Debate Resource - Rationality

[Legion9191]

This is not necessarily a thread for debate. I simply want to introduce the extremely valuable resource of Rationality so that everyone can avoid common fallacies and mistakes that can turn mature debate into heated battles or nonsensical rambling.

 

All content either comes directly from, or is adapted from, Eliezer Yudkowsky's blog "Less Wrong." (http://lesswrong.com/)

It is a community blog in which Yudkowsky explores the science and applications of rationality.

 

All of the material there is extremely useful, if understood properly.

But its rather large, so here I will explain a few of the most important aspects so that you can sift through it at your leisure.

 

Because of the nature of cherry-picking topics, the order may not seem entirely logical.

 

1. Truth Nodes.

 

Assuming that a debate or argument is either a search for truth or optimized beliefs and/or plans, there exists a structure of nodes to accurately describe the relative importance of evidence of varying types.

 

Truth is the node at the center, upon which all other nodes converge.

Connected directly to truth is the node observation. Observation is the closest node to truth.

Connected to Observation is the node Argument.

Connected to Argument is Authority.

There are others, but they are either too abstract or too uncommon to be practical in this diagram.

 

What does it mean for Argument to be separated from Truth by Observation?

It means that Observation shears off Argument.

For example:

The argument at hand is whether or not sodium chlorate will have an energetic decomposition reaction with sugar.

The Truth being looked for is thus a simple "Yes, it will" or "No, it will not."

As debates generally progress, two people with different starting beliefs (One of the opinion of yes and the other of no) will begin by making arguments.

Lets say that person A, who is taking the "No" position, makes a very reasonable argument about why there is no reason sugar should cause the decomposition of sodium chlorate, a non-energetic decomposition by default, to become energetic.

Then, person B, puts sodium chlorate together with sugar, and a violent and energetic decomposition reaction occurs.

 

Person B made no argument. Person B still is closest to truth. Person B observed, they tested. Observation shears off argument, because no matter how good an argument may be, Observation will always be more definitive. (Assuming you know how to observer properly, and I will get to that in a bit)

 

Most people with any knowledge of the scientific method (and many who do not) already grasp the principle of testing hypothesis.

It becomes less intuitive at the next step. As Argument is separated from truth by Observation, Authority is separated by Argument. This means that Argument shears off Authority. This is something that many people do not grasp.

When faced with a debate, most people will defer to authorities on subjects of great complexity, usually waving off someone who is not an authority but presents an argument. (This by the way is actually a kind of Ad Hominem attack. http://plover.net/~bonds/adhominem.html )

 

Lets illustrate this with an example.

Bob is a trained and well known geologist.

Jake is your average 14 year old male.

 

Bob makes an unlikely-sounding assertion about rocks, and provides no argument.

Jake makes an equally unlikely assertion about rocks, and also provides no argument.

 

In this instance, it is best to assign a higher probability of truth to the geologist. Neither made an argument, (and there were no observations) so the closest available node to truth is Authority.

 

Lets say that then Jake presented a very reasonable-sounding argument, that had no logical fallacies.

Lets also say that Bob still did not make an argument.

 

Now, the highest probability of accuracy actually has to go to Jake, not Bob, because Argument shears off Authority, being a closer node to truth. (Source: http://lesswrong.com/lw/lx/argument_screens_off_authority/)

 

Observation is the closest node to truth, but making proper observations is difficult in and of itself.

Lets say I have a classic cartoon bomb: a metal ball with a wick.

I light the wick, after a few seconds, the bomb explodes.

A possible observation here is that lighting wicks causes explosions.

This is, obviously, a false observation.

The trick then in making proper observations can also be described in a node-diagram.

Though this one is quite a bit more complicated.

 

In the center of this node tree is the "correct" observation. (essentially Truth again)

Surrounding it are nodes that represent every aspect of the current universe.

The closer a node is to the center, the more direct the causality.

 

The goal is to remove as many nodes as possible from the node tree without removing any nodes that should still be there. Its kind of like minesweeper. 1 wrong move and the whole thing is ruined, so when in doubt you do not remove something. Removing even one node that should be there makes the observation inaccurate, and that makes it worse than useless.

 

For example, with the bomb-with-a-wick scenario.

At the center, the "truth" is that the chemical compounds contained within the metal bombshell cause a highly energetic combustion reaction when provided with sufficient activation energy. (The wick.) The node "combustion reaction" would be found very close to the center, whereas the node "The speed of light is a constant" would be fairly far away (But still in the causality chain) The node "Metamorphic rock" is totally unrelated, and while still in the node tree, does not have a connection to the center. It is an aspect of the universe that has nothing to do with that particular explosive device.

 

These unrelated nodes are the first things to be screened off. When you have screened off a node, you can also screen off all other nodes that lead only to that one. At the very edge of ever observation node diagram will be basic tenants of the universe. These fundamentals, however indirectly, are part of almost every causal chain. "electrons" is a good example of a node that is very far away from everything but part of pretty much every single causal chain. This is why you can only screen off nodes that lead to only the screened off node. You can screen off Metamorphic rock when talking about explosives. However, electrons lead to both explosives and metamorphic rock, so you cannot screen off electrons.

 

Once many nodes have been removed, the next step is to screen off the "implied" nodes. In the case of explosives, once you are down to combustion reactions, you do not still need electrons. You know that electrons is a pre-requisite of explosives, so you can remove it from the tree, making it implied.

 

The observation you make should represent every node left on the node tree.

The observation of "Lighting wicks causes explosions" is faulty because somewhere along the line the observer screened off combustion reactions when they should not have.

Essentially, make an observation as you would make a wish to a geni. Use precise language, and think of every loophole.

 

 

2. Applause lights.

 

One of the most common things for someone to do in an argument is speak with what Yudkowsky refers to as "Applause Lights." (http://lesswrong.com/lw/jb/applause_lights/) They are things that are said that do not actually bear any substance, but rather are meant to let the audience know when to clap. (Hence the name) A basic test to determine if something is an applause light is to reverse the statement and see if the product seems abnormal. If the inverse seems abnormal, then the original statement is probably normal, and adds no information to the argument. Here is a speech made entirely of applause lights: (Credit to Yudokowsky)

"I am here to propose to you today that we need to balance the risks and opportunities of advanced Artificial Intelligence. We should avoid the risks and, insofar as it is possible, realize the opportunities. We should not needlessly confront entirely unnecessary dangers. To achieve these goals, we must plan wisely and rationally. We should not act in fear and panic, or give in to technophobia; but neither should we act in blind enthusiasm. We should respect the interests of all parties with a stake in the Singularity. We must try to ensure that the benefits of advanced technologies accrue to as many individuals as possible, rather than being restricted to a few. We must try to avoid, as much as possible, violent conflicts using these technologies; and we must prevent massive destructive capability from falling into the hands of individuals. We should think through these issues before, not after, it is too late to do anything about them..."

 

All these things sound good, but they do not help anything. They offer nothing in the search for truth. Done knowingly, applause lights are employed to attempt to garner support from people who do not have a strong opinion or are particularly ignorant, or to seem to demonize someone disagreeing with your premise by making it seem as though they disagree with your applause lights. Done unknowingly, they add a lot of words to an argument without adding any substance, and will serve to distract from any actual specifics contained within.

They are also brilliant opportunities for the opposition to do the time-tested "quote each sentence of your argument and attack it individually, not necessarily in the context of the whole."

 

They are best always avoided.

 

3. Definitions and words.

 

For this, I seriously recommend you read the related sections on lesswrong.com. I know for a fact I cannot convey the specifics as well as Yudkowsky (as well as I know I do not understand them as well.)

However, it is of paramount importance to realize what exactly is being argued about in a debate.

Lets say, (back to bombs I'm afraid. ) that I was having an argument with my brother about whether or not mixing dry ice and water inside a plastic bottle will cause an explosion. I claim it will, he claims it does not. We make arguments, cannot agree, and so do the experiment. The water causes the dry ice to sublime, the vapor has a larger volume than the solid, the pressure builds, the bottle bursts.

I believe that the observation confirms my assertion: that mixing dry ice and water will cause the plastic bottle to explode.

My brother counters that by saying that the bottle did not explode, rather it burst. He claims that because there was no combustion, it cannot be called an explosion.

 

This scenario is extremely similar to scenarios that happen all the time. My brother and I spent time arguing about whether or not the bottle would explode. We thought we were arguing about chemistry and physics and the properties of the natural world, but after all is said and done, we find out we were just arguing about the definition of the word "explode." My argument worked on the premise that the event of the dry ice and water battle counted as an explosion, but the argument I was actually trying to make was whether or not the event would happen given the priors.

Therefore when in a debate, it is of great importance to establish the definitions of ambiguous or uncertain terminology well in advance, and be as precise as possible. If the argument does slip into arguing about the definitions of words, its important to recognize it early, and then cease argument and agree on a definition. Arguing about the definition of a word is not helpful in the search for truth. Arguing about the definition of "renewable" does not get anyone closer to the truth in an energy debate.

 

4. You want to find truth. You do not want to "win"

 

A debate is only useful when it is the means of pursuing truth. If you are not trying to pursue truth, debating is a bad idea. (There is of course and exception for recreational debating, but even then, only if it does not devolve into conflict.)

 

It is natural for us to think in a debate that arguments are soldiers. We must shoot at all of the enemies arguments and protect all of our own, otherwise we are stabbing our own soldiers in the back, or giving support to the enemy.

 

You must not do this. Arguments are not soldiers. There can be truth on both sides of an argument. Conceding a point you previously held or agreeing with a point that your opponent has made may feel like you are "losing" the debate, but it is only "losing" if the intention is not to find truth. By conceding to truth wherever you see it, you will work towards the intended goal of argument.

In the end, the man who comes out of an argument having learned something is the winner.

The person who was right just wasted a lot of time without gaining anything from it.

 

This litany, the Litany of Tarski, helps convey the principle:

 

If the box contains a diamond,
I desire to believe that the box contains a diamond;
If the box does not contain a diamond,
I desire to believe that the box does not contain a diamond;
Let me not become attached to beliefs I may not want.

—The Meditation on Curiosity

 

Thats all the input I have to give personally. If you would like to read more, I suggest starting here:

http://wiki.lesswrong.com/wiki/How_To_Actually_Change_Your_Mind

 

Happy Rationality everyone.

Edited August 1, 2013 by Legion9191

[Marxist ßastard]

It's funny because LessWrong website is ostensibly about rationality, but they take it to such extremes of zealotry that they end up sounding like a cult.

[Legion9191]

What exactly do you mean by that phrase, and can you provide examples?

[Marxist ßastard]

IDK, I just get a weird vibe from that place, but if you want to stay in that camp then I guess the best I can say is to go with your gut and feel it out.

[rizon72]

The problem I have with everything in the OP, is it works for smaller, simple things. The more complex the debate, the less you find a single truth supporting just one side. For example, five years ago when there was the huge debate about what to do with the economy (in the USA). There were many different sides, each with supporting 'facts' and other 'facts' to disprove someone else's solution. Unfortunately this is how most debates are.

 

I see it as, the larger and more complex a debate becomes, the more grey the truth becomes. Add in personal experiences and things become even more muddled.

[Hardwaremaster]

Interesting stuff might be useful depending on context but like the poster before me I would add. Here's the thing about information, the more dots you have to connect, the more freedom you have to connect them in anyway you want. The real world is complex enough to tell pretty much any story you please, regardless if the story is relative to each other. This makes the notion of facts a problem because opinion and imagination gets in the way. Now this becomes a right proper mess when someone lies for whatever reason. Blurring the problem even more to the point that you don't know that this or that is true. At best, you know that this person told you that this or that, and he had his reasons for doing so. I'll give it a thorough read through when I have the time. But in my opinion It all comes down to a debate I was having with a colleague of mine once: is data Subjective or Objective?

Edited August 6, 2013 by Hardwaremaster