An interview with Gordon Bell
Contents
1. u y MaAIAADZUI 43 m interview Sa Gordon Bell has been a major computer industry pundit since the 1960s Known for his role as project leader for the VAX Digital Equipment Corporation s famous mini computer his wide expertise has influenced the design of many other products at Digital Encore Ardent and a score of other companies He recently became a Senior Researcher at Microsoft Corporation Bell has also been a Professor of Computer Sci ence and Electrical Engineering at Carnegie Mellon University and influenced tech nology policy as the first Assistant Director of the National Science Foundation s Computing Directorate He led the National Research and Education Network panel that became the NII GII and was an author of the original High Performance Com puter and Communications Initiative Recognized for his contributions by the many honors he has received Bell is an ACM Fellow and was awarded the 1991 National Medal of Technology for his continuing intellectual and industrial achievements in the field of computer design and for his lead ing role in establishing computers that serve as a significant tool for engineering science and industry In his book High Tech Ventures The Guide to Entrepreneurial Success Addison Wesley 1991 Bell describes the Bell Mason Diagnostic for analyzing new ventures an expert system for ven ture development to startups entrepreneurial ventures investors and governments2. Is your Microsoft group going to look like a university research group No I believe uni versities should do the research and we should support them I want to look at their work and bring it into Microsoft as a prelude to products The place we re probably going to collaborate with more than anybody else is Berkeley Berkeley s got a multimedia research institute that s headed by Larry Rowe So I expect great ideas prototypes and students Berkeley has a fine record of this We officially report to Rick Rashid s office the Vice Presi dent for Research And how many people have you chosen It will be a small group probably not more than half a dozen The group is remote and one of the nice experiments is going to be to see whether or not we can work in the remote fashion First Microsoft in Redmond must 1995 have the hottest most advanced development and engineering groups focused in the telep resence area So my initial task in starting a telepresence lab is to help build those groups Because I feel so strongly that this is an impor tant area I don t want to leave it to just research because at times research is so unpre dictable Thats why I m concentrating on see ing that there will be a videophone channel in the PC this will stimulate more technology than any research activity could How is this research going to be funded and to what level It comes out of the Microsoft research budget What
3. right choice I believe the fundamental problem with telecommunications is their pre occupation concern and regulation that breaks bandwidth up into 64 Kbps voice grade line chunks for tariffs This whole industry sits around trying to maintain the cost and pricing structure based on voice grade lines That s what screws everything up Pacific Bell could offer me a T1 at the same cost as my two POTS lines But the minute they do that they re scared that I ll take the bandwidth slice it up into 64 kilobit chunks use it to get myself 24 phone lines or sell it to my friends for voice and undercut their market And of course there will be people who will I won t but there s a big business for small companies who slice up T1 and ship it across the country and unslice it at the other end Large compa nies do that with their own networks This is occurring with the Internet Phone today We ve got a dichotomy between datacom running at the megabit level telecom for voice which is priced and runs at 64 kilobits and the Internet which is POTS limited at least for home users The Internet is ready to wreak havoc with the telecom industry because the PC will dominate as the telecom switch for POTS As a by product I think we ll see a new instrument that I call the telecomput er a PC plus phone but aimed at Internet access At this point the Internet replaces the telephone network What exactly is the telecomput
4. Bell sits on the boards and technical advisory boards of Ambit Design Cirrus Logic Inc Disk Excellaration Systems Inc Fakespace Inc University Video Communications and CSC s Vanguard Group Computer Science Corpo ration Now that he is with Microsoft Research Bell hopes to pave the way for telep resence by colliding telephones with computers developing instruments that he believes will become the next platform to change our lives Below he describes the evolution of the Internet why telecommunications companies have to withhold band width and the requirements and challenges that must be met for us to be telepresent Bell interactions lt lt october 1995 Heres the gee whiz growth of hosts Karen Frenkel You re in a position of a rare few in our industry because you were involved in the Internet in the earliest days Let s talk about the vision back then and how much of what we have today was anticipated Gordon Bell If you go back to the very beginning of ARPAnet the goal was to have researchers share machines Unfortunately the users of those machines didn t want that to happen at all Sharing was top down from ARPA ARPA also wanted people to be able to collaborate and exchange software Sharing was achieved using file transfers but the killer apps turned out to be email and then bulletin boards Chat groups and computer conferencing that is just synchronous email came into being when more people w
5. interactions october interview a CATV Internet mainframes and SNA dist process client server peer peer Head end Video eze Sources Mainframes or UNIX Servers SNA LAN Networks Couch Potatoes or Clients others evolved this into an industry When AT amp T started they began with a dozen dif ferent Picturephone sites I was one of the first users when DEC was doing the Ethernet deal with Intel and Xerox We needed to meet but didn t have the time to travel so we spent a couple of hours meeting via Picturephone and agreed to go ahead with the deal even though it was our first meeting together At that point I got very intrigued with teleconferencing and convinced DEC to install several sites Each room cost over a quarter of a million dollars because we focused on the quality of service The audio was very good the pictures were good it had lots of cameras and ways to transmit information it was a high cost solution Today companies sell 50 000 products So since AT amp T was doing this fairly early on then does that mean you don t include them with the telcos you find so backwards and were flaming about That was just a service they introduced that no doubt lost money It wasnt a computer service at all It was an experiment to try and sell band width It enabled and encouraged others to start successful businesses and AT amp T could have been the pio neer in selling video pho
6. and takes a chip that would ultimately go on the mother board Who s doing the chip A company that I know but can t disclose There are probably several companies Mechanisms How Synchronous The Who What and How of Telepresence Asynchronous person computer 1 1 personal communication 2 site site conferencing n site conferencing 1 p broadcasts computer management distributed groups with gt 2 10 lt 100 Group Interaction Who interactions october When can you talk about it It will be out by the end of the year But the breakthrough is that we are going to be able to have a nice videophone using just POTS The quality is better than the videophones or picture phones that we see today using ISDN Who do you think wants this Well first and foremost I m a technologist who uses business insight I trust my intuition since I ve been right before even though the world doesn t always get there as fast as I d like I want this and I want this now And from the little mar ket research Ive done there are others like me who prefer to have face to face conversations or face to face collaboration And that drives me into what I want to work on telepresence The greatest inhibitor to telepresence is the fact that we don t all have the ability to see each other The only way to achieve ubiquitous telepresence is to provide it to everyone at very low cost T
7. age l Emultek Corp 1 203 891 8495 cover II Georgia Tech gru search cc gatech edu page 30 Mit Press 1 800 356 0343 page 46 Multimedia 95 1 80 0 5 24 7857 page 47 To advertise in interactions contact Walter Andrzejewski ACM 1515 Broadway 17th Floor New York NY 10036 1 212 626 0685 Fax 212 869 0481 Walter ACMVM bitnet 191915
8. ber about 9 000 scientists I believe were using the centers and the only support they had from the scientific community came because they were free since the money for them came from someone else s budget What all the scientists engineers and educators wanted was their own computers anda reli able and fast network not linking to a time sharing system They wanted communication i n trehet On ss ale Ove t OEE 1 Sey interview aes among themselves They couldn t have cared less about dialing into a timesharing system But this was 1987 so they already had email But they wanted more capacity and reliability because at the time email was croaking It was taking days to get mail through the system and not everyone could access it It was croaking even for just 9 000 supercom puter users and the rest of the computer world Well no You had many academics on the system industrial researchers wanted to be connected and the 56 kilobit links were just not enough When you remote anything response time becomes a network and band width question as much as a human interface question That s why I spend much of my time worrying about the network I see the network as the critical limiter for ubiquitous computing whether it s a cost issue a bandwidth issue a symmetry issue a Washington is doing it wrong issue or the telcos have a damn monopoly on the last mile issue and aren t doing it at all issue You might
9. d RealAudio is providing Internet audio pages How If it takes so much time to load real time audio This runs continuously It isn t loaded before it s played back This is lower quality audio at 14 4 kilobits per second It s AM radio quality Yes These apps includ ing video telephony are enabled by today s capacity is only doubling every two years why the discrepancy interview ai emonstrated laboratory communication data rates are growing a per year D trated laboratory tion data rat g g at 62 per y just like Moores Law for semiconductors or magnetic disk density But installed its not changing prices or introducing services to soak up capacity But more importantly as sub scribers have we gotten any more bits from plain old telephone service in decades Internet 2 0 and just POTS Plain Old Tele phone Service Now Internet 3 0 is where we solve the last mile problem where we ve got symmetry to the home and adequate band width so that we can do complete video tele phony However for the near term I think when youre home we re stuck with POTS and compression to give say 50 Kbps and then you use Internet down line loads from cable TV at Mbps speeds This costs nothing except a cheap modem that attaches to cable Let s help ISDN die so we can make the investment in Internet 3 0 That s what I want Can you do anything very exciting with just POTS We have to since POTS is the onl
10. er Main frames dominated the generation of computing in 1960 Minis came in 70 PCs and worksta interactions october tions came in the 80s Now the Internet is the new infrastructure of the 90s I see the ubiqui ty of the Internet as the catalyst for the tele com puter the next dominant platform and a significant instrument to change our lives The telecomputer is an Internet browser probably a PC without storage and is exter nally maintained via the Net It has a cam era and optional printer So it has a limited and well defined state and it doesn t require a net work or system support organization that char acterizes us as PC or workstation users Disks filled with arbitrary programs and files are where all the direct and indirect user costs are Incidentally AT amp T announced a 300 Internet instrument but I have no confidence in their ability to manufacture it Their technol ogy seems to be totally in their minds and ads Will the telecomputer will be available in the late 90s Perhaps but that s a little soon By 2002 is better It s because of Internet and the datacom telecom collision This will remove the two barriers that hinder telepresence One is bandwidth and the second is the ubiquity of a low cost instrument I m assuming time will solve both problems Metcalfe s law governs when and whether the telecomputer video phone and telepresence exist What is the ubiquity
11. ere connected using 56 kilobit per second backbone links Phase One was to get Tl for the backbone Phase Two kicked in the late 1980s with 45 megabits per sec ond And Phase Three was to go for fiber providing gigabit links There was no research involved through Phase Two Phase Three research was to deal with higher speeds that created the Bay Area Gigabit Network and other test beds But in reality what s created the 1995 Inter net craze what I call Internet 2 0 was the World Wide Web and Mosaic viewers That s again where serendipity comes in It occurred because there s been the factor of three plus orders of magnitude increase in bandwidth going from Phase Zero 56 kilobits to 45 and 155 megabits per second links Can you talk about the interesting state ment you made in your Internet World 95 speech about selling this in Washington using supercomputers and supercomputer centers When l went to NSF to start up the Computing Directorate networking was under the Supercomputing Center Division The first thing I did was to hire Steve Wolfe to head a new division on networking and remove it from the supercomputing division We phased in the NREN plan in broad terms including talking about supercomput ers But most of us thought the supercomput er connection was a stupid bogus idea Why was it a stupid bogus idea Because there were hardly any users of the timeshar ing Supercomputer Centers Only a small num
12. ere online A theme that those of us who have a vision or create technology should always observe is that how something finally gets used has little bearing on what we anticipated when we began This is the serendipity that we hope for when funding research Let me claim that the Growth in thousands of Internet Hosts 4745 growth from 1985 to 1995 5000 4500 4000 3500 3000 2500 2000 1500 1000 500 graph Thats 4 8 million hosts in 0 Data from OpenMarket Net s apps have been constant from 72 until 90 even though use has doubled every year Can I just stop you for one second You were saying there was reluctance to share comput ers back then Now that seems to be all everyone wants to do Why was there reluc tance and what happened in 1990 With ARPAnet people were using timesharing com puters and they were all overloaded The last thing anyone wanted was another user on their system With distributed computing everyone had access Now the challenge is to get them tied together and sharing again The big transition in network apps occurred when the NSF Backbone came into existence to increase bandwidth This came with our National Research and Education Network NREN plan that is now called NI and GII National and Global Information Infrastructure respectively In 1992 the perception was that the 20 year old Internet had just been born The rate of change of number of users has not changed at all It s
13. f competi you double every year tion by lobbying or buying a competitor There s for 22 years you end competition in long distance communications up with 4 million We but there s no alternative for the last mile We tend to notice things need competition for connecting homes and when they get to be in small offices In large cities there are a few alter the millions natives Some cities or towns could put in their own plants similar to what Palo Alto did You mean from the curb to the home Con necting phones to the 20 000 or so central offices The wiring could support reasonable symmetrical bandwidth so that we could have higher speeds But they would have to become more aggressive about providing this service They would have to bet on Internet or on video telephony or data communications This connection is our biggest limiter today and for the foreseeable future Today the threat to telephony and our alter native for high speed data is cable Of course reliability of service is cable s key issue so in a sense their threat is limited But for connecting to the Internet we can probably live with cable reliability Cable represents a medium for deliv ering one way high speed data to our desktops What bandwidth do you need In 1965 when we were first timesharing terminals were This is my proposal at a meeting in February 1987 where we planned the 10G NREN National Research and Education Network The proposal was t
14. his means it cant use ISDN due to its cost and nonuniform access The same chip can be put in a telephone let s call it a tele phone like device a low cost videophone Just like the PC and Internet brought the power of the mainframe and the network cost effectively Internet phone and phone cont RealAudio and simple graphics Workspace CU SeeMe Mbone and conference Room Video Whiteboard Remote Rover Robot Videophone email file transport SW and doc dist bulletin boards schedule Notes Videomail video fax Video lectures and courses view trol hallways with informal interaction 1 1 videophone calls for problem solving authoring interviews Work formal meeting 1995 to the masses so should telepresence products deliver interactive communications Would you please define telepresence What does it mean to you How would it be differ ent from CUSeeMe with good audio Telep resence is being there without having to or being there while being here I want to explore various apps and especially focus on business I want to virtually be at Microsoft in Redmond or San Francisco while physically staying at home in Silicon Valley I think of a new world like telepresence as having various dimensions The three key dimensions depicting Telepresence for Busi ness is given in my figure The three dimensions are 1 the mecha nism how is telepresence accomplished 2 the ap
15. iring and enable it to act as a LAN because wiring an existing house with Ethernet or ATM is an expensive task Can you elaborate on the reference you just made to Intel and the 1500 PC They think that PCs are going to remain fixed at the 1500 price level for the next decade More importantly they believe that the PC will have increasing functionality for the next decade And to some extent they re right But in the meantime another device with lesser or fixed functionality is going to emerge and take a sig nificant part of the PC s marketshare Today s PC industry talks like the mainframe guys used to And meanwhile the telecomputer will come in and wipe them out Now is this device what you re going to be working on at Microsoft research What s Microsoft s role What I hope to work on will build on this technology and enable the need for more telepresence Again what are the inhibitors to telepresence Bandwidth cost and then the question is is there a need for it As a good technologist I assume the usual we will build it and they will come Less than half the time that s true In this instance I think I can clearly see products and needs Microsoft has a history of putting software on the dominant platforms They were right What info comes via the post office that could come Internetted catalogues direct mail letters to buy things scr newspapers newsletters magazines personal purchases including g
16. issue and what s Met calfe s Law It s having enough deployed instruments so that everyone is communicat ing via the instrument with one another We saw how faxes became important after every one got them Metcalfe s law is simply that the value number of possible conversations 1995 So who does it affect communication snail mail radio video substitution telestuff medicine science government wor content accessing Intellectual Property dbases reports and papers SW backup catalogues order goods and services recreation games groups mousing around up and down the trees computing and content education of a network increases as the square of the number of nodes Having the videophone in the computer is the enabler This could get us almost 100 mil lion instruments almost instantaneously But on the other hand it s too expensive and the computer is too hard to use and maintain compared to the telephone If I want to fire up a phone on a computer it s a lot more painful than just dialing And you ve got to deal with software different ver sions and other maintenance issues Ultimate ly we need the telecomputer to replace the telephone However it probably wont replace the telephone or computer but it will replace some of their uses since it will become our pri mary communications instrument I suggested this recently to several people and got a reaction from friends at Intel
17. just when it went from 2 million to 4 8 and 16 million that people begin to notice the phenomenon These are the same people that believe that two Steve s invented the PC when in fact the PC had been around for over a decade Yes that s your point about how when we get to the mil lions everyone tends to notice we So you thought of that I guess because once there was more bandwidth available then we saw a change in the If you plot it on a semi log graph thats a 100 per year growth I dont understand the 86 to 88 shift but that was when NSF got serious about networking I was there Thats not Internet as we know it today It was email and file transport Growth in thousands of Internet Hosts 198 per year or 6 per month 10 000 1000 2x year 100 10 1 or T T Gi s we oe P P interactions october use from ARPAnet That s right It was one of the best techno logical plans Ive ever lead It was done as head of a federal interagency task force that was responding to the Gore Bill to propose an information super highway to look at linking the research and education commu nities including linking high schools to supercomputers 1995 The 1987 plan had four phases Phase Zero was to make the current NSFNet ARPAnet and other nets work and inter operate They were all overloaded and the operations were really poor the service was lousy We w
18. nce on PCs is a good platform for experiments and pro vides a base for useful products but I think by 2002 there will be other devices better tar geted for this Recall the mechanism dimen sion in the first figure which is going from audio graphics white boards and control 1 gave a talk to the financial community about computing and there were 200 people online I had pre faxed them overheads and then presented via a phone conference That worked really well Better yet all of us could hold many of our small technical conferences this way using the Mbone the Internet mul ticast backbone that is the Internet s broad cast like system and even get embryonic video conferencing to boot Within three years I expect to be able to do that same thing only better with a telepresence conference system My slides are online and there s video and audio of me they see my over heads and I see them And this is all done using existing POTS conference facilities or Mbone Is that what you re going to start to do at Microsoft No no This is just a capability of telepresence IIl find all sorts of projects For 1995 october interview Ciela thy lae WES DEEL WWW Book Where did I get the data for this graph I made it all up This one shows the growth in pe cae hype vs reality The first curve shows the rise of speculation on how great itll be Infoway A Ve i speculation nfoway politicians telecom presents and fut
19. nes for conferencing They could have leased videophones So they were on the right track They worry about selling bandwidth or offering a service to sell bandwidth not necessarily making the right choice for the right thing to happen or being creative with a new venture 1 293 3270 terminals As I look at it cable is to Internet as main frames SNA are to dis tributed processing As client server is to true peer to peer 3270 terminals are the couch potatoes of the office world unable to interact with the host This is an impor tant table because it shows that about a quarter trillion dol lars of industries are affected by Internet As Internet can begin to deliver more complex media it can com pete with other industries With just graphics the entire book newspaper magazine and printing industries are affected When video becomes avail able everything from TV to video phones will change Internet affects 240 B industries Media Who When Industry Texttwww CERN 89 Graphics NCSA 3 93 print media Audio SoundM 3 93 Telephony gt 3 cos 3 95 long distance Contin Prog Net 495 radio record Video Qtime 3 93 CU SeeMe 10 93 teleph video cnf Mbone 7 192 tv cable film 3D and animation 3 95 Size 153 65 21 book 24 newspaper 48 magazine 21 printing 60 tv 28 cable 21 filmed entertainment 29 from Technologic partners 4 10 95 Why do you suppose that they don t make the
20. ning a building block There s a parallel with the telephone When the telephone was invented it began with a single one to one conversation Now look at all the applications it can do com merce social interaction shopping ordering tickets sex and everything else we talk about but don t understand Telepresence won t be any different to users it will just be more enriching to have pictures and video com bined with audio Recall that about half the population is visually oriented and the other half verbal How will we interact with it What do you think the user interface is going to be to interact with our telecomputer It must be as uncomplicated as a simple telephone anyone from five to ninety five can use it Note that I said a simple telephone the telecomputer can t be as complicated as the new feature rich telephones where you look at every but ton recognize none of the words and you can t even get a dial tone or make a phone call without a user manual or telephony course Well when I turn on my PC what am I going to do Am I going to click an icon on the TV telephone part of my PC First you dont turn on this gadget it s on all the time It s just there and it s as simple as touching a point on the screen with a finger not a mouse and eventually with your voice Am I going to touch that point and then get a dial tone Sure But youd also better be able to speak to it soon too S
21. note a minimum of 30 years of hostili ty in my voice when I talk about networking Every bit per second and every link along the way has been a fight To get more bandwidth To get more band width at a reasonable or any price to get less restrictions to get openness to get it to work The only reason we ve got networks today is because a separate datacom industry developed them using the installed communications wires and fibers to lay protocols over Net working occurred in spite of telecommunica tions not because of it Now when you Say in spite of telecommuni cations do you mean the long distance carri ers Long distance carriers regional operating companies and PT Ts These organizations have stood in the way of data communications for three decades and remain firmly entrenched in the past Why isn t this need obvious to them How could we make them see that they could make money The telecom companies have operated 69 All the graphs go with regulatory agencies for to the right so long that they only and straight up after 25 years understand how to negoti ate tariffs with govern ments And in the world market you have the same problem All the foreign telcos know how to do is So the growth rate is be state suppliers They dont respond to carrots constant and hasn t When they do take a risk such as ISDN it s a changed for 25 years If screw up They respond to the threat o
22. ny applications programming interface is the foundation for the telecomputer What about the cost issue Technology and time solve this Pm predicating that putting videophones in every computer will create the market for such applications Now let s go back to telepresence because in the long run telepresence will develop because we ve experimented on our mainframes Today our mainframes are our PCs and they re as com plicated as any mainframe We ve all become system managers since we spend an enormous amount of time maintaining software to keep PCs running and all of that PCs are truly mainframes that you baby sit back up and maintain I m willing to have one that never fails in my house but all the others should be slaved to it and cant have much state Do you want to tell us more about how Pm forming an opinion right now My desire is that all the computer engineers in the world to take an oath that reliability is their number one goal Our goal is to build computers that always work never fail and never have to be rebooted Performance should be goal num ber two that is the user shouldn t wait That ties in with having a simple consistent no overhead and no metaphors please interface A garbage can or recycle bin is ok Just the thing that interactions magazine concentrates on is what I would eliminate Then if there are any computer resources left the engineers can start working on brand new function
23. o start with 56 kilobits per second and get what we had working get to T1 quickly late 803 and then go to T3 45 megabits in the 1993 timeframe The factor of 1000 really makes a difference With a very fat pipe you get 10M not the ability to send pictures or video to the desktop but rather the abili 1M 1G ty for a lot of people to send a lot of small messages and get them instanta neously This is what makes Internet 2 0 work in 1995 interactions at 100 bits per second We ve grown a factor of 288 in bandwidth over 30 years But ARPAnet went from 56 kilobits to 155 megabits which is a 3 000 gain in about 20 years So the dis crepancy between what can be and what s actually delivered is great And now the phone companies have invest ed in ISDN which is really too little too late It s only 128 kilobits or less finicky hard to install sales and service folks don t understand it and its expensive They could give us T1 for the same price and installation pain because its all labor and overhead Further more it gets in the way of putting in a decent system Can I urge everyone just to say no to ISDN I m ready to take mine out Pm advocating that telcos deliver five to 10 megabits per second to our homes and shoot to make it symmetrical That is I ve got to be able to send back those speeds as well as receive them However for now my next step is going to be to connect to a cable service to get 10 Mbits
24. o it s as simple as picking up the receiver Yes it s got to be that simple You can t spend any more time than youre doing now It can t be any more difficult or frustrating than a simple phone interactions october So is that hard to do Yes engineering sim plicity is always hard it s an art Why Because most engineers would rather create a multitude of irrelevant options building Cadillacs when only Volkswagens are needed But why can t I click on an icon to my modem which is now a 28 Kbit modem Just the issue of sending email to me has been dif ficult when I m home Or if you re not on a good corporate LAN system then you must go through the pain of a dial Clicking on an icon having it dial watching it log on is not acceptable from a 40 seconds of lost time standpoint It must be instantaneous Today I can physically find a number and have a dialer act faster than I can find it on a computer and have a computer dial a modem So now that Microsoft is going to be able to have you click on an icon to get into MSN you would still have to hang around and wait and do are you there and do that whole thing Hopefully that s better than making an Internet interconnection today It s not the clicking of the icon it s waiting for the service that is the issue The modem dialing and handshaking is where most of the time goes I should wait no longer than I wait for a phone call completion
25. oods personal bills personal finances reports transactions personal and formal letters 1295 october The NY Times Fax service in PDF format is really impressive and gives us a glimpse of what could come Bills are another favorite item for elec tronic delivery I ve invented BillFree which Tve asked to trademark Im assum ing CheckFree is going to buy it some day Its the converse of Check Free for bills And faxes are disap pearing thank god We owe it to the world to get rid of faxes Finally personal let ters which are mak ing something of a comeback In a lot of cases Internet brings people closer together because theyre on line already and its easy and natural interview a PERMISSION TO COPY WITHOUT FEE ALL OR PART OF THIS MATERIAL IS GRANTED PROVIDED THAT THE COPIES ARE NOT MADE OR DISTRIBUTED FOR DIRECT COMMERCIAL ADVANTAGE THE ACM COPYRIGHT NOTICE AND THE TITLE OF THE PUBLICATION AND ITS DATE APPEAR AND NOTICE IS GIVEN THAT COPYING IS BY PERMISSION OF THE ASSOCIATION FOR COMPUTING MACHINERY TO Copy OTHERWISE OR PUBLISH REQUIRES A FEE AND OR SPECIFIC PERMISSION ACM 1072 5520 95 1000 3 50 about the PC and they were right about build ing NT to deal with corporate computing I hope they ll be equally aggressive about the Internet and the telecomputer I m a new employee and don t understand their commit ment to these products However I know that TAPI telepho
26. papers on the net Cctutitc t G VECE MOULE Cii Ler VE ULE uel WELCUS TeChities Is Internet 4 gameboy for adults This is my last graph This is the evolution of gamer age years hours of play vs eee ee time for various machines That is if youre 24 years old then you get counted as in hand eye co ordination two times a 12 year old An hour of my time counts a lot Fey ies HEEL intendo Nintendo and Sega are 12 year old boys machines I can t play them well intendo and Sega are 12 year old boys machines I cant play them we ae Segan Computer games may be for girls because theyre head games and the Nintendos are handleye coordination and shooting And then Internet is the final and ultimate game Its gonna cross over ight chtearcat ONDes a o Oetker 1905 Gordon Bell Microsoft Inc gbell microsoft com Karen A Frenkel is a science and technology journal ist and producer She has just com pleted a documen tary Minerva s Machine Women and Computing which will premiere on San Jose s PBS station KTEH Channel 54 on October 23 10 pm PST frenkel acm org now the first thing I want to accomplish is to get video telephony in all of the operating sys tems as a data type therefore into all the prod ucts and especially TAPI so that we can use it as a building block for doing many apps That s not the end but a building block Just like the telephone isn t the end of the world it s the begin
27. per second from the Internet and Td like a separate T1 line for the reverse channel But I thought you said 25 Mbits per second in your talk That s what I want I believe the technology exists to send and receive 25 Mbps on a 4 wire 2 pair service at distances up to 20 thousand feet Five to 10 Mbps is the mini mum although this only gets one channel of high quality video Fortunately swirling around are other providers cable broadcast and satellite as I ve sketched out in a figure I called The Collid ing Worlds of Television Communications Data Communication and the Internet At least four guys are trying to provide video or television and since the cable guys are threat FCCSET NREN Plan 11 1987 A factor of 1000 makes a difference Optical 1988 1990 1992 1994 1996 1998 2000 october 1995 1000 100 10 0 1 0 01 Installed vs laboratory fiber Gbps it just doesn t ring true Laboratory 62 yr Installed_ 43 yr 45 Mb 1980 2000 1990 ened by the telcos they want to enter into the lucrative phone business But an even greater threat to the telcos core business is the Inter net and its ability to carry worldwide long dis tance phone email and video conversations You mean you could carry voice via the Inter net Today Internet is being used for long distance phone calls using only 14 4 Kbps local connections A new service calle
28. plication what is achieved using telepres ence and 3 the group structure who is using telepresence The first is based on technology and the other two are social We engineers tend to concentrate on the mechanism That is what are we providing from a pure channel stand point And what should go from text to video to graphics white boards to sketch on and the control of shared programs and data Then the second dimension is what are we going to do with telepresence Are we simply communicating are we doing inter views or are we creating a virtual hallway to stroll down in order to accomplish management by walking around Are we attempting to design something or solve a problem Or are we conducting a formal meeting run by Robert s Rules of Order These ques tions are answered only when we have enough real telepresence users Research on collaboration doesnt mean anything unless you ve got enough instruments deployed in real world situations The critical social dimension is the struc ture of who s communicating who s collabo rating who s being teleported and how is the teleporting occurring This begins with simple one to one interaction goes through highly distributed groups and finally mob scenes with an unlimited number Two site conferencing including person to person and video conferencing provided by AT amp T s Picturephone Meeting Service PMS in 1978 is the most common PictureTel and
29. s and hopefully not ways of doing the same function just differently If we ever get to the point of using speech then that will cause the interface to change But today a different way of doing something or a bizarre feature is goal number one which creates an increas ingly complex interface performance is number five and reliability is number 10 and consistency is 11 No actually that s not true To the desktop industry including its interactions buyers purchase price appears to be number one That s what they design for How do you feel about Bob I dont know it I hope it adds user value But if it won t do anything new for any of us increases complex ity by being on top of an already too complex environment and gets in the way of apps I doubt if PI like it I sure feel that way about General Magic s cute mac redo Just like a two year old I want to get to my app ASAP Unfortunately the real cost of a computer is a user s time including the time to learn install maintain relearn and attend to its flaws The hardware software cost has remained constant or even increased over the last ten years Hardware now costs nothing and software costs a lot But the cost that s increased is that users are now the system managers and that s costing a minimum of 50 billion a year to the 50 or 100 million computer users in lost time Let s get back to what you re going to do at Microsoft Engineering teleprese
30. s the budget I have no idea understand that there will also be parallel processing research I work with Jim Gray and he s focused on scaleable computing which I m interested in too The idea is to build very large systems with hundreds of PCs The PCs will be quad processors that are the most cost effective platform on the market This structure is the root of upsiz ing which has been dominating the last several years of mainframe mini downsizing trends Also Jim s lab will be a test site for the telepresence work So what s the first problem you re going to tackle Right now I want to have video phones all around my home and the San Fran cisco lab that link to San Francisco and Redmond Jim Gray is our user The first projects are tools for doing video and there are a number of products we can begin to use Like what Intel s Pro Share CUSeeMe and Vivo So you want to improve on those Yes and use them as components for telepresence sys tems Within a few years the platform will be combined with a robot and used to roam and attend meetings in another space instead of being constrained to just exist in cyberspace Once you ve opened your mind to the possi bilities of telepresence it s hard to go back to just the simple telephone interactions october Index to Advertisers ACM CHI 96 1 410 263 5382 cover III Andersen Consulting cover IV CACI 1 619 457 9681 p
31. that the computer would never cost less than 1500 Counter to what they believe a 1500 device is too expensive for ubiquity and worse yet the programmability of the computer is why it s so expensive to use and maintain I m not going to install many more programma ble computers in my house even though I can afford it I did however finish installing Ethernet ports by every phone as a prelude to replacing phones with the telecomputer I have a relatively small house and about a half dozen rooms with phones If all these had computers I wouldn t get a thing done I would just be maintaining the machines and worse yet a system manager would have to live in my garage or I would have to build an addition So what we need is the non pro grammable telecomputer However the home LAN is a great new product and service opportunity It s clear that everyone will need this The need is to be able interactions Most of this is obvious As an avid catalog shopper I see Internet as the ultimate catalog And games and education will change dramatically What about CDs You might ask which CDs can I get rid of personally in favor of network access Which magazines and reports would you rather get electronically Per haps reference materials like Consumer Reports Which do you not need on hand at all I can see the end of some news type magazines that I just want to browse to use existing telephone w
32. urists all making statements about cyberspace Then how great it ll Conferences addiction Genial aan 5 7 l we had an impulse a function that is infinitely high and takes zero time The impulse telecoms and ma was the World Wide Web with hyperlinks followed by the Mosaic viewers futurists Conferences about Internet follows along with growth in browswers And then there x will be lawsuits Since we are all spending hours and hours browsing there will be Infoway Infoway addiction And thats followed by Infoway regulations regulation NACLES about security Privacy and GUCE Cenu Gie E1 This graph shows articles about security privacy and Actual commerce s fraud vs actual commerce We continue to see articles about how you shouldnt use your credit card on Internet Articles about risk and NOT doing Commerce is taking off and none of us worry about sending credit card numbers With encoded browsers its commerce much safer Finally well see organized crime ae Organized crime on Internet JN TCLES GL DEMIS EpeEL WS CUG GLIE PEL SECC FEU Vic Leu Here we are with articles per newspaper vs orders per Orders per second m A P pap T R second Its crossed over with more orders This is major since every newspaper has an article or two on why you shouldn t order and why you shouldn t advertise They Articles per newspaper n want to keep their subscribers They are also working on putting their news
33. y ubiquitous service I m involved in a project that I think will change how we all interact It s the ability to put a videophone in every Television Games world VCR Telephony CDs Gable world gt 97 Cable gt Phone RBOCs ee DB S Telcos Inter active Cable ___ BA Iyterner Wireless Phone o The colliding worlds of television telephony and data communications f Clients a k a computing and Servers Internet Wirel Datacom iveless world L295 interactions october Internet 3 0 PC running Windows PC My goal is to have Microsoft ship a free POTS based videophone with every copy Universal digital dial tone Scaleable symmetrical service 25 Mbps not 0 0144 or 128 Mbps Fungible bits cre e datacom images and video like of Windows within a year or e RealAudio two I really need this in 2d and 3d video order to work on the telep e phone Cots of telephone and table resence that we first wanted to talk about today And will Microsoft deliver it When will it be ready I dont know But I want them to A product should be released by the end of the year Furthermore other companies also need to be shipping videophone only versions This is the big enabler for telepresence because everyone can have one Is this a software product It s software but for the best quality you really need about 200 million operations per second That s well over a Pentium or P6
Download Pdf Manuals
Related Search