Gaussian Humans

Eliezer S. Yudkowsky (
Fri, 11 Jul 1997 01:23:52 -0500

Neutral Terminology for the 21st Century.
)1997 by Eliezer S. Yudkowsky.

Fairly soon, our society is going to be faced with altered humans. Some
naturally altered humans already exist. Proposals for deliberately
altering humans exist. Now would be a good time to start defusing the
panic that will result when the issue becomes popular.

This article deals with the subject of what to call unaltered humans.
I feel strongly that the term chosen should be neutral and should remain
so, more for the altereds' sake than to spare the unaltereds'
feelings. While there's no need to deliberately promote a sense of
class distinction that will retard research or spread fear, megalomania
has always been the great cliche danger of mind-tampering.

Bear in mind also, as you read this, that you are almost certainly an
unaltered human; as you review the appellations, reject any you feel
would be insulting if applied to you. If you are, or plan to be, an
altered human, reject any terms which make you feel megalomaniacal when
used to refer to others.


Historically, terms used to denote unaltered humans include "normals",
"mundanes", "ordinary humans", "humans"/"mere humans", "mortals", and so
on. I feel all of these are insulting.

"Normals", as used on Babylon 5 and many novels, seems slightly
derogatory; "mundane" is even worse. In both cases the implication is
that the person is flat/dull/obsolete. Few people become famous or
admired for being "normal" or "ordinary". For that matter, how many
"normal" people are there on this list? Your chances of being "normal"
aren't much better than your chance of being "modified".

Calling unaltered humans "humans" is worse still; the implication is
that the altered are a new species. Taxonomically, and ethically, this
will not be the case. Human/nonhuman distinctions should be reserved
for human-computer distinctions, or humans who have been neurally
reprogrammed to the point that they have entirely different cognitive
architectures and are essentially neurally based computers.

"Mortals" is worse still. Mortal/nonmortal distinctions should be
reserved for {humans, AIs} vs. {Vingean Powers, Singularity Children}.
Using the term to refer to others, if meant seriously, is a short step
from megalomania. There may be some slight justification for medical
immortals using this term, but even these long-lived folk could fall
prey to a nova and are thus mortal.


I therefore propose the phrase "Gaussian", as in "Gaussian
distribution", also known as a bell curve. A Gaussian can be anywhere
on the curve. A Gaussian can be an idiot or a genius or an average
fellow. No geis of normality or uninterestingness is implied.

The implication is that the person is a legitimate part of that curve,
which an altered human is not. Let's say we implant 1,000 people with
neural interfaces to pocket calculators. If we now look at the total
bell curve for all human calculating ability, we'll find an ordinary
looking curve with a huge spike way off on the right. Moreover, the
curve and the spike will have been produced by entirely different
internal models, one natural, the other artificial.

Of course, that's a best-case scenario (for the altered humans). Other
alterations (such as those I've proposed in "Algernon's Law") would
produce somewhat odder disturbances in the bell curve. In addition to
spikes on the right, there'd be spikes on the left, flat areas in the
middle, broken correspondences, and all sorts of statistical havoc. I
daresay that cognitive scientists of the future will have to exclude
altered humans from the sample of any study not specifically geared to

Which is the general idea. You have Gaussian humans, who tend to be
distributed smoothly and statistically, with most in the middle and a
few on the edges. Then there are altered humans, whose curves come in
all sorts of odd shapes and sizes - not necessarily with any net trend
towards the right. (In other words, the altered will be weird but not
necessarily superior.)

(Picky readers may note that many altereds' curves will be Gaussian in
their own right; however, the mean will probably be different, which
would distort the unaltereds' curve if factored in. And since an
altered human is likely to be the product of a single disturbance
rather than a large set of environmental and genetic variations, the
curves *are* less likely to be Gaussian.)

Finally, the phrase "Gaussian" has never been used in any science
fiction novel (that I am aware of) to refer to a group of humans. It
will therefore have only whatever connotations we choose to give it, and
will initially evoke no prejudices.

Do any of you deem this phrase insulting?
Would it make you feel megalomaniacal if you used it to refer to others?

Eliezer S. Yudkowsky.

--       Eliezer S. Yudkowsky

Disclaimer:  Unless otherwise specified, I'm not telling you
everything I think I know.