At 12:51 PM 3/14/2005, Dirk-Willem van Gulik wrote:


On Mon, 14 Mar 2005, Ceki Gülcü wrote:

> What is so special about the number 3? What's wrong with 5 committers? With
> 4? With 2? With 1?
..
> Does "sustainable" necessarily mean 3 or more committers? Is a project with
> a single committer yet consistent committer less viable than a project with
> 30 disillusioned and inactive committters?

Well - with just one commiter

-> for sure no oversight.

Why not? A single developer can be overseen by a higher supervisory body, as long as that developer agrees to submit to the decisions of the supervisory body. The number of supervised developers should not matter. You are probably referring to a situation where the developers supervise themselves. Social studies have shown that decisions made collectively do not always yield better results. For one, groups have a stronger tendency to persist in their commitment to manifestly bad decisions, compared to the capacity of a single individual to reverse his/her own bad decision. The capacity to reverse decisions (good or bad!) improves dramatically when the person or group reversing a prior decision is different than the person or entity that took the initial decision. Humans have an often devastating tendency to escalate on their initial (seemingly inconsequential) decisions. For example, most investors have a hard time selling loosing stocks and keeping winners -- the exact opposite of the optimal strategy -- because of their committment to their initial decision to buy.

This may sound contrary democracy  (rule of the people by the people.)
However, one core  component of democracy is the  separation of powers
and independence between the 3 branches of government.  So instead of
"supervision  of  developers  by  the  developers,"  another  approach
consists  of   supervision  of  developers  by  an   elected  body  of
supervisors, for example the PMC.

You could  cite Jakarta as a  counter example. However,  we should not
forget  that the  scope of  LS is  much more  restricted than  that of
Jakarta. Moreover, LS has 3 sub-projects compared to the dozen or more
in Jakarta.


->      for sure an issue if he or she gets hit by a bus, a baby,
        whatever.

and it is very questionable if what that group produces is an ASF
community product.


What is  a community product? If  a community product is  defined as a
product produced by the collaboration of multiple committers, then you
are right. However,  if a community product is  defined as the product
developed  by *one*  or  more developers  with  *input* from  multiple
constituents,  be  they  simple  users  asking  for  help,  developers
submitting  patches or  making suggestions,  then a  project  with one
committer can produce a community product.

This brings me to formulate the subject question "What is a healthy
community?" as "What is a community?"

Two is more than one - but it is at only three that group dynamics first
start to kick in. And it is that group dynamic which is vital in building
and maintaining a community.

Can you give examples of group dynamics kicking in within a group of 3 but not in a group of 2? The difference is not so obvious to me.

Assuming that  everyone votes,  a group  of 3 will  always be  able to
decide  by  majority vote,  but  that is  true  for  all odd  numbered
groups. Is the difference in group  dynamics between groups of 2 and 3
solely due to the fact that 2 is pair and 3 is odd?

Many successful partnerships have 2 partners, not 3 or more.

If we  consider a  community as a  partnership between  committers and
take nature  as a possible  model to emulate,  then 2 would  win hands
down. Indeed,  the vast majority of  living plants and  animals on our
planet are the result of the partnership between 2 parents, not 3.

Sure 5 or 15 is even better - but if you have to draw a line - 3 is as low
as you can go.

> a single committer yet consistent committer less viable than a project with
> 30 disillusioned and inactive committters?

Both should, and will, ultimately be garbage collected.

If garbage collection is cheap, then why not take chances with projects consisting of just one committer supervised by a distinct body?


Dw

-- Ceki Gülcü

  The complete log4j manual: http://www.qos.ch/log4j/



---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]



Reply via email to