satt comments on Wrong Questions - Less Wrong

34 Post author: Eliezer_Yudkowsky 08 March 2008 05:11PM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Article Navigation

Comments (126)

Comments (126)

Sort By: Old

You are viewing a single comment's thread. Show more comments above.

Comment author: satt 20 April 2013 11:52:18AM 1 point [-]

Or there could be a fourth explanation neither of us has thought of.

"There could be an (n+1)th explanation neither of us has thought of" is a fully general counterargument to any argument by cases.

Comment author: PrawnOfFate 20 April 2013 12:25:40PM 0 points [-]

It's valid too. Which is one reason not to put p=1.0 on anything.

Comment author: satt 20 April 2013 01:20:59PM 0 points [-]

Most fully general counterarguments are valid, taken at face value. This does not mean they're worth giving much weight. For example, someone could answer any argument I post on LW with "but satt, it's always possible you are wrong about that!" Which would be correct but rarely helpful.

Similarly, although I'm sympathetic to the idea of never assigning p=0 or p=1 to anything, any well-specified model I make is going to leave something out. So for me to make any inferences at all, I have to implicitly assign p=0 or p=1 to something. If I started throwing out models on that basis I'd have nothing left.

Comment author: MugaSofer comment score below threshold [+] (11 children)

Comment author: satt 23 April 2013 10:40:35PM 1 point [-]

You don't know why that objection is wrong? Because it is. It's not a valid argument we reject anyway, it's an invalid argument.

Alright, call it "invalid" and "wrong" if you like. I'm not trying to make some clever-clever semantic nitpick about the meanings of the words "valid" & "invalid"; I'm trying to communicate why "but satt, it's always possible you are wrong about that!" is all but useless to me.

I'll try it again without using the words "valid" & "invalid": although the BS,IAPYAWAT! counterargument is literally true (which gives it a veneer of reasonable-soundingness) it rarely tells me anything new, because when I post something on LW I usually already know I could be wrong. Being told BS,IAPYAWAT! isn't substantial evidence for me being wrong, because someone can just as easily say it whether I'm wrong or not.

There's an important difference between "valid" and "valid, taken at face value."

Exactly.

If the objection is invalid, answer it! Let your arguments screen off your labels.

OK, gimme a sec.

Comment author: MugaSofer comment score below threshold [+] (9 children)

Comment author: satt 28 April 2013 06:54:04PM 0 points [-]

Sat, we're not saying that the fact you might be wrong invalidates all arguments ever, because you can never be totally certain.

I know. Nonetheless, if someone says to me "there could be an (n+1)th explanation neither of us has thought of" without elaborating, that does amount to a but-you-might-be-wrong-about-that argument (even if not intended as such).

We're saying it invalidates the argument "X is wrong, therefore Y", unless you have a proof that X and Y are the only possibilities.

I don't have a proof, just the plausibility argument I gave earlier. A plausibility argument is not a proof, but this plausibility argument is so straightforward I find it pretty convincing.

Comment author: MugaSofer 29 April 2013 08:18:41AM -2 points [-]

But - it doesn't matter how plausible your axioms are if they give paradoxical results! Is it really more plausible that there's an invisible flaw in our reasoning than that we've failed to think of another possibility?

Hell, there are unlimited possibilities! It could be a duck, for example. That doesn't make any sense as an answer, but neither do those three answers. So why are we privileging them?

Comment author: satt 29 April 2013 11:10:28PM 0 points [-]

But - it doesn't matter how plausible your axioms are if they give paradoxical results!

One way I differ is that the results don't feel paradoxical to me. They feel a bit counterintuitive, but not so much so that my internal paradox alarm goes off.

Is it really more plausible that there's an invisible flaw in our reasoning than that we've failed to think of another possibility?

I'll bite that bullet, sure.

Some people find quantum mechanics paradoxical because it directly contradicts a deep intuition that any & every physical object necessarily has to have an unambiguous position & velocity. That philosophical intuition is simply false; it's a flawed insistence that the universe conform to a flawed induction. The right course of action is to throw out the intuition, not the axioms, despite the apparently paradoxical results.

I think I'm applying the same basic decision rule here: when a robust formalism clashes with an informal, inductive philosophical intuition, let the formalism bulldoze the intuition.

Comment author: MugaSofer 01 May 2013 05:54:29PM -1 points [-]

I don't think the "intuition" that we probably didn't make a mistake in our proof is analogous to the "intuition" that objects have unambiguous positions & velocities.

More to the point, I could say the same thing about your "intuition" that there are no other possible explanations for the universe.

continue this thread »

Comment author: MugaSofer 23 April 2013 10:26:55AM -2 points [-]

Why yes, yes it is. Arguing that someone else is wrong, therefore you are right is a well-known cheap debating trick.

Would you care to explain why I'm wrong, rather than sorting my argument into a low-status category?

Comment author: satt 23 April 2013 11:47:09PM 2 points [-]

Arguing that someone else is wrong, therefore you are right is a well-known cheap debating trick.

When I was complaining about the "but satt, it's always possible you are wrong about that!" argument, I wasn't complaining about all arguments that have "you are wrong, satt, therefore I am right" as a conclusion. I'm only taking issue with people mumbling "well, have you ever considered you might be wrong?" without elaborating. There's nothing wrong with someone arguing I might be wrong about something. But they should at least give a hint as to why I'd be wrong.

Would you care to explain why I'm wrong, rather than sorting my argument into a low-status category?

In this case, "there could be a fourth explanation neither of us has thought of" amounts to saying "there could be a fourth possible terminal state for a causal chain". Well, sure, it's always possible. But why should I assign that possibility any substantial probability?

Causal chains are pretty basic, abstract objects — directed graphs. I'm not talking about a set of concrete objects, where a fourth example could be hiding somewhere in the physical world where no one can see it. I'm not talking about some abstruse mathematical object that's liable to have weird properties I'm not even aware of. I'm talking about boxes connected by arrows. If there were some fourth terminal state I could arrange them to have I'd expect to know about it.

What I've just said might be mistaken. But you haven't given any specifics as to where or how it goes wrong, so your comment is just another form of "but satt, it's always possible you are wrong about that!", which doesn't help me.

Comment author: MugaSofer comment score below threshold [+] (3 children)

satt comments on Wrong Questions - Less Wrong

Article Navigation

Comments (126)

Nearest Meetups

Virtual Study Room

Recent Comments

Recent Posts

Recent Rationality Quotes

Recent Wiki Edits

Top Contributors, 30 Days

Recent Karma Awards

Comments (126)

Top Contributors, 30 Days

You'll need to login or register to do that

(Don't worry, it only takes a few seconds)

Create

Login