- 1940s ~ 1970s
- Develope driven by universities or company labs. e.g., MIT, AT&T Bell Labs
- Openness and Collaboration culture
- "Software is a tool to use hardware.
- System users in person make software and share that.
- The beginning in Bell Labs in 1969.
- Version 4 - rewritten in C (1973)
- Disappear need to produce OS for each hardware - The First Portable OS
- Version 6 (1975)
- Start licensing for universities and companies.
- Restrict "software as a business" due to US government antitrust
- UNIX SYSTEM V (1983)
- Commercialization of UNIX -> UNIX WARS
- Market domination of IBM hardware & Microsoft OS
- not need source code due to standardized PC system structure.
- Shareware
- software that is distributed free on a trial basis with the understanding that the user may need or want to pay for it later.
- "Try before you buy"
- not Open Source
- MIT AI Lab
- "Pure hacker paradise"
- US Copyright Act of 1976
- Program productor stop providing source code
- Lisp Progrmming Language
- Commercialization (1979)
- Symbolics vs Lisp Machin Inc. War
- By Richard Stallman (1983)
- Unix-compatible
- A world where can be lived even with free software
- GNU Public License (GPL, 1989)
- In case of distribution by revising GPL program, must release source code.
- Copyleft, reciprocal licence
- In past, produce OS specified various hardware
- As become standardizing hardware as Intel x86 architecture, occur competition for domination for os market
- Microsoft
- OS/2 with IBM (1985) : fail due to opinion difference with IBM
- Windows NT (1993) : Due to convenience of installation and using, be more competitive
- World Wide Web (1990)
- popularization of Internet
- Distributed computing trend
- One large System (scale-up) -> Many small systems (scale-out)
- OS for Internet Servers
- Researchers and DIY users prefer modularity of unix
- Declination other OS like BSD UNIX.
- New Challenger : LINUX
- Linus Torvalds, Finland undergraduate student, start as hobby.
- Unix-like OS based on the Linux kernel, GPL licenced
- Free download, UNIX program and skills compatible
- Explosive growth by getting popularity for students, research institute.
- Advent of Internet companies (Google, Amazon, Facebook)
- prefer software that can be customization as low price
- Open source is best
By developing and starting from providing technologies for Low-end customers, process that pushes existed competitors that are in high-end.
- Ecosystem suitable for innovation
- Cloud computing, big data, containers, AI
- e.g. Hadoop MapReduce + Apache Spark + Ceph
- Destruction of Software Monolith
- Minimaize License fee, installation cost and tying
- Distributed and flexible computing landscape
- Mix and match
- Focus on implementation than Standard setting
- Richard Stallman
- GNU Manifesto (1985)
- Philosophy against of commercialization of software
- Free as in beer vs. free as in freedom
- Share source code
- Freely revision, distribution
"Open Source Summit" - Tim O'Reilly (1998)
- Include widely range of concept than free software.
- organization, collaboration, innovation, development
- Pragmatism / Commercialism
- Active investment of big companies. e.g. IBM
- Laying the foundation for Internet service in the late 1990s
- Open Source Initiative (OSI)
- Free redistribution
- Source code
- Allow modifications
- No prohibitions about who or why
- e.g. "Educational purposes only" - impossible!
- Copyleft License
- GNU General Public License (GPL): GCC, Linux
- If revise or distribute program, must distribute with source code of revised one. (reciprocity requirement)
- Prevent Free rider : tragedy of the commons
- Permissive Licensee
- Apache, BSD, MIT License: Swift, Angular.js
- Users grant authority that is freely available to program.
- Can use commercially through Copyright
- Focus on activating community
- Trend that prefer to Permissive License.
- Flexibility
- Convenient for companies to use Open Source and closed source program together.
- License Compatibility Problem
- Permissive license project can be granted GPL, but once it becomes GPL, it never come back.
- “Linux is a cancer” - Steve Ballmer, CEO of Microsoft (2001)
- Participation
- The more users, the easier it is to activate
- License is only requirement, activating of open source project is contingent on governance.
- Operational methods that drive user compliance
- First step : Policy
- Organize available license list and scope of use
- Preliminary investigation not to be problem legally
- Second step : Process
- Continuing while project live
- By using scanning tool, organize license conflict and dependency between software component
- Projects (upstream)
- Code that developers contribute to
- Community : core contributors, mailing lists, websites
- Products (downstream)
- product made for solving costomer's business problem.
- Combine, sell and support project according to specific usage
- Project and product are interdependent
- Product is concept that includes support to service by project satisfying customer's requirements
- Basic customer service : installation, solving error.
- Offering solution and required documentation when Several open source projects are connected
- Life-cycle support : Version control, dependency solution
- Legal support : Management of intellectual property disputes, etc.
- e.g. Red Hat: Enterprise Linux subscription service
“Security through obscurity is not true security”
- Not guarantee security even if implementation is opaque
- Because cyber attack penetrates already known weakness
- "Many Eyes" : can recongnize and revise problem of project fastly
- Continuously need security activity even if there are many users.
- Start new Open Source Project develop
- Is new project needed?
- What is business objective through project?
- What is project develop requirement?
- Stand-alone project, foundation/company-backed?
- Linux Foundation, OpenStack Foundation, CNCF, etc.
- Hosting / communication / governance model …
- Participate existed Open Source Project
- Identify community stucture and, lurk to understand norm and culture
- Start small : indirect participation (documentation, funding)
- Code talks: quality, quantity, consistency of contributions
- Occasionally, companies contribute bug fixes and feature updates created for business needs.
- Need considerable time and effort to understand Licensing, compliance, and participation, etc.
- A Toolbox of Practices
- Not one of standardized model, but change according to objective and requirement
- Success factor of Linux
- System that can release by applying numerous user feedback fastly
- In early days, frequently updating more than twice a day
- The Cathedral
- According to a prior plan, skilled workers divide the labor
- Thoroughly management toward fixed destination
- e.g. GCC (GNU C Compiler)
- The Bazaar
- No central control
- The quality of work varies and the efficiency may be lowered
- Value wide participation and experimentation, and continue organic evolution
- e.g. Linux
- It is difficult to say that a certain system is superior
- Because insight comes about from individual, need develop model that recongnizes and reward good insight
- Social and decision-making framework for a project
- Decision formation process
- Project's core principles
- Project's core goals
- community
- Some decisions are required at the beginning of the project (License, IP, Goal, Range, Role assignment), but it is effective that most decision is taken over to community
- Benevolent Dictator for Life (BDFL)
- A principal has right of final decision(mainly project initiator) about main matter
- e.g. Linux(Linus Torvalds)
- Generally the way in small project
- It is difficult for companies to participate
- May not be benevolent to new developer
- If not be recognized to BDFL, occur second-class citizen issue
- Risk of fragmentation between BDFL and community
- Meritocracy
- If be recognized ability, be granted permmision of code repository from develop community.
- e.g. Apache Foundation
- Merit is evaluated mainly based on experience and knowledge
- Consensus (a.k.a. liberal contribution model)
- Get bigger effectness according to recent contribution
- Decision is made through consensus process like vote, etc.
- Open First
- Make decision related project openly
- Transparency, openness and encouraging participation are important
- Separate between skill and business decision
- Open source projects should be steered, at least in part, by those contributing the code
- Upstream development handled by contributors, downstream product decisions lead by companies
- Example : Kubernetes
- Should be detail and clear by giving direction
- What value do we make?, How?, For Who?
- Leaders
- Have right of final decision about feature, release, etc.
- BDFL, committee, or voting group
- Maintainers
- Maintain/manage the part of the project you are responsible for
- close to editor than programmer
- Legal work, event planning, bug tracker management, etc.
- Committers
- Developers who can summit in person to code base
- Reviewed by maintainer
- Contributors
- All Contributions are reviewed before submission
- Reflect review, feedback and resubmission by passing through more than one time
- Occasionally, program user becomes contributor
As it is frequent that contributor gets out project, community should always receive new contributor
- Exchange control for influence
- The more freely it is, the more contributions there are
- Reducing Friction of Tools
- Develop environment or infrastructure could form barriers to entry
- simplify set-up by using virtual machine, container
- Through modular design, give environment that contributor can develop independently from other parts
- Mentoring
- Short term expense is occured by requiring mentor's time investment and sincerity, but it is necessary to maintain project in the long run
- Start from giving a small work
- Prompt feedback and update are important
- Culture
- Culture orienting empathy, respect, and inclusiveness
- Members could feel as sense of belonging is reward
“Humans adjust their behavior based on the metrics”
- Selection of evaluation criteria
- e.g. batting average vs. On-Base Percentage (Moneyball)
- An important metric selection process required
- Interpretation of indicators
- e.g. Are bug report increments always bad? What if the number of users increased?
- Soft Metrics
- Elements that occur costs, such as surveys and interviews, but are necessary for the project to be sustainable in the long run.
- Community Composition: What is the ratio of independent developers to enterprise developers? Isn't the influence of developers from specific companies too great?.
- How satisfied is the community? Is communication between members working well?
- Free technical support service at the Apple Store Genius Bar
- Facebook gives user free service and gives companies Ads and information of user.
- “If you’re not paying, you’re the product”
- Basic products is available for free, but cost to upgrade.
- Free app of Appstore
- Main goal : Proper balance between free and pay.
- If free product is so good, would not upgrade.
- If free product is poor, would not use.
- "Sales funnel"
- Potential paying customers
- Raising awareness and inducing evaluation
- As a kind of Freemium, give software for free, but sell additional functions for pay.
- e.g. MySQL database
- Similar to Open Source, Anyone participates develop freely, but grant owner selling permmision about some code
- Contributor license agreement (CLA)
- Not a true Open Source development model
- Difficulty in encouraging contributors to voluntarily participate due to the perception that the seller owns the project and community
- Red Hat(est. 1993)
- By starting selling own Linux distribution to companies, give suppport through subscription service.
- Many of Red Hat are core contributors of Linux
- Customer service based on professionalism about Linux
- In person, Update at Linux according to customer's needs
- The organization can focus on core competency and adjust material needed for develop
- License vs. Subscription
- Selling License model has incentive to release new version than to improve existed products
- It is subscription model's goal to retain customers, it is core to update continuously
- Expecially, Subscription model in Open Source has advantage of retaining available permmision about software if supscription is ended
- Traditionally, Cooperation between competition companies is loss economically
- Prisoner’s dilemma, free rider problem
- But, in modern software industry, be rising of cooperative competition inspired by Open Source Development Model.
- Competition in traditional industry
- If competition company has company information, decline profit(patents, trade secrets, etc.)
- Collaborative Innovation (MIT economist Eric von Hippel)
- Software Industry
- Low communication cost
- High design cost
- If cooperation is successful, pay only a portion of the design cost, enjoy the value of the entire project, and increase the possibility of innovation
- It is economically beneficial to grow whole pie rather than “free riding”.
- Software Industry
- Through cooperation of competition companies through Open Source
- Development of interoperable solutions and standards.
- R&D risk sharing.
- Capable of responding to complex customer needs (DevOps).
- Formal standard
- Industry representatives form a standards organization and establish and publish standards as a formal meeting
- e.g. OSI model
- Standards based on usage
- Implicit standards based on users' usage patterns
- TCP/IP (1960s) by DARPA
- Open Source
- Companies cooperate according to requirement
- From beginning, All details are not assigned
Methodology focused on practical coding rather than systemical plan
- Individuals and Interactions over processes and tools
- Working Software over comprehensive documentation
- Customer Collaboration over contract negotiation
- Responding to Change over following a plan
As expended concept of Agile, apply principle and convention of Open source to project management and culture.
Faster, more flexible, incremental : continuous improvement based on metrics
- Separate role between developer(Dev) and Operation organization(Ops)
- Ops give environment for Dev to be active, and do not interfere
- Communication is effective in small team (2-pizza teams -- not more than 8 or 9 people) -> O(n^2)
- Communication minimization through Well-documented public interface.
- Modularity : Design to reduce task dependence of each team and to be active independently
- General characteristic of Open Soruce development
- Unbundling, mix & match, tinkering center
- Infrastructure tool-oriented development (Linux, Apache web server, etc.)
- Trends threntening Open Source
- Advent of Cloud Computing
- Change commercial value of software
- Pursuit convenience of costumers
- In 2000s : PC age distriuted computing
- In 2010/2020s : Due to advent of integrated stack of Cloud-based companies, re-centralization
- Cloud giants
- Cloud giants contribute opensource a little(e.g. Google Kubernetes), but because it is more that developers in companies use opensource only at Enterprise program, There is Open Source Project sustainability crisis
As popularlization of computer, convenience and simplicity are important
- Presence as a part of complete product by bundling with hardware
- Decline commercial value of software itself
- Incline simple program for End-user.
- e.g. Apple iPhone -> Not sell iPhone OS and capable of development only in App Store ecosystem
- Serverless computing
- Cloud provider is charge of distribution of server resources
- the most active infrastructure in open source become abstract.
- e.g. AWS Lambda
- Event-driven
- Program flow is decided from events or information
- Optimization in small on-demand application development
Open Source apart from software code and development model is applied variously
- Value
- Rising value by prevailing Machine Learning model.
- Google, Facebook
- Even if the program is open source, data is not
- Transparency
- Trend that public institute data is opened to public
- There is trade off between transparency and privacy
- When anonymized data is also mixed with other information, privacy concerns arise.
- e.g. Netflix Prize(2006)
Users produce new contents under Creative Commons open source license
Lowering barriers to entry into education
- MOOC (Massive Open Online Courses)
- MIT OpenCourseWare, Coursera, Udacity, edX
- Arduino Project
- Raspberry Pi
Reference
- Gordon Haff, "How Open Source Ate Software"