Open Source Projects Manage Themselves? Dream On!

1
Open Source Projects Manage Themselves? Dream On!

• (Another take on CatB)
• Chuck Connell, Tufts Univ.

2
Historical Background

• Software industry began about 1950.
• Eric Raymond wrote Cathedral and the Bazaar in
  1997.
• For those 50 years, all large software was
  created by companies, government agencies, or
  universities. (Never hobbyists!)
• Microsoft controlled small computer market in 90s
  with much fear and loathing.

3
CatB Seemed Revolutionary

• Eric Raymond said There is another way to
  develop large software systems. We can do it
  ourselves!
• Like someone saying today We are tired of
  bankers. Well create our own mortgages
  companies, savings banks, and stock brokers.
  Well do it better, and it will be free for
  everyone!(This seems impossible. How could we
  do it?)

4
Praise for CatB

• Worldwide programming community is
  revolutionizing software and business models.
• You cant afford to not read this to do
  business in the next century.
• The ability of open-source software is simply
  amazing.
• A watershed declaration of independence.
• And, there was religious fervor... The people
  were finally standing up to the evil overlords in
  Redmond!

5
CatBs Main Points

• There is a wide world of programmers eager to
  write free software.
• OSS projects succeed without traditional
  management.
• With a lot of people looking at some code, all
  bugs are easy to see and fix.
• A fluid group of programmers can create complex,
  high-quality software, with no one telling them
  what to do. (a babbling bazaar)
• Other ??

6
Much of the Hype Was True

• OSS model was a new way to develop software.
• Thousands of programmers really did work (hard)
  for free.
• The resulting software was serious and good not
  just for hobbyists.
• Nothing like this had ever been done in any
  discipline.
• And the netizens really did compete with
  Microsoft.

7
But (IMHO) Went Too Far

• View of OSS management
• Assertions about debugging
• Overall topology of the work model

8
No Regular Management?

• What is traditional management? (according to
  CatB)
• Define goals and keep everybody pointed in the
  same direction
• Monitor the project and make sure details don't
  get skipped
• Motivate people to do boring but necessary work
• Organize the deployment of people for best
  productivity
• Marshal resources needed to sustain the project
• ESR claims these functions not needed by OSS.
• But this is just what Eric Raymond (and Linus
  Torvalds) did!
• CatB spends most of its text explaining how.
• Only difference is manager and workers did them
  for free.

9
Debugging Is Easy?

• "Debugging is parallelizable.
• "Given enough eyeballs, all bugs are shallow.
• Are these statements true?
• In fetchmail and Linux, many people in parallel
  looked for bugs and proposed fixes.
• But one person (ESR, Linus) actually made fixes,
  after thinking about each one.
• Deciding which bug fix to use is not trivial.
• Some bugs are not shallow! What about design
  bugs?

10
Cathedral or Bazaar?

• Three models of work flow

Cathedral
Bazaar
Traditional
11
Cathedral or Bazaar?

• Most OSS projects are one layer of management
• One visionary, many free laborers
• ESRs bazaar is really a cathedral!
• (With the same religious fervor)
• What would a bazaar model OSS project look like?
• Anyone aware of such a project?

12
My Prediction (from 2000)

• OSS projects need strong, smart, centralized
  management to succeed.
• Has the last eight years proven/disproven this?

13
ESR Said In Reply

• Mr. Connell's analysis has grave flaws
• He is completely out of contact with the reality
  open-source programmers live in
• This paper was badly wrongheaded
• See www.chc-3.com/pub/manage_themselves.htm,www.
  chc-3.com/pub/manage_themselves_r1.htm,
  www.chc-3.com/pub/manage_themselves_r2.htm
  (IBM.com, Sept 2000)