1 Networks of Tinkerers: a model of open-source innovation Peter B. Meyer Office of Productivity and Technology, U.S. Bureau of Labor Statistics At OSS 2007, June 14, 2007 All views expressed in this paper are those of the author and do not necessarily reflect the views or policies of the U.S. Bureau of Labor Statistics.
2 Open-source technologies Goal here: make micro model of individuals creating open-source technology Defn: Advanced through openly-shared designs Open source software Early microcomputers at Homebrew Club, 1975 Development of first airplanes Clearly documented, slowly over decades Since 1860s: aeronautical journals and conferences 1894 – Octave Chanute’s overview book 1903 – Wrights fly famous powered glider 1910 – an industry exists
3 Chanute’s 1894 book Progress in Flying Machines cites 200 experimenters, from all over Experimenter / group Pages location (background) Maxim33 Britain (US) Lilienthal31 Germany Penaud22 France Mouillard21 Algeria, Egypt (Fr) Hargrave19 Australia (Br) Moy19 Britain Le Bris17 France Langley16 US Wenham15 Britain Phillips14 Britain Later technological histories treat these top several as central. Their findings were mostly public. Patents in the US and Germany look different – there are many patentees, with one each. Chanute says many are “worthless”. Hargrave says it’s too soon to patent or keep secrets.
4 Hargrave box kites experiment, 1894 He thinks all aerial navigation worked should be published and nothing patented till something really works.
5 For decades Otto Lilienthal studied birds and experimented on shapes in wind Published Birdflight as the Basis of Aviation, 1889 Lilienthal curved wing experiments
6 Motivations of experimenters Would like to fly Curiosity, interest in the problem Prestige, recognition Belief in making world a better place Make one nation safer Nobody refers to expected profits “... A desire takes possession of man. He longs to soar upward and to glide, free as the bird...” -- Otto Lilienthal 1889 “The glory of a great discovery or an invention which is destined to benefit humanity [seemed] dazzling.... Enthusiasm seized [us] at an early age.” - Gustav Lilienthal
7 For micro model Assume there are motivated tinkerers Assume total technological uncertainty no identifiable feasible market/product now so no competition Can show The tinkerers would share information
8
9
10
11
12 Investment to gain more from network Programmers specialize on aspects of a project avoids duplication; develops focused skills and tools And standardize code interfaces (or design, or part) so some progress by other tinkerers will snap right in In model: Tinkerers receive fraction f of progress by others Suppose for cost c s, could raise that inflow to f 2 > f
13 Specialization and standardization That’s worth doing if: A player benefits more from this if, ceteris paribus: other tinkerers produce a large flow of innovations p2; gain in useful innovations from the others (f 2 -f) is large. cost c s is small; For β=.95, p=.07, f2=.55, f=.5, payback is 1.33; worth usual investment. Specialization and standardization are natural in tinkerers’ network. Tinkerers in model would be willing to pay. Don’t need market processes to explain this behavior
14 Role for author / moderator / evangelist Chanute corresponded with, visited, introduced experimenters, and published book In model: A tinkerer’s best opportunity for a better program may be evangelism – give up p 1 for p 2 To welcome future tinkerers who could generate progress To avoid duplicate efforts, thru standards and specialization a uthors/evangelists are another kind of specialist tinkerer Octave Chanute, 1894: “The writer’s object in preparing these articles was threefold: 1. To satisfy himself whether... men might reasonably hope eventually to fly To save... effort on the part of experimenters trying again devices which have already failed. 3. To... render it less chimerical... to experiment with a flying machine....” Analogously: Lilienthal’s public demonstrations; Felsenstein at Homebrew; open source programmers Stallman, Torvalds, etc.
15 Wright brothers left the network to start a business Wrights were of open-source type. "I am an enthusiast... I wish to... if possible add my mite to help on the future worker who will attain final success." -- Wilbur Wright, 1899 "At the beginning we had no thought of recovering what we were expending, which was not great..." -- Orville Wright, How We Invented the Airplane, [1953] p. 87 They published, spoke at meetings, had visits from Chanute and others, showed everything. In late 1902 and subsequently they were more secretive. Why? Probably their patented wings from their wind tunnel experiments. Analogously, Apple founders left Homebrew Club
16 Intellectual property and secrecy In each episode (airplanes, computers, open source software) many people want to avoid intellectual property and/or secrecy Hargrave thought all aerial navigation work should be published and nothing patented till something really worked. Chanute wanted to get all information out in the open Analogously: Stallman, Fogel, other open source programmers In model: intellectual property payments for sharing results of experiments would just introduce noise and friction on net. Incentives aren’t needed, and don’t help, and there are no profits to split.
17 Entrepreneurial exits in model Suppose a tinkerer envsions how to make a profit from project A worth more than the present value of staying in the tinkerers’ network. Then tinkerer can exit network agreement conducts directed R&D stops listening to network becomes an entrepreneur
18 Modeling exit to startup firm Suppose each player tries to see through the technological uncertainty and imagine selling a product based on A. At the start of each turn with probabilities π 0 and π 1 each player envisions how to do this, and exits to take a large expected utility payoff of M for quitting the network and starting a firm. Anticipating this could happen they agree on a fee x to exit the network Then the payoff is something like: Previous results hold, while the tinkerer’s network exists But incentives change once he makes the quantum leap to being an entrepreneur.
19 Alternative models of invention (1) Network: a population of agents with i nterest in a problem (a 0 ), worthwhile opportunities (p), information flows between them (f) experimentation and socially constructed “progress” No pool of information, or incentive structure, or technical measure of improvement. (2) Race to be first (space race; genome project) (3) Collective invention (Allen, 1983) but those are (a) firms, (b) not paying costs to experiment (4) To earn income or wealth indirectly Start company, or license patented invention signal to employers; get hired as engineer (Lerner and Tirole, 2002)
20 Conclusion Key assumptions of this model: motivated tinkerers, perceiving progress no perceived path to profit cheap communication This generates inventions, as by: Hobbyists “Skunkworks” inside organizations Basic researchers Better communications – an effect of the Internet is that it enables networks. For our session: an industry can arise this way