bang" the -- this week communicated -- this week community bang visits microsoft. peter: joining us is fred humphries, who serves as the vice president of government affairs for the microsoft corporation. what do you do for a living? fred: i am microsoft's chief advocate. i am the ambassador to the capitol hill, administration. leading a wonderful team of professionals. advocacy public policy issues that range from private security surveillance, trade tax, a host of issues. and one that is real important education. peter: you don't come from a tech background. fred: i worked on capitol hill. i worked for governor -- from

tennessee. i had the opportunity to work with and meet a person named al gore, who was with president bill clinton at the time. the work at the democratic national committee seems years ago. he to bank when it comes to microsoft, what are two of the egg issues that you -- of the big issues that you advocate for us? fred: on his privacy and security. about a week and a half ago the passage of the usa freedom act one that is a significant step in the right direction has yet to take on transparency and bulk collections. on the privacy side, those are

important issues. another one would be education. of science, technology engineering, and math. we are at microsoft's tech fair. we have a lot of innovation g oing on. we need computer sciences. we need scientists and engineers. we have 1000 scientists and engineers of microsoft and we need more. there are jobs out there. by 2018 there are 1.2 million jobs. you know what you are going to have to half? a background. hopefully so many innovative companies like microsoft and others, we need leaders wanted comes to information technology. the event that applies to the immigration issue. read bang absolutely. i am -- fred: absolutely. i am hopeful at some point congress will take on high scale immigration.

i don't know the exact number but when we have some of the innovators that are here and researchers that are here, they come in from all over the world and may contributions for our scientists and engineers. and for our other companies as well. still a need when you look at it from a job perspective. i say it is a step in the right direction because of the post activities. there is a lot of concern that troubled folks and many people we didn't know. to think about how the government is collecting data it is important that it is done right. the rules of the road and do -- and due process. what happened is the end when it

comes to section 215, the end of bulk collection. i think there are -- there is greater transparency. when i think of other issues when it comes to surveillance or privacy, i think of one bill we are quite interested in. it is a bill that senator hatch and heller have advanced. it deals with law enforcement access to data. the congressmen on the house side advanced a same, similar type of bill. how does the government collect information? what rights do we have? one of the thing's -- one of the things that takes place with the act is the collection of information what it comes to data. when the data may be overseas and what is the process to be able to get that data, particularly when you have computing.

data that could be in a center in ireland, for example. the individual may be domiciled. what is the process? maybe because you are able to download it, what is the process to access that data? what is the due process? peter: that leads into the tech issue. fred: one of the things that took place that was really great to see, all of the tech companies come together. it wasn't just microsoft, it was apple. it was yahoo!. it was twitter. all having the same views and all advocating together working with civil liberties groups and many others. in advocating for the usa freddie mac and the work of chairman goodlatte. of course the president signed it and was supportive of the usa

freedom act. you need to have a strong interpretation of the laws and not weaken them. everybody is sensitive to national security needs and issues. you have to have trust. peter: how often do you work with the other big five, big six tech companies? fred: we work on many issues. there are some things we can't compete with on the business side. one of the best practices has been usa freedom act. it is good when we all work together. i think another one with similar support is when it comes to trade. today there is a markup on the house side of the innovation act, one that is important when

it comes to litigation reform on the passage of the piece. in the next couple of days, maybe tomorrow, we are looking at tpa and taa and customs on the trade side. issues that are very important technology companies, because we are all multinational companies doing business all over the world and making sure there are trade agreements respect to i.t. and the rule of law. if those -- so those are some of the issues that we are working together. when it comes to public policy issues there really isn't any major differences when it comes to it. the truth of the matter is we work a lot together. not just with the companies. it is important to work with startups. you look at the tech ecosystem and you want to foster and

nurture and make sure startups are doing well. i think of it as one of our main goals is the freedom to innovate. make sure there are no barriers to be able to continue to innovate. we are fortunate where there is a lot of innovation. so many companies are doing so many neat things. peter: this washington understand what you all do? fred: washington is very interested. i also think it is an issue that it is not a erick wright issue, it is not a republican issue -- it is not a democrat issue, it is not a republican issue. for us at microsoft we find a receptive audience. at least two out of three branches of government, there definitely has been an interest. one of our main goals is what our applicant does to educate

and share. that our applicant does to educate and share. have members of congress and staffers, thought leaders, trade association leaders. look at the 12 different displays and to see the power of the cloud, the power of data and visualization of the data and predictive analytics. we have a skype translator and many other neat projects that you will have an opportunity to see that make a difference in people's lives in society. peter: have you increased it here in washington, have you found a need?

fred: our government affairs offices here in d.c.. we also have a public sector off out in virginia. we have another office in chevy chase, maryland. it is right there on the border. peter: you mentioned patent reform. what would you like to see have done on patent reform? fred: one of the most important things is the litigation reform that will take place. it could be a big deterrence so far as we are looking to bring patent controls. i also think venues is an important aspect. i think those are two specific things in mind.

the sausage making an legislative process that takes place gets started. modernizing the litigation reform, dealing with venue issues. i think they are two things that are very important. thank you, peter. peter: now we want to introduce you to microsoft's jeannette wing, she's in charge of microsoft research. what do you do? jeannette: the labs to open basic research. by open i mean we public own -- we publish openly.

we are open with what we do in our research. the public knows what we do. by basic research we use foundational research, research that is bold, envisioned, long-term. hopefully that will lead to new innovations that lead to new products and services. peter: can you give us an example of open research and basic research? jeannette: what we're doing in this particular text there is project premonition -- particular tech fair is project from edition -- project premonition. the application of project premonition is actually to collect mosquitoes that have been people -- that have written people -- that have bitten peop le and determine what kind of

viruses and diseases mayb be around through taking blood samples from the mosquitoes and figuring out the genetic code of the constituents of their blood. that is the application of collecting mosquitoes. the mosquitoes are going to be collected through drones. these drones are flying around with mosquito traps. we are going to use the cloud to figure out the genetic makeup of the blood. in fact, there is some science behind all this drone-carrying mosquito-collectiong equipment. jones or just an example of the systems. when drones are flying around in the air you don't want drones

to bump into people or collide into buildings and died. we want them to be safe, we want them to do no harm to people. the underlying science behind this particular project is actually to verify that these drones are safe and secure. the application is pretty are out, collecting mosquitoes and figuring out what their blood really contains. the underlying science is where the real challenge is. peter: where does an idea like that come from? jeannette: that is a great part of microsoft research and other -- and any other research organization. our general strategy is to hire great people and let them be creative, let them think out-of-the-box, let them do great things.

in particular, this one project i mentioned is really the brainchild of one of our researchers who really thought about the mosquito application knowing all well about drones and the challenge of verification. and he put together a team of people within microsoft research, but also with academic partners to come together with a multidisciplinary team over years. we have people building mosquito traps all the way to people who know how to verify embedded systems in computers. peter: you are an expert at computational --

jeannette: -- an expert at competition of thinking. jeannette: it is the use of computer science to solve problems and understand human behavior. what that really means is it is a new way of tackling complex problems from the lens of a computer scientist. and why that is different and important is first of all in mathematics we already have ways to solve problems. in computers we can make those solutions, alive by executing the solution. we can operationalize a lot of the distraction. with the power of computing we can tackle problems to a scale beyond what a human being can do. computational thinking is

understanding how to harness the power of computing to solve really complex problems. peter: described microsoft labs to us. it is a building where everyone gathers at 9 a.m. in the morning? jeannette: we have a lab in one building at the redmond campus. we also have a lab at new york city, a lab in cambridge massachusetts, a lab in cambridge england, a lab in bangalore india, and a sister lab in beijing china. all of us do basic research in computer science and other disciplines. peter: what are you most excited about in the pipeline?

jeannette: one of the most exciting thing is happening now which you can see in some of the demos today is how we are tackling the big ai problem the ability for machines to actually solve problems that humans are good at solving. humans are really good at vision. they can look at a picture and tell you immediately what are the objects in that picture. machines had a hard time doing that, but we have been making some breakthroughs in machine learning and artificial attend luncheon's -- and artificial intelligence. a we are going to see this in speech, vision, natural language processing, and all sorts of tasks that humans are good at.

what we can do in computing is put all of these things together. by putting all of these things together with integrative ai, then we can have a single machine actually start making -- start creating the full capacity of human intelligence. peter: are you separate kind of a think tank? jeannette: we are part of one microsoft, we are part of the company. we are an organization that is part of many parts and microsoft. you can think of us as the think tank of the company. i joined microsoft research only two and a half years ago. before that i was a professor at carnegie mellon university over 27 years.

i also served at the national science foundation. the national science foundation is the foundation for the united states that funds recent research. it is the primary source of funding for academic research. in the ecosystem of government, industry, and academia, the national science foundation plays an incredibly important role for the academic research enterprise, not just in doing research but in producing the next generation of scientists and engineers. peter: carnegie mellon has been in the news for their technological break rooms. jeannette: carnegie mellon is

the number one computer science department in the country, in the world i would say. it continues to have technological innovations especially in court computer sciences. it is known for his work in robotics. there were recent challenges that carnegie mellon has already -- had already come in -- carnegie mellon has always come in. carnegie mellon is especially strong in computer science. peter: we have been talking with jeannette wing, corporate vice president of microsoft research. she got her bachelors, masters, and doctors at m.i.t.. thank you. "the communicators" is at the washington office for microsoft. we are visiting their tech center. mike zyskowski is here.

what do you do here? mike: we do applications of research technologies are proof of concept and demonstration purposes and also doing things to scale to see how they's bailout for application in different scenarios. peter: what is that mean practically? mike: we love to explore different ways technology can improve the world. peter: what is your background? mike: varied. i am a classically trained aerospace engineer and did a lot of computer science projects in that space, real-time situation -- real-time simulation, production technologies, which then allow me to a radiant microsoft and microsoft -- allow me to foray into microsoft and microsoft research. peter: we have a demonstration.

what are you demonstrating? mike: this is wind flow. the premise of this research project is around what we were able to do with data that is freely available in the environment today. we have noticed there are a lot aircraft flying around the united states that could be considered sensitive. they are providing information that is relatively freely available, provided by the faa. there are companies who use that information to provide information to the community about what airplanes are doing. we decided to take that information and see if we could use that to help us predict a more accurate forecast. what the wind is doing in terms of speed of direction and various altitudes on the earth.

peter: what is the potential advantage? mike: understanding atmospherics is a very important element to a lot of different things. imagine being able to predict where wildfires are going to go next. in our case, we are interested in predicting where a more optimal path through the sky might be that could save them fuel costs and would help the environment. it comes down to a triangle. what we have is an airplane positioned in space that is real. we know exactly where it is. this is the information we get from the aircraft itself. what we want to know is where the wind velocity and wind direction is.

the third leg of the triangle is a neat trick to determine. this is where machine learning comes into play. two health merge on a solution that wouldn't otherwise be given with the data set you have. we were able to constrain this problem in two different ways. they tell the air traffic control systems where they inc. they are going to go. they follow a flight plan. that helps us constrain the problems with an initial set of conditions. we are able to provide a similarity distribution that an aircraft is flying near another aircraft and probably experiencing the same wind. by using that assumption they were able to distribute that. it is a distribution model from the aircraft.

in this particular area this is probably what they are doing. by applying that across a broad set of aircraft and broad set of surface area, we have come up with a prediction model that looks to gives us an order of magnitude better of understanding of the wins -- of the wins than what the entire airline is providing. peter: is this a brainstorm of microsoft research? mike: it was done by one or two people. my collaborator was a well-known machine learning expert. he and i are both pilots. we understood problem space very well and we also understood the algorithms and engineering problems. by marrying those two things together we were able to come up with this unique approach to

solving it, which is pretty simple in its concept that pretty powerful in its application. peter: is it beneficial to have boeing headquarters nearby? mike: we have not talked with boeing about this. they have expressed interest. we see this broad that we see this benefiting the entire industry across. our lost their benefit. peter: a big monitor here, what do you want to show us on the monitor? mike: ok, so i'm using a program here called roadblock telescope. we are looking at a 3-d model of the solar system. we decided to do an experiment to see if our algorithm actually worked. these arrows depict what the

current forecast says it was for that particular day. peter: that wind is directional up to the northeast. mike: generally northeast and direction. these little purple guys are under the airplanes. what they might predict is this a very nonuniform distribution of this aircraft. it is a high density of airplanes. write about here. one of the things we weren't sure was if this would actually be enough information to have a broader model. we created a new prediction that

gave us this sort of distribution. if you look they are a little bit more northerly than northeasterly. if i overlaid them you might be able to tell the difference. in general we did have something that was different. then what we did is we set giving these two different predictions if we let a balloon go in the air where would it go? our prediction forecast would take this trajectory. given the northeasterly flow this is where it would land. then we use our new forecast to predict the endpoint. this is about a 100 miles different. a relatively strong difference in magnitude. when the balloon was launched and actually landed, the little red flag there shows where it landed.

so we feel really good about our prediction. it was an order of magnitude more accurate than what it would have been if we used the earlier predicted wins. -- predicted winsds. we have taken this algorithm and provided early tesh provided an early version of surface -- and provided an early version of this service. how it might benefit those today that count on better understanding of aerospace. peter: microsoft is also a commercial company out to make a profit. where is the business model? mike: we believe strongly in basic research. our mission is really just to try things without a business model. we are taking