[FoRK] Open Source Databases
Contempt for Meatheads
jbone at place.org
Tue Jun 15 12:34:28 PDT 2004
> There's a third possibility. In many cases, and I
> would be surprised if this didn't include Oracle,
> a lot of that complexity is created by business
> decisions, rather than engineering ones.
While there's some validity to this statement, too, the idea that the
process of building software involves distinct engineering and business
decisions is ultimately bogus. There's no clear distinction;
engineering decisions *must* be informed by business reality --- but by
the same token, "business" heads are not the only and final arbiters of
what's best for either users or the business. Too often engineering
lets itself fall into the trap of not representing the user's interests
(e.g., failing to quash ambiguity and complexity in the spec because of
lack of clarity in the minds of the end-users or, worse, "business"
intermediaries like product managers) or, indeed, their own (we often
end up making concessions that undermine the maintainability or
extensibility / evolvability of the software, both of which have costs
to the business. We shouldn't let this happen. The responsibility is
ours to defend "the right thing" and to communicate the costs of not
doing so to the *other* decision-makers; any failure is at least
The fine line is in finding the balance between "must ship" and "the
right thing." It's hard to find. But the best way to find it and
maintain it, IMHO, is by adamant insistence on clarity, simplicity, and
maximum parsimony in problem statement, design, and implementation.
The best line of code is the line that doesn't need to be written.
More information about the FoRK