wall. We then take it down to 12 volts to put it into our UPS. We then bring it up to 110 or 220 to feed it to the PC. We take it down to 12 volts to actually drive the PC. That's crazy. Take it down to 12 volts and stick it in the PC and run it right off of the battery. That's what we should be doing. We want to make it such a small amount of current, such a small amount of wattage so we can always leave it on. Why? Because when you turn off your PC, it's worse than a boat anchor. At least my boat anchor will keep my boat in place. I tossed my PC over the side of my boat and it still dragged along, you know. It didn't hold the boat in place. So you want to leave your PC on all the time. Why? So it becomes your telephone, your alarm clock, your security system. You turn it off, those functions stop. You leave it on, it can do those functions. That's what we want to do. We want to make it multifunction with virtualization. We want to be able to make this little PC also be able to support a wireless mesh infrastructure. We'll get more into that a little bit later. We would like to put some of the new functionalities coming out into it. 60 gigahertz ethernet. That's going to give you about 7 gigabits per second over the next 10 meters from it. That eliminates a lot of the wires that's in your house. It also allows you, through wireless ethernet, to transfer about two times faster than Iceda. Two times faster than USB 3.0. We want to put a cellular modem in. Why? Because here in New York City you can't make a 3G call anymore. Try and use your 3G phone without sticking AT&T network we keep hearing about and you can't do it. Why? Because so many people are downloading porn to their iPod that nobody can get through. Now we can't do anything about that because again we're limited in bandwidth. Don't believe those people that tell you bandwidth is unlimited. It's crap. It's limited. What you have to do is you have to shrink the cell. You have to make the cell smaller. And that way you can share the bandwidth over the spatial area. But we can't do that in New York City. Why? Because it's expensive. It's expensive to put up these additional antennas. It's expensive and people don't want them on their buildings anymore. So what we're going to do is we're going to make every single one of these wireless, of these thin clients, in effect, offend the cell. It's going to pick up your cellular phone. It's going to send it through the regular wired internet, back to the telephone company, and then it'll stick into the cellular network. And we will have fixed that problem. These things will have no fan and no moving disc in them. Why? Because fans and discs are the first thing to break it into your computer system. How many of you have a stereo system that's older and 10 years old? How many of you have one that's older and 20 years old? I do. Why does it last for 20 years? Because there's no fan and no disc in it. I guarantee you put a fan and a disc in it, it's going to crap out almost immediately. You take it down and put it on your boat and it'll crap out even faster. Why? Because it's salt air. Salt air just eats fans and eats discs. So keep that fan and disc out of there and it'll last a long time. Now, you could have many of these little thin clients throughout your house because they're only drawing 10 watts a piece. You could have 10 of them, or six of them, and still less wires than one light bulb. And it replaces many of the other control units which you have in your house. It saves space on your desktop because it doesn't sit on your desktop. It sits behind your LCD monitor. Hospitals. PCs are death in hospitals. Why? Because they're nice and warm. That's where the staff germs live. Okay? And thrive. And then all of a sudden the fan comes on and blows them all over every place. Thin clients without the fan sealed make a nice environment where the staff germs can't get into. And like I said, we can support Microsoft Windows and OS X, but we're going to have to charge more for it. Now, the servers in detail. We're going to have standard servers. I call them baby bear, mama bear, and papa bear. And you're going to be able to realize parts as you need more and more capacity. How am I doing in time? Realize parts as you grow in capacity. I'm going to just talk about mama bear and papa bear for a second. It can be made out of standard industry parts. Unlike OLPC, we really don't have to invent anything new here. Everything is standard and all the software is open. The baby bear is actually going to be made up of two thin clients put together with a heartbeat in between to create a high availability server. Why? Because I want to send those out to the same places with the water wheels and the windmills and stuff like that. And if half of the server goes bad, I can sacrifice one of the thin clients and rebuild the server, send the thin client off to be repaired, and get it back. Because in a lot of these areas, it may take two weeks, three weeks, four weeks, a month. There's no UPS, no FedEx. No FedEx overnight, anyway. More like FedEx and a Mule. So you have to be able to have some time to get the repair parts in. Oh, and again, by having them at 12 volts, it becomes real easy to create a UPS because that becomes a car battery. Now, the networking. Typically what happens today with networking is you call up the telephone company, they send in DSL into your house, maybe 1.5 megabits a second download and so many kill bits a second upload. Maybe you have cable and they say, okay, 50 megabits a second if you're really lucky some places. But what we want to do is bring in all of this internet into the base of the building, attach it to the server so we can accumulate perhaps 300 megabits a second into the server and remember from the server to the thin client you're hooked up with a gigabit per second ethernet or higher. This means if you're the only person who's actually accessing the internet you might be accessing it at 300 megabits a second. If all 300 of the clients that are attached to the server access it at the same time you're getting 1.5 or 1 megabit a second but that's highly unlikely. So you're actually going to get a fairly substantial internet usage from this. And these back calls will increase the latency over the traditional wireless mesh. A lot of times when we talk about wireless mesh we're talking about trying to get to far-flung remote areas by having packets retransmitted from one mesh client to the other. What I'm talking about wireless mesh is perhaps one or two hops before it actually gets to a back call and you can actually get into the wired internet with quality of service types of functions. The internal network of this system is going to be protected from the wireless mesh network by virtualization techniques and using VPNs. And we'll be able to control how much of the ethernet wireless network the system's wireless network goes out to people using the wireless connectivity by throttling it down to perhaps 1 to 4 megabits a second. Now if you're a thin client owner and you know that some of your wireless internet is going out and being used by people you say well I don't want them to use up all of my ethernet so you throttle it back to 1 to 4 megabits a second and the wireless people get quite a bit of ethernet that way and because you have a gigabit per second to the server you're not really concerned with losing that much. Now because of this we think we can create a model much like the FON model. This has been in use for years. In the FON model you're either a Linus or you're a Bill. If you're a Linus you give away your ethernet, your wireless ethernet so that when you go outside of your little bubble and you go somewhere you can connect up for free. If you're a Bill you charge people to use your wireless ethernet and then when you go out you expect to pay for it but at least you get access. We like the Linus model the best and what we like to do is have everybody have the Linus model and make this so prevalent that people don't have to sign any type of terms of agreement, they don't have to start up the web browser, click here 16 times, simply open up the wireless device and use it the way that ethernet and internet should be used. Now because of this we want to have everything virtualized and all data encrypted so that people can feel relatively secure and this is the type of thing that the systems administrator entrepreneur can also help to assure. We want to have virtual application development environment. I'm a great believer in the cloud. Unfortunately there's four different models for the cloud and three of them stink. The only one that makes any sense is the one where you create a virtualized container. You build your application inside of it and then you move that virtualized container from place to place when you need more resources. Need more CPU power? Move it to a larger machine. Need more space? Move it to a larger data farm. You know, that's what you do. So we would like to be able to create this virtualized container first in a thin client so you could run it on that if you wanted to. Then you could move it down to the server in the basement of the building and run it there and then from there you could move it out to one of the providers but you're still in charge of what happens inside of that container. That gives you the flexibility of getting the extra resources that you want at the same time as having control over the software that's inside that container. Now, we also are very conscious about the data and we think that you should be able to have control over the data. Well, you want to store that data on your thumb drive, store the data down in the server, or store the data off in the cloud and you should be able to determine that at any time. Now, the interesting thing about this is remember you've got all those servers sitting in the basement that are all connected to the internet at perhaps 200 megabits a second. During the weekends, the people are all at home or they're also out at the beach or something like that and so a lot of the servers in companies are very low utilization. During the weekdays, a lot of the people are at work or at school and so the servers that are at the apartment houses have low utilization. If you think about SETI at home, you can actually utilize those services back and forth and basically create a very large Beowulf supercomputer that with the proper programming be able to utilize millions of spare CPU cycles to solve problems. Because Eastern Client is also a wireless mesh repeater, we believe that, and it has the capability of doing 60 gigahertz ethernet, we think that we should be able to create this wireless bubble over large portions of the city and then give people a cheap wireless interconnect device to access the internet for the information and the training that they need. We would also like the SETI's and Client's for less home ground and this is especially true in Brazil. In Brazil, the government has a 100% duty on all finished products that come into Brazil. If you however, you have some type of assembly or some type of integration that's done in Brazil, then the duty on those parts goes down to 6%. And so at least for the Brazilian economy, having the machines designed and built inside of Brazil makes a lot of sense. But in addition to that, there's a lot of small companies that are happy to assemble things but they don't have the ability to design them. So you can actually get a design, a reference design done at a university and allow these small companies to manufacture them. Now people say to me, how much are you talking about? And for a moment I'm going to take you into fantasy land and then we'll come back to reality. There's 192 million Brazilians and if 80% of those live in an urban environment that means that's 154 million people. If you think that each one of them might have two thin clients kind of associated with them, one at home and one at work or one at home and one at school, that means that basically you have about 300 million thin clients. Toss in point of sale terminals, cash registers, that type of thing, about 92 million and you have about 400 million thin clients you're talking about. If you say that there's an average of 300 thin clients per server, you're talking about 1.3 million servers but because they're highly available servers, redundant, you're really talking about 2.6 million. Now again, I'm not a Pollyanna. I don't expect 100% penetration. But even if you only had 1% penetration, one out of 100 people in Brazil utilizing this method of getting their computing, that's still a heck of a lot of systems. And people say to me, well this is going to be really expensive because all this hardware stuff is going to be expensive but remember we're going to be paying for this over time. These people are going to go to the bank, borrow the money, buy the equipment and then pay for it over time like you pay for a car, like you pay for a house, like you pay back a loan when you're buying a business. And therefore the hardware is actually the cheapest part of all of this. The most expensive part is buying the services, buying the internet connection, buying the media that people want to watch. Paying the salary of the systems administrator, all these things is more expensive than the hardware so you can afford to pay the top dollar for the hardware so that it lasts a long time. And the company I said I've been talking to, I say, I go to them, I say look, I'd like to have your best server. Please don't scrimp on this for me because I want it to last a long time and I need a service contract where you will go out and repair this thing on site because you can't send the whole server back to have it repaired. And I say, I don't care. Just sell it to me at the price you would normally sell it to me. I'm not looking for a discount because it's the least part of the cost of this. And it's much better to get something that's highly reliable that doesn't break as much than it is to get something that breaks once or twice if it breaks one time, the costs go way up. So what we want to do is to have the high quality long life equipment over time and pay it back. Other costs that go into this, we want to look at LCD lit panels because fluorescent lit ones have mercury in them, bad for the environment. We'll look at security and home automation controls that make sense. And we're going to be buying services in bulk. Now people say to me, the telephone company is going to fight this. The internet company is going to fight this. And the reason they're not is because right now, because they're selling to individuals, because they're selling to individual companies, their cost of sales is around 36%. Because here they're actually selling to the systems administrator who then apportions this out to all of their customers, their cost of sales goes down dramatically. The cost of support also goes down because instead of talking to my mother and father and having to start off with, is your mouse plugged in? Have you rebooted your system? Instead of talking to a trained, licensed, certified systems administrator, and they can start at a much higher level. When this person calls them, they know they've got a problem. So people say to me, what about digital inclusion? Because that's a large portion of what we want to do. We want to bring the internet to the people that don't have it so they can get training and jobs. And again, if we create this wireless bubble, then we can connect them with a very inexpensive wireless device and have them holding their programs and stuff back on the server in a virtualized environment much along the lines of Google Apps, Gmail, and that type of thing. We also would like to target people who are currently on unemployment and welfare to have these jobs as systems administrators. We particularly think that single parents, people who have children at home could do this job very well because they could do the monitoring right out of their own apartment, right out of their own house. We think the physically challenged people could do this, people in wheelchairs because you don't have to be able to walk to do this job. If you needed somebody to help you with the installation, you could hire them temporarily and then continue to have the job as a systems administrator. And what we're trying to do, particularly in Brazil, is on unemployment and welfare and make them taxpayers. And we think that this would be a big advantage and to be able to bring money into these favelas so that they can have jobs and generate an economy inside of the slum. Now, over time, we assume that the cost of these is going to drop dramatically because we're going to be turning these things out like cookie cutters. And we think that over time, the remaining costs will go down. So more and more people will be able to afford this type of capability and bring it even eventually into the favela itself. But also, over time, because these things have such a long life, we assume that people will eventually want to upgrade their thin clients, pay a little bit more for them, and their older thin clients, which are still perfectly good systems, would then be sold to somebody who can't afford newer thin clients and are given to them or donated to them as a tax advantage. We think we can generate basic service packages that would be either very low cost or paid for through advertising. So how does the systems administrator afford to buy this business? Well, the first thing they do is they're going out to get a loan from a private bank or they can get loans from friends and family. The amount of money to buy these servers and buy, say, 300 thin clients is approximately the same as it would take to buy a house. And home mortgages are relatively easy to get. We're also going to start an underwriting program from the state bank, so the state bank of Brazil, to underwrite these loans so that the smaller banks cannot lose any money. Every time the smaller bank creates a loan, there'll be a certain amount of the money that will go into an underwriting fund, and then if one of the loans does not be paid, then that underwriting fund will cover that. And so these banks will have no reason not to loan out the money other than maybe they can get a better deal and interest some other place. But the concept of not having the loan be paid is not going to exist. So how about how much money would a systems administrator entrepreneur make? And this is one reason why it may not work as well in North America. In Brazil, a beginning systems administrator makes about a thousand U.S. dollars a month. A good one might make about 2,000 U.S. dollars a month. Well, if you have 300 thin clients and you charge each person 6 dollars a month that goes to the systems administrator's salary, that pays for that base salary. Now, if the systems administrator does other things like gives classes, charges if people always are restoring files, they have additional hardware they sell to people. They do printing services, high-speed, high-capacity printing services or simple web programming or web design and things like that for some of their clients. They could generate a lot more money than just the base salary. So this person is an entrepreneur. They continually look for new ways to make additional money for themselves and they keep their servers responsive and continually moving them forward. Eventually, probably after about 3 years they'll have paid off the loan they got from the bank. They have a choice then of either lowering the amount of money they charge for their customers or continuing to put that money into expanding the capacity of their servers to meet new demands for their customers. But whatever it is, they own the business. So when they want to move or if they want to retire, they can go find another systems administrator entrepreneur that wants to take over an already running business. That second person goes to the bank, gets a loan, buys the business and now the first systems administrator entrepreneur walks away with a nice pile of money which they can use to start their next business. This eliminates the problem we've seen with a lot of government-sponsored things like this because when the government loses interest in it the systems administrator loses their salary, leaves and then the equipment is without a systems administrator or somebody to take care of it. I've seen this time and time and time again in various projects. How hard would the job be? Remember these are all fixed configuration systems. This isn't like you're moving into, you're being a systems administrator of all these different combinations of systems. We think, I think that I could, with intensive training, that we could train somebody in about 6 months to a year otherwise these people are on unemployment and welfare. Many of you could probably do this job right off the bat. We're going to supply as many things to help them as humanly possible including upper level support. And we're going to have forums so other systems administrators can help them as well as point them to solutions which will be out in places like Sourceforge. This is the basic steps of a person to become a Project Cower entrepreneur. Get the training, get certified, licensed and bonded. There'll be programs for that. Figure out which vertical market you really want to go after. Go and become the salesperson to sign up people with letters of intent. Take all of that to the bank along with business plans, spreadsheets of things that would be generated by the project. And then buy the materials and install them and start delivering services to the customers. And the message we have is that the harder you work, the more money you will make. We're going to implement this as completely open. We're going to have to publish everything out on the server. Nobody will have to ask anybody's permission to become a entrepreneur. If they see something they like, they'll be able to go out and implement it themselves. They will have to be licensed by whatever their local government is because they're going to be handling things like telephony. They're going to be handling things like service business. And I'm sure the government will need to license that. But the project itself is going to be owned by the community. The timeline is we are getting funding right now to do some of the difficult projects that it's hard to get community people interested in doing. We're planning our first vertical markets to go after. We're finding the specifications for those vertical markets now. And after that we start to integrate the software needed to supply those vertical markets. Starting next year, we hope to provide training courses to people that wouldn't be entrepreneurs. This is the Project Coward Board of Directors. Most of these people are Brazilian because we're starting this project in Brazil. After we get it going in Brazil, we're going to translate all of the documentation over to Spanish and go through Spanish-speaking countries. And finally we have a technical board and Fidel has been kind enough to grace us being on that. We're going to expand the technical board as we realize what technical issues and technical qualifications we need. And finally, I'd like to thank Linux Foundation for giving me support in Linux Pro Magazine for writing about this and the Debian program that we do. With that, I want to thank you very much. Oh, by the way, I love this quote from Edwin Lanz, the creator of the Land Polaroid Land Camera. Do not undertake a project that lets us manifestly important and nearly impossible. Thank you. Okay, we have maybe a few minutes for some questions or comments. Yes, ma'am. Could you use the microphone, please? We have an ability to use more than one sysadmin for a particular building. And the scenario I'm thinking of is suppose the tenant on the fifth floor decides they just don't get along with the personality of the sysadmin. My initial understanding of your project is that they'd be required to use those services. We've went into this a couple times, particularly with condominiums that we've been talking to. And there's several different business bottles that could actually happen. As an example, it could be that the condominium owns the hardware and owns the networking and simply hires the systems administrator to do that. Having more systems administrators than just one is not the problem that we're really looking at. So if you wanted to have multiple systems administrators, that'd be fine. What we're really kind of hesitant about are the people that say, well, what happens if I have like six systems administrators working for me and I'm the actual owner and these other people are just employees of mine. So they start to form like a corporation of these systems administrators. And that's really something I wouldn't like to see. Because I really believe that one of the things that is good about Project Calwa is seeing millions of small jobs, millions of small businesses that could then compete with each other. And as soon as you start to amalgamate that, you lose that competitiveness. So each one of these people is an individual business person. I would like to see at least that happen. But if you worked in something where two people went together and had 600 thin clients and then you had to in effect two systems administrator for the two thin clients, I think that would be okay. And that way, if you really hated one person, you could go to the other one. There's another question here. Yeah, I have Hello. Okay, I see you back there. I just wanted to comment that ARM processors are also very, very good for these requirements. And have you talked to Debian edu people that are doing similar projects with similar, with thin clients and similar requirements or other projects like LTSP? Actually, I am talking to the LTSP people. I was talking with Jim McQuillen when he was at a conference in Joppa Soa for so last month. And at first, I didn't think the LTSP had all the facilities we needed but version 5.2 of LTSP looks like it has about 99.9%. I talked with Jim, told him the 1.1% that I needed, he got this big grin on his face and said, oh, that's a great thing. And then by the time he actually gave his presentation, he had incorporated my need into his presentation for the next release. So either I'm a very good convincer or else, he really did think it was a good idea. And that particular thing has allowed us to cut a lot of development time off of the project, because we think we are going to be able to use LTSP basically as it is. I just got the throat cut from Phil. Thank you very much. I'll be around tomorrow. And if you have any other questions or things like that, please feel free to stop me.