1	The IT Innovation Ecosystem Ed Lazowska IT & Public Policy Autumn 2004
2
3
4	Overview of “Tire Tracks Diagram” Shows 19 $1B (or larger) sub-sectors of IT Shows university research (federal funding), industry research (industry or federal funding), product introduction, $1B market Shows flows within sub-sectors, and between sub-sectors Shows a subset of the contributors, for illustrative purposes
5	Key concepts illustrated Every major $1B IT sub-sector bears the stamp of federal research funding Every sub-sector shows a rich interplay between university and industry It’s not a “pipeline” – there’s lots of “back-and-forth” It typically takes 10-15 years from idea to $1B industry There are many research interactions across sub-fields
6	Key concepts not illustrated (but I’ll get to them) Unanticipated results are often as important as anticipated results It’s hard to predict the next “big hit” Research puts ideas in the storehouse for later use University research trains people University and industry research tend to be complementary Visionary and flexible program managers have played a critical role
7	The Internet 1966: First experiments in digital packet switched technology 1968: ARPA issues RFQ for IMPs AT&T says it’ll never work, and even if it does, no one will care 1969: ARPANET inaugurated with 4 hosts Len Kleinrock’s student/programmer Charley Kline attempts remote login from UCLA SDS Sigma 7 to SRI SDS 940 System crashed partway through – thus, the first message on the Internet was “lo”
8
9	"1975:" 1975: ARPANET has 100 hosts 1977: Crufty internetworking demonstration 4-network demonstration of ARPANET, SATNET, Ethernet, and PRnet – from a truck on 101 to England 1980: Design of TCP/IP completed 1983: Conversion to TCP/IP completed Routers allowed full internetworking – “network of networks” Roughly 500 hosts
10	"1988:" 1988: ARPANET becomes NSFNET Regional networks established Backbone speed 56kbps Roughly 100,000 hosts and 200 networks 1989: CNRI interconnects MCImail to the Internet Wise policy choice 1990: Backbone speed increased to 1.5Mbps by IBM and MCI Roughly 250,000 hosts and 1,500 networks Note: There still was “a backbone”!
11	"1992:" 1992: NCSA Mosaic stimulates explosive growth of WWW 1995: Full commercialization, at 45Mbps 6,000,000 hosts, 50,000 networks
12
13
14	Key concepts illustrated Bears the stamp of federal research funding Shows a rich interplay between university and industry Not a “pipeline” – there’s lots of “back-and-forth” 10-15 years from idea to $1B industry
15	(D)ARPA I(P)TO JCR Licklider, 1962-64 Ivan Sutherland, 1964-65 Bob Taylor, 1965-69 Larry Roberts, 1969-73 Al Blue (acting), 1973-74 JCR Licklider, 1974-75 Dave Russell, 1975-79 Bob Kahn, 1979-85 Saul Amarel, 1985-87 Jack Schwartz, 1987-89 Barry Boehm, 1989-91 Steve Squires, 1991-93 John Toole (acting), 1993-94 Howard Frank, 1994-97 David Tennenhouse, 1997-99 Shankar Sastry 1999-01 Kathy McDonald (acting), 2001-02 Ron Brachman, 2002-present
16	IPTO under Bob Kahn, 1979-85 VLSI program Mead-Conway methodology MOSIS (Metal Oxide Silicon Implementation Service) Berkeley Unix Needed Unix with virtual memory for the VLSI program (big designs) and the Image Understanding program (big images) Also a Trojan horse for TCP/IP And a common platform for much systems and application research
17	"SUN workstation" SUN workstation Baskett said no existing workstations could adequately handle VLSI designs (Bechtolsheim’s frame buffer approach was unique) Kahn insisted that it run Berkeley Unix Clear byproducts Sun SGI RISC (MIPS, SPARC) TCP/IP adoption Internet routers (Cisco)
18	Additional key concepts illustrated Many research interactions across sub-fields Graphics, workstations, VLSI, computer architecture, operating systems, and networking were being synergistically advanced!
19
20	"Visionary and flexible program managers..." Visionary and flexible program managers have played a critical role
21	ISAT Study: Impact of AI on DoD August 2004 Co-Chairs: Ed Lazowska Al McLaughlin
22	Study Charter Review impact of AI technology on DoD Major systems enabled by AI technology Significant demonstrations and new capabilities Spin-offs – DoD to civilian “Spin-ons” – civilian to DoD
23
24
25	Dynamic Analysis Replanning Tool (DART)
26	PackBot
27
28	Image Understanding: BCAMS
29	USPS HandWritten Address Interpretation System
30	Image Guided Surgery
31	Commercial Airport Operations
32	Impact of AI on DoD: Observations AI technology is having significant impact on DoD. Metrics include: saving lives: CPOF expediting planning and logistics: DART keeping troops from harm’s way: PackBot large operational cost savings: ASF improved intelligence: TIDES/EARS reduced training costs/manpower: TacAir-Soar more effective surveillance/monitoring: BCAMS
33	"AI yields new capabilities:" AI yields new capabilities: speech recognition: Phraselator automated language translation: TIDES planning: DART decision support: CPOF simulation/training: TacAir-Soar image understanding: BCAMS robotics: PackBot
34	"Some of the specific systems..." Some of the specific systems were quickly engineered in response to DoD/wartime needs – e.g., DART, ACPT, Phraselator All systems were built upon three or more decades of sustained DARPA investments in AI and other technologies technologies, prototypes trained people, synergistic interactions ability for quick reaction response
35	“Ideas in the storehouse” Electronic commerce draws upon: Internet Web browsers Public key cryptography Databases and transaction processing Search
36	Unanticipated results are often as important as anticipated results The development of timesharing in the 1960s (in Tenex, Multics, CalTSS) gave us electronic mail and instant messaging
37	It’s hard to predict the next “big hit” “Tire Tracks Diagram,” 1995 vs. 2003
38
39
40	"In our despondency in 1995" In our despondency in 1995, we failed to foresee … Client/Server computing Entertainment technology Data mining Portable communication World Wide Web Speech recognition Broadband last mile
41
42	Research institutions come in many different shapes and sizes
43	The correlation between high-tech success and top universities is clear Boston: MIT, Harvard Research Triangle Park: Duke, UNC, NC State Austin: University of Texas So. California: UCSD, UCLA, Caltech No. California: Stanford, Berkeley, UCSF Puget Sound region: University of Washington
44	Why? Education Technology attraction Company attraction Innovation (technology creation) Entrepreneurship (company creation) Leadership and intangibles
45	“Competitive advantages” of universities Students Long-term research, not tied to today’s products Inherently multi-disciplinary Neutral meeting ground “Open”
46
47	The nature of industry R&D Entirely appropriately, industry R&D (at least in IT) is heavily focused on D – product and process development Microsoft’s investment in Microsoft Research – unquestionably one of the world’s great IT research enterprises – is nearly unique 30 years ago, IBM, Xerox, and AT&T represented a huge proportion of the “IT pie” Each had a great research laboratory focused more than 18 months out
48	"Today," Today, the “IT pie” is far larger And the industry’s investment in R&D is far greater (all technology companies do R&D) But of the newer companies – the ones that have grown the pie – Microsoft stands almost alone in its investment in fundamental research Dell? Oracle? Cisco? Nada! Microsoft began this investment in 1991 – when it was a far-from-dominant $1B company – Microsoft (particularly Gates and Myhrvold) should receive enormous credit for taking this step
49	"So," So, how much of Microsoft’s $7B in R&D (>15% of revenues) is “research”? Microsoft Research – the part of Microsoft’s R&D enterprise that’s looking more than 18 months ahead – is about 700 heads, <5% of this total This is extraordinary by the standards of other companies … but don’t confuse Microsoft’s R&D expenditures – much less the rest of the industry’s R&D expenditures – with an investment in fundamental research!
50	"Why might companies be reluctant..." Why might companies be reluctant to invest in R&D that looks ahead more than one product cycle?
51	"Established companies generally don’t capitalize..." Established companies generally don’t capitalize on innovations The culprit is good management (and shareholder behavior), not bad management Evolutionary vs. disruptive innovation “It’s a zero billion dollar market”
52	"Example:" Example: RISC (Reduced Instruction Set Computer) processors
53	Federal support of science Old history NIH (National Institutes of Health) as a small unit of the Public Health Service since the late 1800s Army Ballistic Missile Laboratory supported ENIAC at Penn 1945: Vannevar Bush, Science: The Endless Frontier 1947: ONR (Office of Naval Research) established
54	"1950:" 1950: NSF (National Science Foundation) established Bush had advocated one agency, but got 3+ Civilian natural and physical sciences: NSF Civilian life sciences: NIH Defense sciences: ONR, etc. 1957: Sputnik 1958: (D)ARPA ((Defense) Advanced Research Projects Agency) established 1958: ARPA / 1972: DARPA / 1993: ARPA / 1996: DARPA
55	"1962:" 1962: I(P)TO (Information (Processing) Techniques/Technology Office) established within DARPA More on DARPA IPTO shortly
56	Recent history in IT specifically 1985-86: NSF Supercomputer Centers established 1986: NSF CISE Directorate established HPC (High Performance Computing) Act of 1991 (the “Al Gore created the Internet” Act) Multi-agency coordination Presidential advisory committee 1992: NCO/HPCC (National Coordination Office for High Performance Computing & Communication) established
57	"1997:" 1997: PITAC (President’s Information Technology Advisory Committee) established 1998: PITAC interim report 1999: PITAC final report
58	Characterizing research “Fundamental research” and “application-motivated research” are compatible
59	Traditional view
60	Alternative view
61	Trends in federal research funding How has the federal research investment (basic and applied) fared over the years? It’s increasing significantly, in constant dollars – a factor of more than 2 in less than 20 years [NSF data analyzed by AAAS, 2003]
62
63	"What’s the balance of the..." What’s the balance of the nation’s research portfolio? A dramatic shift towards the biomedical sciences in the past 20 years, accelerating in the past 5 years Biomedical research is important But it relies critically on advances in other fields, such as physics, engineering, and information technology There is broad agreement that the nation’s R&D portfolio has become unbalanced [NSF data analyzed by AAAS, 2003]
64
65	"How does support for computing..." How does support for computing research stack up against the recommendations of PITAC? It’s fallen off the train
66
67	"Research investments are closely linked..." Research investments are closely linked to creation of the nation’s Science & Technology workforce So, in what fields are the nation’s Science & Technology jobs? [John Sargent, U.S. Department of Commerce, 2004] [First chart: employment growth, 1996-2000] [Second chart: projected employment growth, 2002-2012] [Third chart: total projected job openings, 2002-2012] [Fourth chart: projected degree production vs. projected job openings, 2002-2012, annualized]
68	Recent Occupational Growth Growth in Numbers
69	IT, Science and Engineering Occupational Projections, 2002-2012 Employment Growth: Numbers
70	IT, Science and Engineering Occupational Projections, 2002-2012 Total Job Openings
71	The Market Perspective Degree Production vs. Projected Job Openings
72	NSF CISE Cyber Trust program FY04 awards announced 9/21/2004 Funded 8.2% of proposals 32 of 390 proposals 2 of 25 Center proposals 12 of 135 Team proposals 18 of 230 Small Group proposals Awarded 6.2% of requested funds $31.5M of $510M
73	Department of Homeland Security FY05 budget request $1,069M Science & Technology budget request $17.8M for Cyber Security – 1.67% One is led to conclude that DHS simply does not care about Cyber Security (Also, 90% of the DHS S&T budget goes to Development/Deployment rather than Research – fails to prepare us for the future)
74	DARPA Cyber Security research DARPA’s new Cyber Security research programs have been classified Let’s assume there are good reasons. There still are two major negative consequences: Many of the nation’s leading cyber security researchers (namely, those at universities) are excluded from participation The results may not rapidly impact commercial networks and systems – upon which much of the government, and much of the nation’s critical infrastructure, rely
75	21^st century vs. 19^th century industries In 2003, the US government spent: $5B on basic research in the physical science and engineering $25B on direct agricultural subsidies
76	"Recap:" Recap: About $55B of the nation’s $2,319B budget goes to basic and applied research More than half of this goes to the life sciences (IT is less than 4%) IT research funding is actually decreasing More than 80% of the employment growth in all of S&T in the next decade will be in IT – and more than 70% of all job openings (including those due to retirements) Recent news provides little encouragement!
77	"“What the hell were..." “What the hell were you thinking?”
78	The federal budget: How the sausage is made Most of the budget is mandatory Half of what’s discretionary is defense The rest involves dozens of agencies They are grouped irrationally, and tradeoffs must be made within those groups “Balancing the budget” is a foreign concept
79
80
81
82
83
84
85	IT, economic growth, and productivity “Advances in information technology are changing our lives, driving our economy, and transforming the conduct of science.” Computing Research Association
86
87	Productivity In the US, our wages are high, so our productivity needs to be high, or we’re SOL A US worker who is twice as productive can compete with a foreign worker who makes half as much
88	The productivity paradox We all “believe” that IT increases productivity There have been continuous investments in the application of IT for more than 40 years But there were at most very modest signs of any increase in organizational productivity from 1975-1995 “Computers show up everywhere except in the productivity statistics” – Robert Solow, Nobel prize winning Economist, 1987
89	Between 1995 and 2000 A huge surge in economic growth, driven by dramatic increases in productivity (double the average pace of the preceding 25 years), attributed almost entirely to IT! “We are now living through a pivotal period in American economic history … It is the growing use of information technology that makes the current period unique.” Alan Greenspan, Chairman of the Fed, 2000
90	So, what happened? Not clear the economic data was capturing the right things Also, it was measuring entire industries, not individual firms (accounting for quality differences) Changes in processes, stimulated by changes in technology, take time to show impact
91	Impact of IT on the economy, 2004 “We have completed our program of attributing US economic growth to its sources at the industry level. … Our first conclusion is that many of the concepts used in earlier industry-level growth accounting should be replaced … investments in information technology and higher education stand out as the most important sources of growth at both industry and economy-wide levels … the restructuring of the American economy in response to the progress of information technology has been massive and continuous …” Dale W. Jorgenson, Harvard, Mun S. Ho, Resources for the Future, and Kevin J. Stiroh, Federal Reserve Bank of NY, “Growth of US Industries and Investments in Information Technology and Higher Education”
92
93
94	Education for the “innovation economy” Once upon a time, the “content” of the goods we produced was largely physical
95	"Then we transitioned to goods..." Then we transitioned to goods whose “content” was a balance of physical and intellectual
96	"In the “innovation economy..." In the “innovation economy,” the content of goods is almost entirely intellectual rather than physical
97	"Every state consumes “innovation..." Every state consumes “innovation economy” goods Information technology, biotechnology, telecommunications, … We produce these goods! Over the past 20 years, the Puget Sound region has had the fastest pro-rata growth in the nation in the “high tech services” sector
98	What kind of education is needed to produce “innovation economy” goods? National and regional studies conclude the 3/4^ths of the jobs in software require a Bachelors degree or greater (and it’s highly competitive among those with this credential!)
99
100	"In Washington State:" In Washington State: We rank 48th out of the 50 states in the participation rate in public 4-year higher education (1997 federal data presented by OFM) We rank 41^st in upper-division enrollment – “Bachelors degree granting capacity” – still in the bottom 20% of states We rank 4^th in community college participation Washington’s public higher education system is structured for a manufacturing economy, not an innovation economy!
101	"On a per capita basis" On a per capita basis, Washington ranks 32^nd among the states in the number of Bachelors degrees granted by all colleges and universities, public and private, and 35^th in the percentage of our Bachelors degrees that are granted in science and engineering (1997-98 data, Dept. of Ed.) Private institutions are not filling the gap
102	"We rank 43^rd in graduate..." We rank 43^rd in graduate and professional participation rate at public institutions (1997 federal data presented by OFM) We rank 41^st in the number of students pursuing graduate degrees in science and engineering at all institutions, public and private (1999 data, NSF) At the graduate level, things are just as grim
103	"We rank 5^th in the..." We rank 5^th in the nation in the percentage of our workforce with a recent Bachelors degree in science or engineering, and 6^th in the percentage of our workforce with a recent Masters degree in science or engineering (1999 data, NSF; “recent degree” = 1990-98) We are creating the jobs – and we are importing young people from elsewhere to fill them!
104	"UW’s state funding per student..." UW’s state funding per student is ~25% below the average of its Olympia-defined “peers” (22% behind 24 HECB peers, 26% behind 8 OFM peers) (1999-2000 data, IPEDS) In 1976, Washington spent $14.35 on higher education per $1,000 of personal income; by 2001, that number had dropped by nearly a factor of two – to $7.65 (Postsecondary Educational Opportunity #115) We under-fund the relatively few student places we have. And it’s getting worse
105
106	"Washington ranks 46^th out of..." Washington ranks 46^th out of the 50 states in state support for research This is the relatively modest “seed corn” from which large-scale federally-funded research programs grow
107	"Washington is all geared up..." Washington is all geared up to fight the last war!
108	More broadly (some data is not current, but nothing much has changed) Bachelors degrees, nationwide, 1997: 222,000 in business 125,000 in the social sciences 105,000 in education
109	"China granted only 1/4 as..." China granted only 1/4 as many Bachelors degrees in 1997 as did the US (325,000 vs. 1.2M)
110	"Proportion of Bachelors degrees that..." Proportion of Bachelors degrees that are in engineering: US: 4% United Kingdom: 12% China: 40%
111	"What’s the fastest-growing undergraduate major..." What’s the fastest-growing undergraduate major in America today?
112
113	"At the doctoral level (..." At the doctoral level (also 1997): 40,000 J.D.’s