I think the first statement and my later clarification are much
the same thing: "Don't say something that might be offensive".
This is the easier of the two memes to attack, and perhaps by
concentrating on it, I reduced my arguments for the other into a
strawman by merely leaving the association in place. So, let me
take on the latter directly: "People can be hurt by words" is a
counterproductive meme.
Let us imagine ourselves as uploaded beings, as Katherine did in
another response to my assertion. In that realm, we receive inputs
from various sources: sensory devices, data storage devices, and
network packets--communications from other beings on the net. We
process these inputs, manipulate them, add our own creations,
evaluate the results, and produce outputs: either physical outputs
(such as buildings, artworks) or messages to other beings. In the
net, there is a standard for exchange of information. A packet
format, addresses, routing protocols; a "language" of sorts. Now
in this environment, let us say that a message is sent from one
being to another. It is addressed properly, in the proper format,
and represents data processed by a similar being. If the being
that receives the packet has designed his input mechanisms in such
a way that certain properly-formatted packets that don't meet some
more specific criterion cause him to malfunction, or to delay the
processing of other information, or to perform counterproductive
work, or to misinterpret other inputs, would you not consider such
an arrangement to be an engineering flaw in the recipient? You
would ask why he made his system fragile to a properly-formatted
input. You would advise him to engineer himself more robustly.
It's not too different in the meat world. We process inputs, we
send messages, and we receive messages in agreed-upon languages.
The suggestion, then, that one should spend time anticipating the
recipient's probable response to a message before we send it seems
equivalent to making the internals of each machine on the net part
of the communication protocol. No one would ever engineer a net
that way. The purpose of protocols is to set a standard so that
we /don't/ have to know about the internals of every machine just
to send information around.
Let us instead engineer ourselves more robustly to accept inputs
without malfunctioning, so that we can make use of a wider diversity
of inputs--experiences and communications--to work from. And let
us design the /protocol/ clear, expandable, and powerful enough to
accommodate our needs. Let's work on the language, and on our
understanding and reasoning, and on our ability to express ideas
in the language, rather than piling on burdens beyond the language
that we must accumulate like a stack of manual errata and release
notes (sorry if I'm getting too deep into the metaphor).
In short, let us stive to make ourselves more adaptable, tolerant,
and less fragile to experiences and ideas instead of putting limits
on the ideas we can express.