The Host's Dilemma: Voluntary Forfeiture in Platform Markets ...Presentation Transcript
The Host’s Dilemma: Voluntary Forfeiture in Platform Markets for Informational Goods Jonathan Barnett USC School of Law March 5, 2010
Easy proposition: Open organizational models for information production are not new.
A little harder proposition: Open organizational models for information production are not really open.
A somewhat harder proposition: The welfare implications of open models for information production are ambiguous, even for users.
Caricatures of Open Innovation
(Some academic) Lawyers; (Many) Policy Advocates: Open source is revolutionary and really good. Let’s subsidize open “anything”.
Economists: Open source appears to deviate from rational choice behavior. Let’s figure it out . . . Reputation effects; warm glow; intrinsic utility.
Management scholars: Open source is the innovation engine of the future. Here’s a link to my book.
Open Innovation Is Neither New Nor Exceptional
Bell Labs (AT&T): “Open” licensing (nominal royalty subject to cross-licensing), plus complementary know-how transfers
Research disclosures, sometimes at “academic”-level volume and quality (IBM Technical Disclosure Bulletin; AT&T; IBM; Xerox)
Open Software Foundation (1988): Not-for-profit organization founded by AT&T rivals to standardize Unix operating system.
Microsoft: API disclosure with no revenue sharing. Result: > 70,000 applications for Windows (cf. 12,000 applications for Mac).
Free software: Adobe Reader, Internet Explorer, Google, etc.
Transaction Paths in ICT Platform Markets End-Users End-Users Developer-Users Developer-Users Platform: Software, Hardware, Operating System
Host’s Dilemma (I)
Platform only has value over non-mediated transactions if it achieves scale, which implies network effects. Host must credibly commit to achieve scale.
Users must make nonsalvageable learning and adaptation investments to use the platform or any other platform. Host must credibly commit against hold-up behavior after achieving scale.
Hold-up opportunities are abundant.
Change price for access
Change level of access
Withdraw or constrain support
So long as host expropriates value < switching costs, users will capitulate. By anticipation, users decline to adopt (or “underadopt”).
Windows Releases (Desktop Systems) 32 & 64-bit Oct. 2009 Windows 7 32 & 64-bit Jan. 2007 Windows Vista (Desktop Versions) 32 & 64-bit Nov. 2006 Windows Vista (Business Versions) 64-bit Apr. 2005 Windows XP Professional x64 Edition 32-bit Oct. 2004 Windows XP Media Center Edition 2005 32-bit Dec. 2003 Windows XP Media Center Edition 2003 32-bit Oct. 2001 Windows XP Home & Professional 32 & 64-bit Sept. 2000 Windows Me 32-bit Feb. 2000 Windows 2000 Prof. & Server 16 & 32-bit May 1999 Windows 98 SE 16 & 32-bit June 1998 Windows 98 16 & 32-bit Aug. 1995 Windows 95 16-bit Apr. 1992 Windows 3.1 16-bit May 1990 Windows 3.0 16-bit Dec.1987 Windows 2.0 16-bit Nov. 1985 Windows 1.0/1.01 Type Release Date Release
Host’s Dilemma (II)
Platform is a theoretically durable good. But host maximizes long-term profits by constraining durability through upgrades and extensions.
In multi-period setting, reputation effects constrain hold-up behavior as platform achieves scale. But reputation is not a complete solution if host can not credibly commit against end-period behavior.
Result: Host’s dilemma persists at every release. This is true even assuming (i) end-user myopia, and/or (ii) developer myopia.
Non-Exclusive Imperfect Solutions
Contract: Not all forms of hold up are perfectly contractible (not anticipated or not verifiable). E.g.: extent/quality of support or backwards compatibility.
Not all users can be integrated (e.g., end-users)
Integration costs can be exorbitant
Set price below short-term profit maximizing price
Disclosure of trade secrets (e.g., source code, specifications)
Disclaim control rights or transfer ownership over platform
Complete forfeiture resolves the credible commitment problem. But it destroys any direct funding stream.
Long-term profit maximization requires that the host select some intermediate combination of incomplete forfeiture and incomplete control over the consumption bundle.
Control can be asserted in three non-exclusive ways:
Assert control over some portion of the platform
Assert control over platform with respect to some users
Give away platform and assert control over complementary goods
Old Models: Closed to Semi-Open
The Mainframe Model (IBM):
Customized software “given away” with hardware and support
Hardware usually leased
The Unix Model (AT&T):
OS developed at Bell Labs
Widely licensed at nominal fee until AT&T breakup (1983)
Funded by implicit user tax through AT&T monopoly
The Windows Model (Microsoft):
OS licensed broadly to OEMs (Dell, etc.)
Widely releases APIs and beta versions to third-party developers
The Mac Model (Apple):
OS preinstalled as part of hardware; not available for license
Limited cultivation of third-party developers
Old Software Model: Semi-Closed Platform Holder (OS) APIs+ support Independent Developers OEMs/ Users OS $ $ Software Applications
New Software Model I: Community (Pure Open Source) Core Contributors Users Bug reports, patches Code releases subject to license at no fee
“ Forking” (multiple versions)
Privatization (depending on “reciprocity” term)
New Software Model II: “Community Lite” (Hybrid Open Source) Foundation Entity Core Contributors User-Contributors Proprietary Sponsors “ Distributions”; Services; Hardware Users $ $ Reports, patches Code subject to license at no fee $
Selected Open Source Foundations Advisory Board with representatives from sponsor entities. Members are exceptional contributors as determined by Board. 40% cap on board representation by single entity. GNOME Foundation GNOME GUI desktop No members. Self-appointing Community Council. Public commitment to zero royalty. Ubuntu Foundation “ Ubuntu” Linux distribution Individual members (300). Self-appointing board; almost no corporate representation. No members. Self-appointing board. No corporate representation. Derives 88% of funding from Google “royalty” contract. Membership entity. Classes based on increasing dues with increasing representation rights on board. Max. 2 reps/member. Governance Structure Foundation Application (market share) Apache Foundation Apache server application (47%) Mozilla Foundation Firefox Browser (20%) Linux Foundation Linux kernel (8% server market)
Linux Platform (OS Kernel) Linux Foundation (subject to Board of 15 sponsor plus 2 independent reps) Core Contributors Sponsors/Members : IBM Novell Intel Red Hat Oracle Fujitsu HP Bank of America Hitachi NetApp AMD NEC Texas Instruments Motorola $500K/yr/p.c $ Staff (>50% total) Contributors-Users Users Code Reports, etc. >1/3 code changes, >1/2 signoffs Distributions; other goods and services $ Patent Invention Network Patents
Leading Sponsors of Linux General microprocessor; Linux-based operating system platform for mobile devices. Sponsor >$500K/yr. Responsible for 6% of code changes and 6.4% of signoffs. Intel Sponsor >$500K/yr. Responsible for 6% of code changes and 8.2% of signoffs. Sponsor >$100K/yr. Responsible for 12% of code changes; 36.4% of signoffs. Sponsor >$500K/yr. $1B commitment announced in 1999. 10,000 Linux-related positions. Responsible for 6% of code changes and 5.3% of signoffs. Level of Involvement SUSE Linux Enterprise distribution; related support and warranty services. Second leading seller of supported Linux operating systems. Novell Red Hat Enterprise Linux distribution; related support and warranty services. Top seller of supported Linux operating systems. Red Hat Linux-compatible hardware (servers), middleware, client applications; related consulting services. IBM Related Products Sponsor
New Software Model III: Market Altruism (Sun Microsystems)
2005: Releases Solaris operating system source code; establishes independent OpenSolaris Governing Board, which sets forth Open Solaris Constitution
Sun Inc. Hardware, Supported Software (Paying) Users OS Code subject to no-fee license Governing Board, subject to Charter, Constitution “ Core Contributors” (Members) Approved Community Groups Nominate members
Non-Profit Strategies in the Mobile OS Wars
Nokia (Symbian OS) (47% worldwide smartphones):
1998: Symbian established by Nokia, Motorola and Ericsson to hold acquired Psion OS.
2008: Nokia pays $410M for OS and transfers it to non-profit Symbian Foundation
Feb 2010: Foundation “open sources” OS.
Foundation founders: AT&T, Fujitsu, Nokia, NTT Docomo, Qualcomm, Samsung, Sony Ericsson, TI, Vodafone
Google (Android OS): Open Handset Alliance, sponsored by Google and various chip, telecom and handset providers.
Linux Mobile OS: LiMo Foundation, sponsored by various telecom and handset providers.
Intel/Nokia (MeeGo OS): Linux Foundation hosts open source software platform for mobile devices, developed by Intel, Nokia
Organizational Competition: Mobile OS Market 2.8% Open Handset Alliance (open source) Android 12.4% Microsoft Windows Mobile 13.7% Apple Apple iPhone 20% RIM Blackberry 47% Symbian Foundation (open source) Symbian -- LiMo Foundation (qualified open source) LiMo -- Palm Palm WebOS -- Intel, Nokia (open source) MeeGo Global market share (2009) Foundation/Parent Operating System
Main Points; Implication
Both for-profit and not-for-profit entities regularly forfeit knowledge assets in order to resolve the “host’s dilemma” and elicit user adoption.
Voluntary forfeiture is a strategic, not ideological, choice. Relaxing access at any point in the consumption bundle requires access restrictions at some other point.
Welfare implications of “open” over “closed” are indeterminate a priori. Open platforms may be welfare-neutral, welfare-improving or welfare-detrimental relative to closed platforms.