[
Lists Home |
Date Index |
Thread Index
]
4/12/2002 4:55:56 AM, "Michael Kay" <michael.h.kay@ntlworld.com> wrote:
>The problem is that the set of features that actually get cut is the
>intersection of the sets of features that each individual is bold enough to
>cut.
Right. Standards groups generally work by consensus, which means that almost
any significant player has a veto over what comes out. Also, the ability of
participants to represent their own organization's interests depends on their
ability to form alliances, win credibility, etc. So there's a "if you want
to get along, go along" dynamic. Expecting standards bodies to cut features
is like expecting the US Congress to cut "pork" [special projects that benefit
one legislative district at the expense of all, negotiated via of off-the-record
vote swapping deals].
I'll guess that the set of features is ultimately trimmed because the standards
(and standards bodies) themselves die out and new ones take their place. I
wonder which standards Gosling was referring to in 1990? If they are still
relevant today, then the technology must have caught up eventually, the
implementations became stable and interoperable, and at long last the spec
became a "standard" despite itself. But I'll guess that many of the specs
he was thinking of simply became irrelevant, i.e., didn't "matter" to
a wide audience in the long run. What would Gosling have been referring to?
Do any of them "matter" today? Did they hit the 80/20 point despite themselves?
Maybe the motto here is "Use Occam's Razor to keep your spec neatly trimmed,
or someone else will use Occam's Razor to cut its throat."
|
