Morality, Semantics, Physics, Java, Reality

Eliezer S. Yudkowsky (
Mon, 13 Oct 1997 00:05:33 -0500

Full subject:

"On Morality, Semantics, Quantum Dynamics, Java Programming, and the Nature of Reality"

Delmar England wrote:
> At 05:23 PM 10/10/97 -0500, you (Eliezer S. Yudkowsky) wrote:
> >Right. You reject the phrase as a contradiction, even though you know
> >perfectly well what I'm talking about (alternate Universes a la "Lord Kalvan
> >of Otherwhen" or "All the Myriad Ways" or two hundred other books), and even
> >though counterfactual processes are built into the basic ontological substrate
> >of reality (quantum mechanics; the Schrodinger equation and linear evolution).
> >
> Say what! What on earth are you talking about? What is it you imagine that
> I "know perfectly well?" Until your post, I had never heard of "Lord
> Kalvan", nor "Otherwhen", nor "All the Myraid Ways."

So you've _never_ heard of parallel worlds? You haven't seen them on Babylon
5, or Star Trek, or in print? And (see below) you've also never skimmed
through quantum mechanics? What are you doing discussing terms like
"objective", then? No matter how many dictionaries you read, you have to be
able to point to something or it's all just a fog of words built on words.

> As for, > and even though counterfactual processes are built into the basic
> >ontological substrate of reality (quantum mechanics; the
> Schrodinger > >equation and linear evolution).
> I haven't heard of this before either. What is it supposed to mean? I don't
> know and considering terminology like "counterfactual processes" seriously
> doubt that anyone else does either. I don't know what your problem is,
> fella, but you sure have me mixed up with a lot of things that ain't me.

Never challenge a scientist to define his terminology... it's not good for
your sanity!

Okay, here's what actually happens, physically speaking. If you fire a photon
at a half-silvered mirror, reality splits into two parallel Universes, one in
which the photon was reflected and one in which it wasn't. If the alternate
photons hit another mirror, the Universe is split into four lines. We know
that all four exist, because we can make two of them cancel out and two of
them add up so that the photon winds up in a specific place. We can set up a
physics experiment such that the photon can go to C via A or B, or to D via A
or B. In "classical" physics, the photon goes to either A or B, and thence to
C or D. In quantum physics, the photon goes to both A and B (in alternate,
counterfactual branches), and then goes to C via A, C via B, D via A, and D
via B. However, D via A and D via B, when added together, sum to zero. So
the photon goes to C. If you prevent the photon from going through B, then
the photon ends up going to C via A or D via A, so it has a 50% chance of
ending up at either.

Perhaps this makes no sense to you. But that's the way it is. Ask any
physicist. Counterfactual processes, or processes in which "it might have
been" exerts an experimentally detectable causal influence on reality, are
part of the current known ontological basis of reality.

> >Wrong. My version was altered from yours almost verbatim, maintained most of
> >the phrasing, and did not alter any meaning except substituting "reality"
> for "morality".
> "substituting 'reality for morality'? And you see nothing wrong with this?
> [rant deleted]
> You call this "paraphrasing?" I think I need to call a linguistic
> paramedic because you sure have butchered the hell out of my words.

Delmar England wrote:
> At 08:38 PM 10/7/97 -0500, you (Eliezer S. Yudkowsky) wrote:
> >Flaws in attacks on "objective morality" can usually be detected by
> >substituting "reality" for "morality".
> A fine substitution I would say, and not at all flawed.

For the second time, try to remember what you said! I'm getting tired of
acting as your long-term memory! If you can't keep track of your own
thoughts, at least don't accuse me of being at fault!

> >I am not debating with someone who mysteriously forgets what the term
> >"objective morality" means after starting an attack on it.
> I sure didn't 'forget' and imagine that objective reality (real) and
> "objective morality"(myth) are the same thing. Somebody did. Guess who.

I take it back. There's nothing "mysterious" about your forgetting the term
"objective morality". Nor should you be blamed for forgetting that the
morality/reality analogy axis is used for the detection of logical flaws,
rather than as a proposed identity. But, I still think you should tone down
the attacks. And maybe get an email program that will let you view entire
threads as they evolve.

> >There is a
> >difference between analyzing a definition and pretending not to understand it.
> Definitions are my forte. See the book.

My "forte", among other things, is extensional definitions. That is, for any
given term, I can define it by pointing to something, rather than in terms of
other terms. I think that our problems in communication derive from a clash
between extensional and intensional definitions of terms such as "morality",
"goal", "desire", and so on. I'm a cognitive engineer, so I define these
terms as being computational structures, or the human implementation of those
structures. Example:

class Goal extends Object {

Goal[] superGoals;
double probability;
double initialValue;

double value () {
double value = initialValue;
for (int i = 0; i < superGoals.length; i++)
value += superGoals[i].value() * superGoals[i].probability;
return value;

My question about the existence of "objective morality" can be phrased as
follows: Given the above definition of "Goal", does there exist a system,
such that all initialValues are zero, and all probabilities are 1, in which a
Goal object exists such that value() returns a nonzero method?

This may seem impossible, but it also seems impossible that such a system
should return a nonzero value() with no initialValues, regardless of
probability. Since there are, in fact, logical ab-initio Goal-sequences with
probabilistic arguments and nonzero values, deriving from the system's ability
to reason reflexively, the issue is in doubt. Since "reality" is as much a
derived quantity as "value", the issue is even more in doubt.

Or to put it another way:

class RealThing extends Object {

Cause[] mCauses;

Boolean isTrue () {
for (int i = 0; i < mCauses.length; i++)
if (!mCauses[i].isTrue())
return false;
return true;

class Cause extends RealThing {
// Here's the paradox; causes are themselves RealThings.

How can anything be true without assuming a First Cause?
(An assumption is a RealThing with no causes.)

Note the similarity to the Goal problem. Insofar as something does exist, I
conjecture that the Goal problem has a solution as well.


You might be saying something about morality and reality, even an important
something, but I can't connect it with my mental model of goal and subgoal, or
my model of cause and effect.

> So you don't want to debate with
> me anymore. That's great, but considering your "paraphrasing" and other
> random attributing to my position, I don't see that you have EVER debated
> with me.

I was using the logical technique known as "reduction ad absurdum", which is
often quite effective against "definitional" attacks. If a given line of
logic results in a falsehood when applied to a set of known facts, the same
logic cannot be used to disprove a set of hypothesized facts. Since your
attacks on "objective morality" work just as well when applied to "objective
reality", it follows that a flaw exists in your logic.

My "attributions" are either verbatim quotes you forgot about, or are
carefully designated as belonging to an "Alternate Delmar". The Alternate
Delmar is a logical construct used for purposes of reductio ad absurdum. I
doubt that any readers, other than yourself, managed to confuse Delmar and his clone.

--       Eliezer S. Yudkowsky

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