detabbing again

[Seebs]

I think it's safer to say 2 this way:

2'. As of this writing, Seebs says he thinks that grilled cheese
sandwiches are delicious.

What you say can be verified objectively (using voice recognition
technology). What someone thinks can't be verified without asking
them, because thought reading technology is not yet advanced enough
(although, there are some indications that this may change sometime
soon).

Not counting the minor quibble, a very good example.

I don't think it matters whether it CAN be verified with existing tools,
only whether it is IN PRINCIPLE a claim which could be definitely true or
false. (Which may be fuzzy, because "delicious" is far enough from raw
base experience that someone's mind state might be ambiguous. Except
in really easy cases like this one.)

Consider the claim "Julius Caesar had precisely 106,215 hairs on his head
at the time of his death." I think it is an objective claim, and probably
one which is false, but I do not think it is remotely possible that we could
know whether or not it is true with any technology I expect to become
possible, ever.

In short, I consider objectivity to be a trait things have regardless of any
possible technological developments or the availability or non-availability
of data.

If you were to trace my paternal line back for three hundred generations,
there would be a man. That man had eyes which either were, or were not,
"brown eyes". Brown or not? Objective truth, but certainly unknowable.

-s

 

[Flash Gordon]

The question isn't whether or not the claims are true, the question
is whether or not they can be verified without reference to any
observer's interpretation.

There are now algorithmic ways of doing certain forms of analysis on
text. There are definitely ways of analysing and proving things about
code. Put these two fields of science together and you can get a
computer (not a person) to analyse the C standard for unambiguous
statements, such as the minimum value of INT_MAX, and then have the
program analyse a C implementation to see if it meets those
requirements. Obviously with current technology it cannot test all
aspects, but it is certainly possible to test some. Therefor it IS
possible to use an instrument (specifically a computer) to test for some
of these things.

I can make sense of both claims. In
the second case the claim can be verified objectively because
(in my opinion) its truth value can be measured by an instrument.
In the first case the claim cannot be verified objectively because
(again in my opinion) its truth value rests on whether certain text
"specifies" something else, and this condition depends on
who is interpreting that text.

You need to study physics more, there is a lot of stuff which is
dependent on the observer but can still be measured.

Also, I have yet to come across any instrument which provided readins
which did not need interpreting.

For another example, consider the claim:

The word "Septermber" appears at the top of most pages of
the n1256 C Standard draft.

and the claim:

Because the word "Septermber" appears at the top of most
pages of the n1256 C Standard draft, all even-numbered
paragraphs should be ignored in determining what it specifies.

The first claim can be verified by direct observation. The
second claim can be verified (or refuted) only by asking
someone whether they think it's right -- its truth value
depends on who you ask.

Some things not being objectively verifiable does not mean that nothing is.

Repeating myself from an earlier message -- the question of
whether or not a statement is objectively verifiable is not
itself verifiable objectively. So other people may disagree with
my assessment of what statements are objectively verifiable, and
if they do there really isn't anything I can say about that,
except perhaps something like, "well, if that's what you want to
think, I can't offer any objective proof you shouldn't."

Yet you *are* arguing about what what is or is not objective.

Right, but I wasn't making any claim about what the statement
about two iron balls "means". My claim was about how the
the physical objects will behave in the actual world, not
about the meaning of some sentence with those words in it.

Now perform the experiment and measure the results *without* affecting
the behaviour. Current theory states this is impossible. Therefor it is
not possible to verify your claim. Therefor, as it cannot be verified,
your claim as to the behaviour of those balls is not objective.

Let me try phrasing it this way: What questions about the
Standard can be answered without asking any person what the
answer is? To be objective, a question must be answerable
without consulting the opinion of any person.

Easy. What is the minimum allowable value of INT_MAX. I'm pretty sure
that with current methods it would be possible to write a program to
analyse the text and determine the answer to this question. It would not
be easy, but it would be possible.

In theory it would actually be possible to develop a system which had
every meaning of every word in the standard (and all referenced
standards) and prove a lot more than this.

 

[Tim Rentsch]

[snip]

... uhm. I propose we take this to email if we want to keep going. [snip]

10-4. I'll send email for this and also the other
un-replied-to posting of yours in this thread.

 

[Tim Rentsch]

There are now algorithmic ways of doing certain forms of analysis on
text. There are definitely ways of analysing and proving things about
code. Put these two fields of science together and you can get a
computer (not a person) to analyse the C standard for unambiguous
statements, such as the minimum value of INT_MAX, and then have the
program analyse a C implementation to see if it meets those
requirements. Obviously with current technology it cannot test all
aspects, but it is certainly possible to test some. Therefor it IS
possible to use an instrument (specifically a computer) to test for
some of these things.

Certainly it's possible to analyze text algorithmically and
process it in possibly useful ways. But that isn't the same as
what the text "means", unless you're using a very different
notion of meaning than most people. Also, note that just because
there is an algorithm that processes linguistic text doesn't mean
there is anything absolute about its results or conclusions.
It's perfectly possible to write an algorithm that would
transform the ISO C standard into "Romeo and Juliet", but that
doesn't mean the C standard was written by Shakespeare (or
that "Romeo and Juliet" was written by the ISO committee).

You need to study physics more, there is a lot of stuff which is
dependent on the observer but can still be measured.

I can pretty much hold my own in physics with anyone
who doesn't have a degree in the subject. But if you
have some insights that you think I would benefit from,
please feel free to send them to me in email.

Also, I have yet to come across any instrument which provided readins
which did not need interpreting.

It's easy to devise an instrument, for example for the
ball-dropping experiment, that turns a light ON if they
hit within one millisecond of each other, and stays off
otherwise.

Some things not being objectively verifiable does not mean that nothing is.

Okay, but that's nothing like anything I've said.

Yet you *are* arguing about what what is or is not objective.

I'm not. I'm only stating my opinion and trying to explain the
meaning of some previous statements. If after understanding my
statements someone still wants to hold a different opinion,
that's up to them, but there isn't any point in saying "I
disagree" without first understanding what the other person is
saying.

Now perform the experiment and measure the results *without* affecting
the behaviour. Current theory states this is impossible. Therefor it
is not possible to verify your claim. Therefor, as it cannot be
verified, your claim as to the behaviour of those balls is not
objective.

Depends on what measurement it is you're trying to do. For some
experiments, the error bars of the approximation being tested are
much larger than the fluctuations due to quantum mechanical
effects; measurements in such cases can be made quite reliably.
The ball-dropping experiment is one of these.

Easy. What is the minimum allowable value of INT_MAX. I'm pretty sure
that with current methods it would be possible to write a program to
analyse the text and determine the answer to this question. It would
not be easy, but it would be possible.

It would be possible to write a program to analyze the text and
determine /an/ answer to this question. It also would be possible
to write a different program to analyze the text and determine
a different answer. Neither program has any inherent claim
to producing /the/ answer.

In theory it would actually be possible to develop a system which had
every meaning of every word in the standard (and all referenced
standards) and prove a lot more than this.

Only if you take "meaning" to mean what the system says it means,
rather than what people who read the standard think it means. I'm
not interested in what the program says, because it could say
whatever the person who wrote the program wanted it to say. I'm
interested in what meaning is understood by the people who read
the standard.

 

[lawrence.jones]

If the authors had meant to require char* and void* to be fully
interchangeable, they could easily have said so in normative text.

If we could have *easily* said so in the normative text, we would have.
But that turns out to be quite difficult to do without having unintended
effects on the type system, the type compatability rules, and all sorts
of other things (many of which we probably didn't even think of), which
is why we said it as strongly as possible in non-normative text and
trusted implementors to do the right thing.

 

[James Kuyper]

If we could have *easily* said so in the normative text, we would have.
But that turns out to be quite difficult to do without having unintended
effects on the type system, the type compatability rules, and all sorts
of other things (many of which we probably didn't even think of), which
is why we said it as strongly as possible in non-normative text and
trusted implementors to do the right thing.

I'd appreciate a more detailed explanation of what those unintended
effects would have been. It seems to me that, if you could not say this
more strongly without producing such effects, then that implies the
existence (at least potentially) of an implementation for which
mandating those effects would be a problem; on such an implementation,
it might be difficult "to do the right thing".