Re: Which Singularity Don't You Want?


"Dr. Dave" <dtate@xxxxxxx> wrote in message
news:1133045901.866448.283770@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

> > > I thought you said evolution always gives you what would be of benefit.

> > No, I said that what we get from evolution always conferred some kind
> > of reproductive benefit.

> Actually, you didn't, but let that pass for the moment.

Well, let's debate on what we both think we think. I can't be bothered
defending earlier misstatements of my position. We're all feeling our
way here.

Specifically, I'm using

> > Evolution is like Santa. If you're good, you get something, but it
> > isn't always what you asked for. It may not even be good for you in the
> > long run. What allows you to survive for the next two million years
> > might be what causes your extinction when the comet hits or the ice age
> > comes back.

> With the caveat that being lucky is as useful as being good, I'll buy
> that so far.



> > > At any rate, if you like you can substitute the version about being
> > > able to run 40 mph. It simply is not valid to argue that if existing
> > > ability X is useful, then evolution will necessarily provide more and
> > > better X (much less optimal or perfect X).
> >
> > No, it's just that evolution won't give you X if X had no value.
>
> You mean I don't have wisdom teeth that I'm going to have to have
> surgically removed? Damn, that's a relief.
....

Dealt with elsewhere. (The sickle cell issue was responded to in
detail, and better than I could, and I piggybacked).

> > > Vision is tremendously important to humans, and we have a huge
> > > evolutionary investment in it. And yet, we nevertheless are subject to
> > > optical illusions, myopia, color-blindness, and a bunch of other
> > > sub-optimal traits, none of which our vision system is capable of
> > > recognizing in itself. So why is reasoning different?

> > See other explanation repeated till we are sick of it - basically, it
> > is a self-correcting mechanism.

> That's not an 'explanation' -- it's an assertion, at the level of "but
> reasoning is magically immune to this problem, and we just know that".

> An assertion of your conclusion, in fact -- if we really knew that
> reasoning were self-correcting in the way you claim, you would not need
> to invoke any kind of evolutionary argument. It would be sufficient to
> jump to the conclusion the way you have.

That's fair enough. I think that the ability of reason to make changes
to itself is fundamental, and a creationist or spaghettiist would
accept it on the same basis.

The reason that I think that reason is capable of amending itself is
that I think that that is fundamental to our understanding of what
reason is. We think about things, and change our minds accordingly. To
someone who doesn't accept this as a description of how reason works,
there is little that I can add to it. Anything further will indeed seem
repetitive.

This doesn't, in itself imply perfection. It is possible to contend
that the ability of reason to correct itself is indeed flawed. I don't
think it's possible to contend that reason has no ability to correct
itself. To say such a thing would be to ignore much of what human
beings spend their time doing - indeed, what we are doing right now on
this thread.

> > > > I'm not claiming that it was inevitable that human beings gained a
> > > > general reasoning ability.

> > > No, you're claiming that, given that they once started down that path,
> > > it was inevitable that it become essentially perfect by now. Which is
> > > not a reasonable claim.

> > I'm claiming that it was unlikely that as a general purpose ability it
> > would come with a built in flaw.

> No, you aren't, because every time I point out other general-purpose
> abilities that come with built-in flaws, you jump to the claim that
> conscious reasoning is different, and is inherently self-correcting,
> and that we can know this. In essence, you abandon your evolutionary
> argument and switch to an "it's obviously true" claim.

I'm asserting that reason is capable of affecting itself. That's
something that doesn't apply to physical abilities. The eye cannot look
at itself to improve its focus. Human reason can, for example, store
data externally for later retrieval.

> > > Under your theory, why did chimpanzees (who clearly are also working
> > > their way down the big-brain brance of the tree) not already evolve
> > > good reasoning abilities? The justifications you have presented so far
> > > for why humans must be superb reasoners would apply equally to them.

> > As I said, I don't believe that chimps are reasoning creatures.

> Well, everyone who has worked with chimps for any length of time
> disagrees with you. What do you base your particular certainty on?

FSVO reasoning. I believe that there is a range of abilities which
humans have which aren't shared by other species. But if some chimp
experts insist that chimps can too create sophisticated internal models
of the world - then I'll just extend my set of reasoning creatures to
include them. It doesn't change my fundamental argument, whatever that
is.

> > They might be thinking, in some sense. I don't think they are able to
> > construct internal models of the world other than those they get out of
> > the box.

> That's true of half the people I know, too.

There might be people who don't use their abilities fully. I still
think that they are able to look at a map of the London Underground and
figure out what station they need to change at.

> > > > > 2) Evolution always finds the optimal solution eventually, and

> > > > Evolution will tend towards a more optimal solution.

> > > No. This is a common fallacy, but wildly wrong. In the first place,
> > > the "objective function" for evolution is species survival, not visual
> > > acuity nor antler size nor reasoning ability nor any other specific
> > > ability of individuals.

It isn't even species survival. It's competitive advantage against
members of the species without the characteristic. It's concievable
that a species could become extinct because individuals acquired some
show-off mating displays that made them more attractive to predators.

> > > Second, it's a stupid randomized local
> > > hillclimbing heuristic, additionally constrained by the limitations of
> > > what kinds of phenotypic variation can be achieved through perturbation
> > > of the current genotype. It does awfully well at the population level
> > > at adapting species to stable environments, but it has serious problems
> > > at the individual level and in dynamic environments. Which, since the
> > > evolution of nearby species makes *every* environment a dynamic one,
> > > leads to the Red Queen problem of having to adapt constantly just to
> > > avoid losing ground.

> > > Species go extinct all the time; was that optimal for those species?

> > It's a local optimum.

> Extinction is a local optimum!?

Extinction will always happen at a less local scale.

> > As I've said or will say elsewhere, a species
> > gets an advantage from a particular facility. Evolution, once it has
> > granted that facility, will tend to make it work better.

> Again, I'll point out that this simply is not true, especially if you
> define 'better' as something other than reproductive success, as you
> have done for reasoning.

What I mean by better is that if evolution produces big antlers, it
will produce bigger and bigger antlers until they are as big as it can
get them. There will be a balance point where the inconvenience of the
huge burden on the head will match with being able to catch the eye of
the lady elk.

> > When the
> > facility becomes less useful, it starts to reduce the capacity of the
> > facility - such as the wings of flightless birds.

> So if evolution gives you an ability, it always continues to make it
> better, except when it makes it worse.

It always makes it better. It continues to improve the wings as wings -
until an environment is encountered which means that the species
doesn't need them anymore.

> How do you know we aren't the reasoning equivalent of flightless birds?

That's possible - but unlikely. So much of human activity centres
around reason, that it seems implausible that it is like the appendix
or cocyx or those pesky wisdom teeth. It isn't something left over from
an earlier stage - it's something we use constantly.

> > > > > 3) In our case it has already gotten there.

> > > > Well, whatever the costs, we do have reasoning,

> > > If you're going to just assume your conclusion, why bother with an
> > > argument for it?

> > I didn't think that anyone was arguing with the fact that we do have
> > reasoning. I thought the question was whether it carried a flaw.

> Your definition of 'reasoning' seems to include the "without a flaw"
> part.

To briefly summarise -

1 - We have a reasoning ability.
2 - That reasoning ability is able to recursively examine and refine
its own conclusions .
3 - I find it likely that if such an ability were granted, evolution
would tend to make it operate optimally.
4 - It seems unlikely that optimal operation would involve creating an
inherently flawed model of the outside world.
5 - If reasoning had an inherently flawed model of the outside world,
or if its methods had some other kind of inherent flaw, then the more
reasoning that took place on a particular problem, the more erratic and
erroneous it would become.

> > > > and it is able to examine itself.

> > > See above -- you're simply asserting your conclusion again here.

> > No, I'm claiming these characteristics as a given. If my axioms aren't
> > accepted, I'll have to backtrack a bit, but I'm working on the basis
> > that reason exists, that it can examine itself, and that human reason
> > seems to work differently from other animals.

> Well, if you admit the possibility that it can't examine itself
> *accurately*, I'll admit the axiom that we do attempt to examine our
> own reasoning. If you make the success/accuracy of such efforts part
> of your axiom, then you're simply assuming your conclusion.

I do accept that reason has limitations of accuracy, whether
considering itself, or the outside world. It's interesting to note that
reason has itself allowed us both to realise that, and make allowance
for it.

It's precisely that capacity to compensate for error which allows us to
keep thinking and discussing and thus possibly getting closer to the
truth.

> > > > I don't claim that evolution always produces the optimal solution.
> > > > However, it tends to produce a useful solution.

> > > Tell that to the Irish elk.

> > The Irish elk evolved its big antlers because they worked better as a
> > means of reproductive advantage over little antlers.

> Yes, I know.

That was an optimised antler. It may have led to a less than optimal
elk, and led to the cumbersome elk being unable to get away from the
predators chasing it.

The faculty of reason may, in the long run, lead to the extinction of
mankind. That doesn't mean that the facility was not optimised.

> > Evolution ran away
> > with itself producing enormous antlers such that the beast could hardly
> > hold its head up. In the general sense, one could say that having to
> > carry a huge weight on your head which is largely to attract the girls
> > will have significant disadvantges in, say, fleeing wolves - but
> > there's no doubt that the elk had some horns on him. You couldn't say
> > that as antlers, those antlers weren't impressive.

> > It may be that human reason is a dead end, and will lead to certain
> > extinction - but that doesn't mean that it works wrong. It means that
> > as a long term guarantee of species survival, it falls short.

> It could also be that we are currently running around with the
> reasoning equivalent of a huge rack of mostly useless antlers. Once
> you admit that evolution doesn't always make things more useful over
> time (wings on kiwis, antlers on Irish elk), you lose the inevitability
> of the perfection of human reasoning.

I totally accept that reasoning might not turn out to be a good idea in
the long term. (The jury is still out). That is a different
consideration though. I do believe that reason is essentially without
an inherent flaw. Since the principle capacity that I ascribe to reason
is the creation of internal models, I'm asserting that there is no
inherent limit to the complexity or coherence of those models, and the
ability to match them to the real world. (That's quite a big claim, and
I might have to come back and layer a few provisos on it).

> > > > A reasoning capability that was inherently flawed or biased would seem
> > > > to be less optimal.

> > > Tell that to the cockroach. Or, if you insist on restricting your
> > > discussion to large-brain strategies like mammals, to the Norway rat.
> > > They seem to be thriving just fine, and there's no chance of them
> > > exterminating themselves, which cannot necessarily be said for us.

> > But they don't have faulty reasoning. They just don't have the kind of
> > reasoning we have at all. They don't, as near as we can tell, form
> > abstract models of the world.

> Again, yes, I know. But you're the one arguing that forming abstract
> models of the world is an evolutionary advantage over (say) rats. I
> don't see any evidence for that.

I'm not claiming that reasoning gives an evolutionary advantage over
anything except faulty reasoning. At some particular stage of
evolution, HS picked up the ability to reason, which gave the members
of the species with the ability an advantage over those without.
Evolution pushed it.

Any objective look at the species of the planet would probably admit
that being a beetle was better than being smart. I don't think that
reason makes human beings the pinnacle of creation from an evolutionary
point of view. We are exactly as evolved as all the other species
currently in existence.

I do think that we have an unflawed reasoning capacity.

> > > Better yet, tell it to the Neanderthals. Remember, the goal of
> > > evolution is not to make you smarter, but to make your great
> > > grandchildren exist. Species that start down the "advanced reasoning"
> > > path don't have much of a track record; only one has survived even this
> > > far.

> > We don't know of any other species that has ever moved down this path.
> > It may be that the Neanderthals disappeared because they lacked the
> > specific reasoning ability that we HS have. (Though that isn't known).

> Are you seriously going to argue that Neanderthals didn't have abstract
> reasoning?

I'm offering as a very tenuous hypothesis that they might have had
weaker abstract reasoning than HS. I find it mildly plausible, but it's
also possible that they had better reasoning capacity than HS, but lost
out due to other factors.

Aha - so if Neanderthals could have better reasoning than us, how could
ours be optimal?

Good point - which leads me to think that I need to figure out what
exactly I mean. Clearly it is possible for intellectual capacity to
differ greatly. However, the process of reasoning is fundamentally the
same among human beings. At the very least, there is a core consensus
on how we reach conclusions.

Is reasoning capable of error? Yes, it is. Is it more likely to produce
error the more we do it? No, the more we reason about things, the more
reliable our conclusions, in general. Are all problems soluble by this
perfect power of reasoning? No, clearly not. Even with as much
information as we could want, and unlimited resources to think about
it, we know that there are things we just don't know - and in fact,
can't know.

It's possible to imagine an alien species with a different form of
thought - reasoning(1) - which is capable of reaching solutions
inaccessible to our reasoning - reasoning(0). Would the existence of
reasoning(1) prove that reasoning(0) is flawed? No, just that it has
limits - which we know already.

> > In any case, I'm not claiming that reason provides any particular
> > benefit

> Well, then that's a major retreat from your earlier position.



> > - except that if it gave no reproductive advantage, we wouldn't have it.

> ...and that still isn't true. Where's the reproductive advantage from
> my wisdom teeth?

You've had that already. Long time gone. But there was an advantage in
having that number of teeth, somewhere, sometime.

> > > > I don't assume that reasoning is perfect - just that it probably
> > > > doesn't include some flaw which makes all it's conclusions suspect.

> > > Fifth.

> > Well, I don't apologise for repeating my assumptions and claim.
> > Otherwise, we might end up arguing in parallel, which gets us nowhere.

> Fine -- but if you're going to state that you believe as an
> unassailable axiom that we are sufficiently rational for all human
> purposes, then you should probably stop presenting that axiom as a
> conclusion from some evolutionary argument. If it's an article of
> faith, that's different.

I think I've repeated both the evolutionary and recursive arguments.
I'm not claiming it as an article of faith.

> > > > It might be that their is a flaw which is
> > > > so cunningly constructed that it limits the accuracy of reasoning, and
> > > > which precisely limits the capacity of reason to spot the flaw in
> > > > itself. I find such an error difficult to imagine.

> > > And yet you can watch the monkey not get his cookie... To me, your
> > > comment sounds exactly like someone saying "there can't be a blind spot
> > > in my field of vision, because if there were I'd notice it".

> > Obviously we can't rule it out. But we can at least consider the
> > possibility that there is no inherent blind spot, and from that point
> > of view, consider what it would look like.

> What what would look like? I have a pronoun referent problem there.

What such a blind spot would look like.

> At any rate, you've turned the argument on its head here. It's not a
> question of "considering the possibility that our reasoning is good" --
> we've all done that already, and we continue to do it all the time in
> everyday life. You're the one who seemed to be claiming that you could
> do better -- that you could prove that there were strong positive
> reasons to believe that our reasoning is good, even if physicalism is
> true. At that point, the burden of proof is on you to provide an
> argument, not just assert a belief.

I don't see this as related to physicalism at all. That we are able to
think about something we previously thought about is simply what we do.


> > Can I rule out a flaw in reason which makes reason not work properly?
> > No - if there is a fundamental flaw in my perceptions, I can never find
> > it. But that gets us nowhere. Such nihilism just leaves us with no view
> > on anything at all.

> I agree completely. That's why I have a big problem with theories
> that, as best I can tell, *imply* that I am flawed in that kind of way.

I think that our capacity for reason is, as I have said, unflawed. The
evolutionary argument is supportive, but inconclusive. The recursion
argument is what I mostly depend on.

> > > > It is the
> > > > capacity of human reasoning to recursively examine itself which leads
> > > > one to surmise that it doesn't contain a hidden flaw which renders it
> > > > inherently inaccurate.

> > > It is the capacity of human reasoning to recursively examine itself
> > > which is precisely in question here. That's six.

> > I don't see how anyone can reasonably question that human reason
> > recursively examines itself.

> If you mean just the attempt, without any guarantee of success or
> accuracy, then I'll agree. But you seemed (to me) to saying that we
> not only examine our own reasoning, we also successfully and accurately
> evaluate its reliability. That we can, by this self-examination, tell
> that there are no "blind spots".

I agree. We can't tell that things are running properly when the tools
we examine ourselves with may be themselves flawed. My contention is
more pragmatic. It's based on the fact that we do lots of reasoning,
and it seems to work.

> > > Sometimes an accurate estimate of one's chances is not a positive
> > > survival trait. The bold survive disproportionately in environments
> > > that reward boldness.

> > Humans are bold in spite of their knowlege. Hence a tendency to use
> > substances that switch reason off when bear-wrestling is called for.

> ...or any other time. But you skipped past my point, which is that
> your claimed alignment between reasoned responses and
> survival/reproduction does not appear to hold in Real Life (tm).

It's a balance of different stimuli. Reason is neutral. It will tell us
that if we wrestle the bear, we will probably die. Then we have
self-preservation "I don't want to wrestle a bear and die!" - which
operates on a visceral level. Then there is the instinct to protect
people, especially those with matching genes - "That bear is going to
eat five of my first cousins, I feel impelled to interpose myself!"

I'm guessing that evolution has tuned the various impulses to balance
out as well as possible. I just don't think that the best way to do
that is to tell us "Yeah, I could take the bear. He ain't so tough."

> > > > But all our thinking is potentially subject to careful reasoning.

> > > For the record, that was seven.

> > And there will be many more. We are trying to find an area of
> > disagreement here, so we can have a proper argument about it.

> ObMontyPython: "This isn't an argument!" "Yes it is." "No it isn't!"

> There is no proper argument to be had if your basic stance is "our
> reasoning is valid, we know it's valid, we can tell it's valid by
> reasoning about reasoning".

> That's a religious belief, and at that
> point we agree to down swords and go have a beer somewhere.

I don't think I have a lot more to say to develop this, but that won't
necessarily stop me. Beer me anyway.

J/

.