incubator-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ben Hyde <bh...@pobox.com>
Subject do no harm
Date Thu, 07 Nov 2002 14:20:07 GMT
Reading some of the posts in general@incubator has triggered one of my 
canonical rants.
Forgive me.

There is a syndrome that often arises inside commercial software 
projects.  The forces
which be agree that task X is a noble and fine task.  Task X is staffed 
and budgeted.
The folks working on task X spin up some processes and procedure.  
Later it is noticed
that they have taken on task X^2 and that the result isn't particularly 
useful.

This tends to happen often on 'we ought to write some doc for this 
code' projects, but
it is also pervasive on 'we should have a process and procedures' 
manual.

In both cases what happens is that people get drawn down the depth 
first rat hole.
Instead of cherry picking each and every thing they can capture that is 
known they
get caught in a tarpit of hill climbing toward perfection on the 
individual topics.

One reason this happens is that the people that know the thing at hand 
are willing
to rant - or to use the jazz term riff - on any topic that comes up.  
This is fun,
and interesting, but it's not cost effective.  So when you notice this 
you need to
break up the 'fun at the water cooler' and get back to work.

Another reason this happens is that projects of this kind are often 
drawn rapidly
to the unsolved problems.  The places Z where in the history of the 
system a forest
fire of problem solving and emotion swept thru the thing; but where the 
the ground
has not yet healed and the consensus as to what was learned hasn't 
emerged.  This
just opens old wounds - or worse leads to task X getting captured by 
one or another
side in the old task Z.  That kind of thing is outrageous, task X 
should stick to
it's brief.  So when you notice this you need act like the policeman 
"Oh folks,
this isn't your concern - move along now."

These, and others, can do a lot of damage.  If you take someplace where 
things are
fuzzy for a good reason and force them to be clear all your doing is 
naively forcing
a choice where the system has already discovered that there isn't an 
obvious good
choice.  This can be way harmful!

To pick a current example.  I'm concerned that I perceive a tendency to 
be overly
fastidious about what the 'veto' rules are.  I think it's best if you 
just capture
pointers to discussions in the existing archives with a gloss of 
summary and
leave it at that.

I think your goal should be to capture a sense of culture, not to write 
the open
source version of Robert's Rules of Order - which by the way are 
fascinating.

  - ben


Mime
View raw message