[ExI] Why AI agents invent their own language if you let them chat.

[John Clark]

On Tue, Oct 6, 2026 at 12:51 PM Adrian Tymes via extropy-chat <
extropy-chat at lists.extropy.org> wrote:

* >>  I promise you this, in the future I will give ONE HELL OF A LOT MORE
>> attention to what the journal SCIENCE has to say than what you have to say!*
>
>
> * > The journal made the same mistake you made.*
>

*So I'm supposed to disbelief science journals but believe you? Don't be
ridiculous!  *

*> **Science journalists are science journalists. *


*The majority of the scientific breakthroughs made in the last 75 years
were first reported in either the journal Science or the journal Nature.
And both have a sterling reputation. In cryptography there's something
called a "chain of trust"; it's a hierarchical sequence of validations that
links an end-entity back to a recognized root authority to prove
authenticity and integrity. Something similar occurs in the way scientists
communicate among themselves, if scientist X trusts scientist Y and
scientist Y trusts scientist Z  then scientist X has some reason to trust Z
too. Reputable journals like Science and Nature provide a valuable service,
they engender a chain of trust.*

 > *You have known, in the past, that appeals to authority are logical
> fallacies.*


*No. Appeals to authority are NOT always fallacies. When I read about the
results of an experiment in Science I do not have the ability or the
resources to repeat the experiment for myself to verify that it is correct,
but journals like Science, Nature, Physical Review Letters, Cell, and the
New England Journal of Medicine have a reputation for double and triple
checking everything before they print anything, and when they make a
mistake, which is rare, they immediately print a retraction. And that's why
it's more logical to believe something I read in Science or Nature than
something I hear on Fox News or Newsmax. *

*But I'm supposed to ignore them but pay attention to you? No Adrian, I'm
not going to do that.*


*John K Clark *


>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20261007/8b296a92/attachment.htm>