[YESNO] What's wrong with arguments having an amount of strength? ("Check Your Understanding" question)

[Alisa Zinov'yevna Rosenbaum]

*Yes or No Philosophy*, part 17, "Check Your Understanding":

> What's wrong with arguments having an amount of strength?

For arguments, "amounts of strength" other than 0% or 100% would be irrelevant. Here's why. The "amount of strength" of an argument is supposed to reflect the amount of *epistemic weight* or *support* that the argument brings to bear on its conclusion. But all arguments can be expressed as negative arguments, and a negative argument either refutes a particular idea or it doesn't. There are no intermediate statuses between refuted and unrefuted. Therefore, for an argument, the only relevant amounts of strength are binary: does the argument refute a particular idea or not?

Another issue with arguments having an amount of strength is that the combined strengths of different arguments is unclear. For example, if I vary an argument by a tiny amount, does the variation have a similar strength? If so, can I make up a bunch of trivial variations so that the combined strength of all those arguments add up to significantly more than the strength of the original?

[Max Kaye]

On 18/06/2020 1:56 pm, Alisa Zinov'yevna Rosenbaum wrote:
> *Yes or No Philosophy*, part 17, "Check Your Understanding":
>
>> What's wrong with arguments having an amount of strength?
>
> For arguments, "amounts of strength" other than 0% or 100% would be irrelevant.

I would replace "would be irrelevant" with "is meaningless" or "is
irrelevant to figuring out if the actual answer is false or true". The
way you've put it doesn't make it clear (to me) that the reasons might
still be relevant (a Bayesian might say GR has a 90% chance of being
correct), and it's the attempt at percent-erizing them that is the
meaningless part.

I'm going to use %probability because ppl who make the mistake of
assigning % strength sorta measurse usually wrap that up with
probability, plus Bayesians are not uncommon, and I don't know what "%
strength" even _can_ mean.

As an example (also I've just come up with this argument 😕): a %
probability would be theory laden. Ignoring the fact objective truth
exists so the only valid answers are 0 or 100% anyway, to calc it you'd
need a deep and far reaching explanation of explanations (because
probability is counting and/or comparison across some population of
things). but we know some things about knowledge that preclude this:
like there are infinitely many ideas (both correct and incorrect), so
you not only need this uber-good explanation of explanations (better
than BoI) but it also needs a new *measure* for comparing infinities of
ideas, which also probably means you need to understand things about the
ways ideas can be encoded and any quirks of encoding (these things we
don't really know), and your other theories need to be able to control
for effects of the encoding on the probability (like english isn't a
good way, are all words equally probably in true explanations? etc).

example: let's presume we can represent all ideas as some arrangement of
all whole numbers > 0. So encoded in the order of elements of the set
{1, 2, 3, 4, ...}. Say every good idea you have to test starts with
_only_ odd numbers in the first 4 terms: {1, 3, 5, 7, ...}, {1337,
289348577, 3, 19, ...}, etc. Can you make any conclusions about what
role odd numbers play? Do odd numbers at the start make it _more_ likely
some encoded idea is true? Is it an artifact of the way you express
them? Or do all ideas actually have some equal and equivalent form, a
"pseudo-reciprocal" form; some _other_ encoding where all odd numbers
were even and vice versa?

This rabbit hole ends up at an oracle: to be accurate you need to - in
essence - understand truths so deep (not to mention contradictory to the
enterprise) that you wouldn't care about the question in the first
place. Furthermore: such an oracle would necessarily need to be error
prone (because objective reality exists), which means that it would only
ever be measuring its own error anyway, so it's just as meaningful as
picking randomly! (because to know what it's error is requires a theory
of it and we're just back to 'turtles all the way down' type stuff).


In essence: knowing about % support or % probability is the same problem
as an oracle that's right 100% of the time. Moreover, even if you could
calculate it, the same problems follow i.e. it doesn't help you
understand _why_ something is right, and you still need to do that to
use the knowledge. I think DD covers this in ch1 of fabric (ofc there's
plenty in BoI too).

> Here's why. The "amount of strength" of an argument is supposed to reflect the amount of *epistemic weight* or *support* that the argument brings to bear on its conclusion. But all arguments can be expressed as negative arguments, and a negative argument either refutes a particular idea or it doesn't. There are no intermediate statuses between refuted and unrefuted. Therefore, for an argument, the only relevant amounts of strength are binary: does the argument refute a particular idea or not?

I haven't heard this before. It's nice.

> Another issue with arguments having an amount of strength is that the combined strengths of different arguments is unclear. For example, if I vary an argument by a tiny amount, does the variation have a similar strength? If so, can I make up a bunch of trivial variations so that the combined strength of all those arguments add up to significantly more than the strength of the original?

Also: you could easily contrive situations where the strengths add up
more than 100% - what does that mean? (Or worse: probability)

Maybe you multiply instead of adding, but two theories that are mutually
consistent (!) but not mutually implying and each has 50% strength (what
does that even mean?), or rather 50% probability of being right:
presuming we take that assumption that this is the case: what do we do?
Does it have 50% chance of being right? 100% doesn't seem right, 25%
doesn't make sense either. Any ofc we can't talk about it in much more
detail because we don't have a good theory of ideas that allows such a
question to be valid!




----
time: 30min

[Elliot Temple]

On Jun 17, 2020, at 8:56 PM, Alisa Zinov'yevna Rosenbaum <petrogradp...@gmail.com> wrote:

> *Yes or No Philosophy*, part 17, "Check Your Understanding":
>
>> What's wrong with arguments having an amount of strength?
>
> For arguments, "amounts of strength" other than 0% or 100% would be irrelevant. Here's why. The "amount of strength" of an argument is supposed to reflect the amount of *epistemic weight* or *support* that the argument brings to bear on its conclusion. But all arguments can be expressed as negative arguments, and a negative argument either refutes a particular idea or it doesn't. There are no intermediate statuses between refuted and unrefuted. Therefore, for an argument, the only relevant amounts of strength are binary: does the argument refute a particular idea or not?

You’re assuming negative arguments are binary. This is basically begging the question.

You’re assuming my positive-negative translatability claim.

I don’t really see anything here but assertions.

> Another issue with arguments having an amount of strength is that the combined strengths of different arguments is unclear. For example, if I vary an argument by a tiny amount, does the variation have a similar strength? If so, can I make up a bunch of trivial variations so that the combined strength of all those arguments add up to significantly more than the strength of the original?

This is a reason instead of just an assertion. But it’s answerable, not decisive.

Elliot Temple
www.curi.us

[Alisa Zinov'yevna Rosenbaum]

On Sat, Jun 20, 2020 at 01:18:41PM -0700, Elliot Temple wrote:

> This is basically begging the question.

1. Verb: "is begging" (present continuous tense)

2. Subject: "This"

3. Object: "question"

4. Modifiers:

- "the" modifies "question".

- "basically" modifies "begging".

http://www.websters1913.com/words/Beg :

> Beg (?), v. t.

> To take for granted; to assume without proof.

> To beg the question, to assume that which was to be proved in a discussion, instead of adducing the proof or sustaining the point by argument.

Conclusions: [The argument] more or less assumes something that it was supposed to prove.

[Alisa Zinov'yevna Rosenbaum]

On Sat, Jun 20, 2020 at 01:18:41PM -0700, Elliot Temple wrote:

> On Jun 17, 2020, at 8:56 PM, Alisa Zinov'yevna Rosenbaum <petrogradp...@gmail.com> wrote:

>> *Yes or No Philosophy*, part 17, "Check Your Understanding":

>>> What's wrong with arguments having an amount of strength?

>> ... the combined strengths of different arguments is unclear. For example, if I vary an argument by a tiny amount, does the variation have a similar strength? If so, can I make up a bunch of trivial variations so that the combined strength of all those arguments add up to significantly more than the strength of the original?

>
> This is a reason instead of just an assertion. But it’s answerable, not decisive.

As far as I'm aware, the idea of arguments having an amount of strength doesn't contain a refutation of the following argument, which concludes that comparing ideas by amounts of support is meaningless.

Suppose that:

- arguments have strength

- we can determine whether an argument supports an idea and we can quantify the strength of that support as a positive number

- the amount of support that an idea has is the sum of the strengths of the arguments that support it

Then we can give any idea an arbitrarily large amount of support, provided we know of at least one argument that supports it. Just vary that argument so that the result is different from the original argument but the result still supports the idea. You could do this by adding things to the argument like "and 2+2=4", "and 3+3=6", and so on, to the original argument. The result would a different argument that still supports whatever idea the original argument supported. You can create as many of these variants as you need to get an arbitrarily large amount of support for any idea. Therefore, comparing ideas by amount of support is meaningless.

[Alisa Zinov'yevna Rosenbaum]

On Sat, Jun 20, 2020 at 01:18:41PM -0700, Elliot Temple wrote:

> On Jun 17, 2020, at 8:56 PM, Alisa Zinov'yevna Rosenbaum <petrogradp...@gmail.com> wrote:

>> *Yes or No Philosophy*, part 17, "Check Your Understanding":

>>> What's wrong with arguments having an amount of strength?

>> For arguments, "amounts of strength" other than 0% or 100% would be irrelevant. Here's why. The "amount of strength" of an argument is supposed to reflect the amount of *epistemic weight* or *support* that the argument brings to bear on its conclusion. But all arguments can be expressed as negative arguments, and a negative argument either refutes a particular idea or it doesn't. There are no intermediate statuses between refuted and unrefuted. Therefore, for an argument, the only relevant amounts of strength are binary: does the argument refute a particular idea or not?
>
> You’re assuming negative arguments are binary. This is basically begging the question.
>
> You’re assuming my positive-negative translatability claim.
>
> I don’t really see anything here but assertions.

Yeah. My answer was poorly done. And it was just assertions.

I also answered the same YESNO "Check Your Understanding" question poorly back in May 2020. I [wrote](https://groups.google.com/g/fallible-ideas/c/rxqL8u7f-XA/m/2q5cZKVmAQAJ):

>> There is no rational way to evaluate arguments as having non-binary amounts of strength, i.e., evaluations other than *decisive* or *worthless*.

Another assertion.

Elliot [replied](https://groups.google.com/g/fallible-ideas/c/rxqL8u7f-XA/m/pqpcdsJnAQAJ):

> Why?

I never answered.