-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathResolving-Confusion.html
243 lines (231 loc) · 23.4 KB
/
Resolving-Confusion.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
<?xml version="1.0" encoding="UTF-8" ?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />
<link rel="icon" type="image/png" href="./images/favicon-32x32.png" sizes="32x32" />
<link rel="icon" type="image/png" href="./images/favicon-16x16.png" sizes="16x16" />
<title>Resolving Confusion - SPK's Rationality Essays</title>
<link rel="stylesheet" type="text/css" href="./css/default.css" />
<link rel="stylesheet" type="text/css" href="./css/highlight.css" />
<!-- <script src="http://ajax.googleapis.com/ajax/libs/jquery/1.10.2/jquery.min.js"></script> -->
<!-- <script type="text/javascript" src="/js/header-links.js"></script> -->
<script type="text/javascript" src="http://cdn.mathjax.org/mathjax/latest/MathJax.js?config=TeX-AMS-MML_HTMLorMML"></script>
<link href="atom.xml" type="application/atom+xml" rel="alternate" title="Sitewide ATOM/RSS Feed" />
<!-- Google Analytics stuff -->
<!-- Google tag (gtag.js) -->
<script async src="https://www.googletagmanager.com/gtag/js?id=G-DEWF2J5BG8"></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag(){dataLayer.push(arguments);}
gtag('js', new Date());
gtag('config', 'G-DEWF2J5BG8');
</script>
<script type="text/javascript" src="https://fast.fonts.net/jsapi/f7f47a40-b25b-44ee-9f9c-cfdfc8bb2741.js"></script>
</head>
<body>
<div id="header">
<div id="logo">
<a href="./">SPK's Rationality Essays</a>
</div>
<div id="navigation">
<a href="./">Home</a>
<a href="./notes.html">Notes</a>
<!-- <a href="/about.html">About</a> -->
<a href="./archive.html">Archive</a>
<a href="./atom.xml" type="application/atom+xml" rel="alternate" title="Sitewide ATOM/RSS Feed">RSS</a>
</div>
</div>
<div id="content">
<h1 id="post-title">Resolving Confusion</h1>
<!-- <center><img src="https://fbcdn-sphotos-d-a.akamaihd.net/hphotos-ak-prn1/t31.0-8/p600x600/10257116_10202295769100492_2438594605053717342_o.jpg" height="400" width="300" class="sujeet-pic" alt="Sujeet pic" /></center> -->
<p><strong>Entry Question</strong>: How do we get rid of Confusion about some topic?</p>
<hr />
<h1 id="why-do-we-want-to-resolve-confusions">Why do we <em>want</em> to Resolve Confusions?</h1>
<ul>
<li><p>More Predictive Power means more Win.</p></li>
<li><p>Confusion means you’re getting lower Predictive Power than you think you should.</p>
<p>Worse, you don’t even realize it.</p>
<p>Here, we will see how to Notice Confusion - places where you don’t actually know as much as you think you do. Now, you will be able to see the places where you actually suck.</p>
<p>Then, we will see how to Resolve Confusion, so that you can know more and thus Win.</p></li>
<li><p>When you are Confused, you won’t even realize what the events in the Territory are.</p>
<p>You think you know all the events, but actually there are many events you don’t know about, for which you consequently make very poor predictions (more or less random predictions or whatever the human default behaviour is).</p>
<p>i.e., You don’t know all the various scenarios that your solution must handle. Your solution doesn’t handle all the problem requirements. Naturally, if you don’t design your solution to handle those scenarios, you’re most probably gonna suck over there. No magic fairy is gonna bail you out.</p></li>
<li><p>When you are Confused, you won’t be able to make accurate predictions about important events</p>
<p>You won’t have the deep models you need to make precise and accurate predictions.</p></li>
<li><p>When you are Confused, you won’t be able to update on evidence and learn usefully from experience</p>
<p>So, learn how to Notice and Resolve Confusion. Else, you’ll go to Bayesian Hell.</p></li>
<li><p>When you are Confused, you won’t be able to tell what was the cause of the outcomes that you see in life</p>
<p>How will you fix things if you don’t know what caused the problem in the first place?</p></li>
</ul>
<p>So, you’re Confused</p>
<ul>
<li>when you think your problem is <em>itself</em> completely impenetrable and confusing,</li>
<li>when you are getting poorer results than you think you should,</li>
<li>when you don’t know all the events that your solution must handle,</li>
<li>when you aren’t able to make accurate predictions about the events,</li>
<li>when you aren’t able to improve performance even after getting lots of evidence, or</li>
<li>when you aren’t able to tell the cause of some particular observation and thus aren’t able to fix things.</li>
</ul>
<h1 id="what-causes-confusion">What causes Confusion?</h1>
<p>What causes us to have Confusion? Why don’t we throw out the beliefs that don’t pay rent? Why did they enter in the first place?</p>
<p>Curiosity-stopper</p>
<p>Naive Realism - You think that reality itself is that confusing and that there is no answer, no way to correctly predict the outcomes.</p>
<p>Test: If you have Confusion without Naive Realism, you will dispel the Confusion very soon.</p>
<p>i.e., if you believe that your Map is different from the Territory, you will try to figure out where your beliefs are lacking and try to fix them or throw them away.</p>
<p>Test: If you are Confused and not making progress, you must be falling prey to Naive Realism</p>
<p>All the stuff that keeps Naive Realism going - Confirmation Bias, etc.</p>
<p>Could Motivation be a big factor? Could the fog surrounding a topic come because of low Motivation for it?</p>
<p>What heuristics and biases are causing and sustaining our confusion? They must have something to do with it (given that they are the reasons why we don’t think rationally even with our bounded resources).</p>
<h1 id="problem-due-to-confusion-hopelessness">#1 Problem due to Confusion: Hopelessness</h1>
<p>Without knowing that Confusions are not in the Territory and are only in your Map, you won’t even realize that your problem can be solved and understood. You will think that your problem is impenetrable.</p>
<p>When you understand how Confusion works, you will realize that either your problem has a solution or there exists a proof that it is impossible to solve. One or the other. But, you will know either way.</p>
<p>You won’t sit around with a false understanding.</p>
<p>This will give you hope and the energy to take action on the problem.</p>
<h1 id="consequences-of-confusion">Consequences of Confusion</h1>
<p>You can’t predict: You have no anticipation-controlling things in your mind for the events. You just stand there confused.</p>
<p>Potential Test for Confusion: You won’t be able to make any (quantitative?) predictions!</p>
<p>You will just stand there with a blank expression. Nothing will come. There is no machinery below that lever in your mind.</p>
<p>You can’t update: You have nothing to falsify or strengthen on the basis of evidence, cos nothing made predictions either way.</p>
<p>What do you do when something goes against your anticipation and you notice it?</p>
<p>You just throw in another label to “explain” that particular outcome.</p>
<p>Or, you stretch your existing labels to be able to “explain” that outcome</p>
<p>Can be only one of the above :P</p>
<p>Potential Test: You won’t be able to improve your model usefully.</p>
<p>You can add more symbols, but you won’t see the implication of those concepts put together.</p>
<p>You can’t do Backward Inference: When the teacher asks, why it’s hot, you can’t give any actual causes. You just Guess the Teacher’s Password (“it’s hot BECAUSE… of heat conduction”). You bring up labels that you have heard of in the past and put in your Map.</p>
<h1 id="beliefs-must-pay-rent">Beliefs must pay Rent</h1>
<p>Ask each Belief to pay rent on its own. It should make fresh constraints on anticipation that you couldn’t have made earlier.</p>
<p>Also, don’t argue about beliefs. Argue about anticipations.</p>
<blockquote>
<p>If a tree falls…</p>
</blockquote>
<p>Tree falling in lonely forest -> makes sound</p>
<p>Tree falling in lonely forest -> doesn’t make sound</p>
<p>The two “beliefs” make the exact same anticipations about reality.</p>
<h1 id="how-to-notice-confusion">How to Notice Confusion</h1>
<p>Before you get some sort of information or answer from others, Predict</p>
<p>Don’t fit your model to the answer somehow</p>
<p>Don’t try to explain it somehow</p>
<p>Just admit that you are confused</p>
<p>The general situation: You think you have a correct model but something happens that contradicts your anticipation. You now don’t think that your model is wrong. You just try to do some mental gymnastics to make your prediction fit the outcome, WITHOUT changing your model.</p>
<p>aka: “The rod is hotter BECAUSE…” followed by a meaningless sequence of words.</p>
<h2 id="example-the-hot-rod-and-the-physics-students">Example: The hot rod and the physics students</h2>
<p>The teacher asks you <em>why</em> this is so</p>
<p>Loaded question</p>
<p>Ask her if the rod has been near the fire for long</p>
<p>If so, you have NO explanations for why it is cold at that end</p>
<p>In other words, Predict what you should have seen</p>
<h1 id="mysterious-answer">Mysterious Answer</h1>
<p>What is a Mysterious Answer?</p>
<p>[Link to Eliezer]</p>
<p>Not having any moving parts is a Mysterious Answer Smell</p>
<p>I think because you have no way of making predictions apart from the predictions that came along with the black box.</p>
<p>You could have black boxes that pay some rent (predictive power) but about whose insides you don’t know. As long as you realize that it is possible for you to understand what is inside, it’s not a problem. Maybe the amount of predictive power it gives you is sufficient for your purposes.</p>
<h1 id="techniques">Techniques</h1>
<p>List out the Events in the Territory.</p>
<p>Break the big bad thing into smaller chunks.</p>
<h1 id="replacing-symbol-with-the-substance">Replacing Symbol with the Substance</h1>
<p>What does that mean?</p>
<p>I think replacing symbol with substance means replacing a label (“Confusion”) with the actual anticipation-constraints over the various events.</p>
<p>Instead of writing a definition that is useless, you identify the exact behaviour you care about.</p>
<p>Why is this useful? Well, if you come to know that some concept X is true, then because you have listed out the anticipation-constraints over the various events, you can easily make all the predictions that X would make.</p>
<h1 id="lost-purposes">Lost Purposes</h1>
<p>The main problem is you <a href>lose your purpose</a> when you think only in terms of the symbol.</p>
<p>You think you’re going to “school” to get an “education”, when in actual detail, you’re gonna sit in classes you’re not interested in and memorize just enough to get through the final exams.</p>
<p>It’s not that you don’t know about those events. You do. So, that’s not a problem. You’re not really confused (I think).</p>
<p>The problem is you think getting an “education” is gonna help you achieve your life goals by giving you knowledge. That right there is the belief that isn’t paying rent. That’s where the Lost Purpose comes in. If you could just remove that one belief, you would immediately realize that you aren’t really gaining helpful knowledge the way you want to. You would immediately either start learning through some other means (personal projects, whatever) or you would accept that it is the final “degree” and the attendant job offers that you desire from your “education”.</p>
<p>In other words, I’m saying that <a href>Lost Purposes</a> come about when you have <em>one belief</em> that doesn’t pay rent - the belief that this activity is going to help you reach your goal. All your other beliefs could be lawful, rent-paying citizens of your mind. You would correctly anticipate the other benefits and costs of that activity. You would just fail to see that it is not helping you reach the original goal.</p>
<blockquote>
<p>Lost Purpose = Confusion about the Benefit</p>
<p>You think it will help you achieve some goal, but it won’t.</p>
</blockquote>
<p>This is an important class of confusions. Here, you think some activity is helping you achieve some goal, but it actually isn’t. This is why we are often irked (and maybe enlightened) when some original-seeing person - maybe a kid or a foreigner or a country-bumpkin or whatever - asks us basic questions about <em>why</em> on earth we are doing some “perfectly obvious” task.</p>
<p>This is the Art of Questioning Basic Assumptions.</p>
<p>“Why do people wear ties?” - We’re aware that there are status benefits to doing so. But we think these are just side-benefits. The main reason we wear ties is because… ??? If you’re unusually advanced, you would have not come up with any Fake Explanations there. But, we do feel tempted to come up with plausible reasons.</p>
<p>“Why do I have to go to school and sit in classrooms for 15 years?” - Paul Graham <a href>argues</a> that the main reason for sending kids to school is to keep them occupied while we are away at work (plus, be eligible to attend college). We realize that that is a definite benefit. But, it seems like a side-benefit. The main reason we send kids to school is… ??? We feel tempted to come up with all kinds of reasons. Knowledge, character development, association with kids of their own age (what? they wouldn’t have met them otherwise?), etc. Even when we find out that kids are super-bored in there and don’t have any real basis for deciding hierarchies and thus end up having lord-of-the-flies schools… we still say, “No… it’s still worth it. They have to get educated.”</p>
<p>Our belief that going to school will give us knowledge, character, and whatnot is a belief that doesn’t pay rent but seems like it does. We don’t <em>really</em> expect to see the massive improvements in knowledge, character, and so on that would actually have come about if school were meant to increase them. But we’re confused, so we don’t notice that surprising absence of improvement. Our confusion that school will help us achieve our knowledge goals is the cause of the Lost Purpose there.</p>
<blockquote>
<p>Purpose is lost whenever the substance (learning, knowledge, health) is displaced by the symbol (a degree, a test score, medical care). To heal a lost purpose, or a lossy categorization, you must do the reverse:</p>
<p>– <a href="http://lesswrong.com/lw/nv/replace_the_symbol_with_the_substance/">Replace the symbol with the substance</a>, Eliezer Yudkowsky</p>
</blockquote>
<p>To discover that you have lost your purpose - that this activity isn’t actually helping your goal - you <em>need</em> to see through to the actual events and your predictions for them.</p>
<p><strong>Note</strong>: By explaining Lost Purposes in terms of Confusion regarding the Benefit, I make a <em>differing prediction</em> here from Eliezer. Unless seeing the events and their actual outcomes vs our predicted outcomes is a cure for confusion, just replacing Symbol with Substance will not get rid of Lost Purposes. You are <em>confused</em> here, not merely wrong. Just telling you that you were mistaken is not gonna get rid of your confusion.</p>
<p>Example: When some kid points out the futility of many of your actions, or when a foreigner questions your basic assumptions, do you immediately forswear that action and change your ways? I don’t think so.</p>
<p>I get the suspicion that most people don’t <a href>change</a> their mind despite getting proof that they’re wrong because of Lost Purposes. Maybe I’m over-reaching here, I don’t know. You may not be able to change your actions, due to <a href>Akrasia</a>, cool, but you have to at least <em>admit</em> you’re wrong.</p>
<p>If I’m right, then, to resolve Lost Purposes, you need to get rid of the Confusion.</p>
<p>However, counter-point: There do seem to be occasions when just replacing symbol with substance seem to make you aware that your action is not achieving the goal you want. For example, with “school” and “education”, or with “baseball” (in the original article), and so on - once they’re pointed out to you, you can probably see the silliness of many things and the fact that they don’t really do anything.</p>
<p>What’s happening there? I think the answer is that there you’re not really confused about the topic. (Isn’t that almost by definition? No True Scotsman, anyone?)</p>
<p>Well, let’s be technical then. I predict that when people with no confusion about the topic (say people who are bored in school and have an incentive to question the assumption that it is useful, or people who are looking back a decade later, after realizing that all those things you worried about in high school never really mattered) question the basic assumptions, they will be able to accept the lost purpose. People who are confused (maybe because they are invested somehow in believing that school is good - parents with kids, kids who can’t or don’t want to question the world around them, administrators) will not be able to accept the lost purpose even if they are given evidence.</p>
<p>Note: I predict that people who are coming in with motivated skepticism (the bored kids) or motivated stopping (the parents who have invested a lot in their kids’ education) will not be able to accept the opposite of their belief. Bored kids may not be able to accept some of the actual benefits.</p>
<p>So, I guess that leaves only the near-objective adult viewer into the past who is not confused about the topic. He or she can accept the evidence as it is.</p>
<p>Maybe it’s about <strong>identity</strong>. The more something is <a href>part of your identity</a>, the more you have a chance of being confused about it (?). You would want to protect that belief, even though it doesn’t pay rent. You would end up believing that it <em>does</em> pay rent and refuse to acknowledge any evidence to the contrary.</p>
<p><strong>Exercise</strong>: Question the assumptions behind your actions. Will doing this really get me what I want?</p>
<h1 id="the-problem-with-definitions">The Problem with Definitions</h1>
<p>We have the need to define ideas several times - in essays, in our own proofs, and so on.</p>
<p>What is Planning? What is Confusion? What is a Hard Problem? What is Akrasia? What is Truth?</p>
<p>Well, that is done easily enough. We saw that we need to replace the Symbol by the Substance - we need to talk about the relevant events and the predictions made for those events by this hypothesis. We can come up with a “definition” for something using this.</p>
<p>However, that’s not enough. Why? Because one of the most pernicious problem with informal thinking is the tendency to ignore alternative hypotheses.</p>
<p>We say that if this is X, then it will have Y and Z properties. Lo and behold, it has Y and Z properties. Therefore, it is an X.</p>
<p>For example: If you are confused, then you won’t be surprised by anything. I tell you a wrong answer. You are not surprised. Therefore, you are confused! Yup!</p>
<p>Right? No! You could have a wrong model that gives that wrong answer and so are not surprised. You are definitely not confused.</p>
<p>I had failed to consider the alternative hypothesis - you could have a wrong model. (There is one more hypothesis, you could have a right model).</p>
<p>What we need to do is consider two or more hypotheses - Confused vs Not Confused, Planning vs Not Planning, Hard Problem vs Not Hard Problem, Truth vs Not Truth.</p>
<p>Instead of looking at the predictions made by one hypothesis and checking whether they’re true, you should look at the <em>differing predictions</em> between the hypotheses and check out who wins there. The other criteria don’t matter at all.</p>
<p>Don’t ask “Am I confused?”.</p>
<p>Ask: “How does a Confused person differ from a non-Confused person?”</p>
<p>Now, put this together with what we learned in replacing Symbol with Substance - talk about the differing predictions for specific events.</p>
<p>Judge whether something is an X or not X by seeing its outcomes in the events where X and not X differ. Resist the temptation to label items based on definitions.</p>
<p><strong>Major Thought Smell</strong>: Talking about “Is this an X?” without mentioning the alternative hypotheses.</p>
<p>Say NO to Definitions!</p>
<hr />
<p><strong>Note</strong>: You can check the predictions of only one hypothesis as a sanity check. If something it predicts doesn’t happen, it is a goner. But to find out how true it is, you <em>need</em> to look at alternative hypotheses. It is meaningless to talk of evidence without alternative hypotheses.</p>
<h1 id="what-am-i-confused-about">What am I confused about?</h1>
<ul>
<li>Math problems (never got 100s consistently)</li>
</ul>
<h1 id="concrete-list-of-confusions-past-and-present">Concrete List of Confusions (past and present)</h1>
<ul>
<li>AI researchers and intelligence</li>
<li>Me and Akrasia</li>
<li>Consciousness</li>
<li>Me and Science (before my Scientific Enlightenment)</li>
</ul>
<h1 id="summary">Summary</h1>
<ul>
<li><p>Confusion means you have lower Predictive Power than you think you do</p></li>
<li><p>When you’re confused, you think your problem is impenetrable and cannot be solved easily. You feel it is hopeless, so you don’t even take action.</p></li>
<li><p>When you’re confused, you don’t know all the requirements that your solution must handle. So, you end up with poor results.</p></li>
<li><p>When you’re confused, you can’t make accurate predictions about events.</p></li>
<li><p>When you’re confused, you aren’t able to improve performance even after getting lots of evidence</p></li>
<li><p>When you’re confused, you aren’t able to tell the cause of some observation, and thus you can’t fix your problems.</p></li>
</ul>
<hr />
<div class="info">Created: November 15, 2014</div>
<div class="info">Last modified: January 22, 2015</div>
<div class="info">Status: finished</div>
<div class="info"><b>Tags</b>: confusion, hard problems, impact, anticipation constraint</div>
<br />
<div id="disqus_thread"></div>
<script type="text/javascript">
/* * * CONFIGURATION VARIABLES: EDIT BEFORE PASTING INTO YOUR WEBPAGE * * */
var disqus_shortname = 'spkrationalitytrainingground'; // required: replace example with your forum shortname
var disqus_identifier = '/Resolving-Confusion.html';
var disqus_title = 'Resolving Confusion';
/* * * DON'T EDIT BELOW THIS LINE * * */
(function() {
var dsq = document.createElement('script'); dsq.type = 'text/javascript'; dsq.async = true;
dsq.src = '//' + disqus_shortname + '.disqus.com/embed.js';
(document.getElementsByTagName('head')[0] || document.getElementsByTagName('body')[0]).appendChild(dsq);
})();
</script>
<script type="text/javascript" src="https://fast.fonts.net/jsapi/f7f47a40-b25b-44ee-9f9c-cfdfc8bb2741.js"></script>
<noscript>Please enable JavaScript to view the <a href="http://disqus.com/?ref_noscript">comments powered by Disqus.</a></noscript>
<a href="http://disqus.com" class="dsq-brlink">comments powered by <span class="logo-disqus">Disqus</span></a>
</div>
<div id="footer">
Site proudly generated by
<a href="http://jaspervdj.be/hakyll">Hakyll</a>
</div>
</body>
</html>