-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathThe-Big-Problems-Reloaded.html
211 lines (199 loc) · 19.4 KB
/
The-Big-Problems-Reloaded.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
<?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>The Big Problems: Reloaded - 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">The Big Problems: Reloaded</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>: What are the important problems we need to solve? What are the areas of life we need to optimize most?</p>
<hr />
<h1 id="the-more-important-questions">The More Important Questions</h1>
<p>Let’s revisit Richard Hamming’s classic questions:</p>
<ul>
<li><p>What are the most important problems in your field?</p></li>
<li><p>Are you working on one of them?</p></li>
<li><p>Why not?</p></li>
</ul>
<h1 id="overall-aim">Overall Aim</h1>
<p>My overall aim is to maximize the amount of utilons I have. How do I get utilons? By doing things that improve my world. That can involve eating tasty food (but staying healthy), partaking in enjoyable activities, becoming better at stuff I like doing, etc.</p>
<p>The more I do those things, the more utilons I will get.</p>
<p>However, that stuff doesn’t scale. There’s only so much food I can eat in a day, or only so many movies I can watch. As Martin Seligman says, that shit satiates really quickly. I will get bored of even the most enjoyable activities.</p>
<p>How else can I get utilons? Is there any way to get utilons that isn’t limited by my appetite or energy level? I thought you’d never ask.</p>
<p>Altruism scales. I want other people to live good lives. That is part of what makes the world a good place. The trick here is that there are a <em>lot</em> of other people, and because of f*cked-up current situations, it needn’t take much money to save a person from a despicable fate. There is a tremendous amount of utilons to be gained from saving even a single life. The world is a much better place with that person alive. And there are a lot of people to be saved. So, if you <em>multiply</em> the amount of utilons you get from helping a person with all the people you can help, you can stack up a huge number of utilons. Even though it pains me to say this, there are a lot of people in the world suffering. Death, disease, old age - these motherf*ckers get us all.</p>
<p>Anyway, how do we go and help people? No, not by helping out part-time at some charity. Or, by donating old clothes. F*ck that shit.</p>
<p><a href>Money is the unit of caring</a>. ’nuff said.</p>
<p>We need to get as much money as possible so that we can donate it to effective charities that will use it to save and improve as many lives as possible.</p>
<p>Another way is to solve important technical problems. If, for example, there is a way to improve the efficiency of life-improving methods, then making it happen is valuable. For example, making some cure cheaper, increasing the productivity of people who work in these areas, inventing cures for diseases, helping people see the f*cking important problems in this world, etc.</p>
<h1 id="brass-tacks">Brass tacks</h1>
<p>What are some good ways to make money? The essential answer to that, as Eliezer Yudkowsky points out, is to find out where your comparative advantage lies and then specialize in it and make lots of money. Comparative advantage, from economics, refers to that domain in life where you can make more money (or utilons or whatever) than in any other domain. Note, this is different from a domain where you can make more than <em>other</em> people. The aim is to maximize utilons.</p>
<p>What are some technical problems that need to be solved? Let’s take this on in the remainder of this essay.</p>
<h1 id="the-big-problems-hypothesis">The Big Problems Hypothesis</h1>
<p>As I’ve written earlier, I think that there a few Big Problems that we need to solve urgently. These are domains where I think increasing our performance level would make a huge difference.</p>
<p>I think we need to be able to fight Akrasia - weakness of will, going against our best intentions. We should be able to generate a lot more Motivation for the things we want to do, and decrease Motivation for the things we don’t want to do.</p>
<p>Next, we should be able to discover the Truth about some topic more efficiently. In other words, we should be able to increase our Predictive Power more efficiently and solve problems more easily.</p>
<p>Finally, I think we should be able to increase our performance level in some area more efficiently. We should be able to gain Skill much faster. For humans, just knowing the abstract procedure for a task isn’t enough, we need to develop our skill at it in order to do it quickly and accurately.</p>
<p>These are the most important problems I can think of. If getting great results is about knowing what to do and doing it well, then Truth (knowing it), Motivation (doing it), and Skill (doing it well) are the skills you need to have.</p>
<p>TODO: There is also the problem of Complexity - constructing the action you need to do. It’s not usually as straightforward as picking an action out of a lineup. You need to build up a plan using smaller plans and so on. This involves Creativity too.</p>
<p>Yes, there is also the problem of coordinating groups to get important tasks done, but I have zero experience there. Will keep that aside for now. Next, there is the problem of emotional health - we humans can’t just keep solving problems mechanically. Things like depression, anxiety, shame, unhappiness, etc. can bring our campaign to a screeching halt. We need to get our emotions in line with our goals. Will probably get going on this in the future too.</p>
<hr />
<p>However, Terence Tao warns against trying to solve “Big Problems” too early <a href="#fn1" class="footnoteRef" id="fnref1"><sup>1</sup></a>. I don’t know how much his advice in the domain of Math problems would apply to the general problems I’m talking about, but let’s assume that it has some truth to it. Maybe, instead of going for the Moon Shot of solving the Big Problem instantly, we can incrementally build up solutions to smaller problems that empirically give us results.</p>
<h1 id="the-important-questions">The Important Questions</h1>
<p>Which domains do we need to focus on? Ok, for now, we’re focussing on Motivation, Truth, and Skill.</p>
<p>These are rather vague terms. Let’s be more specific.</p>
<h2 id="motivation">Motivation</h2>
<p>We need to procrastinate less. We need to increase our <em>probability</em> of doing important tasks in the following areas.</p>
<ul>
<li><p>Going against the current</p>
<p>Doing totally uncomfortable but important stuff that nobody does. Stuff so <strong>scary</strong> that you don’t even think about it. It just doesn’t exist in your mind. This is what we procrastinate on.</p></li>
<li><p>Hard Work - Mission-critical stuff</p>
<p>Thinking deeply about the issues, writing essays maybe, studying the topic, taking a bunch of action on it, etc.</p></li>
<li><p>Boring but necessary things</p>
<p>Habits like revising your notes, exercising regularly, etc.</p></li>
<li><p>Time-wasters - avoiding crap</p>
<p>Not surfing the web randomly, wasting time on unproductive activities, etc.</p></li>
<li><p>Resisting temptations</p>
<p>Avoiding stuff for which your body has a Strong Impulse. It badly wants to do them. Eating junk food, doing addictive things, etc.</p></li>
<li><p>High-Delay Action</p>
<p>Doing things that will only pay dividends in the very long run - stuff we don’t feel like doing at all. Saving money for retirement, keeping fit, changing domains strategically, etc.</p></li>
</ul>
<h2 id="truth">Truth</h2>
<p>Truth = Predictive Power, remember? We need to increase our Predictive Power in the domains we care about. We must be capable of consistently predicting the narrow outcomes that will happen.</p>
<p>In an academic domain, this could be the percentage of textbook problems you’re able to solve. Or your scores on the various tests, whatever. In a real-world setting, it could be the proportion of problems you’re able to solve - say, number of programs you’re able to write.</p>
<h2 id="skill">Skill</h2>
<p>We must be able to increase our performance levels in the domains we care about.</p>
<hr />
<h1 id="performance-measures-come-first">Performance Measures come first</h1>
<p>With all that said and done, let’s get to the meat of the issue.</p>
<p>After my last two essays, I’ve realized - I fell into the Technique Trap.</p>
<p>I thought I had a lot of techniques that would help, but I now realize that they were just “techniques”. I didn’t have any real performance measures for these Big Problems and so there was nothing keeping me honest. I didn’t make sure the “techniques” were evidence-backed, I didn’t learn how to <em>actually</em> use them, I didn’t <em>actually</em> apply them, and, of course, I didn’t <em>actually</em> get any results. I was just fooling myself.</p>
<p>No more.</p>
<p>The first step is to get proper empirical measures for each of the Big Problems. I don’t have a technique unless I can use it to <em>actually</em> get measurable benefits.</p>
<p>Everything else can come afterwards. I can acquire more techniques and test them against the performance measures. I can develop techniques of my own and see if they actually work. I can practice my techniques and use the Measurement tools to assure myself I’m making progress. But, first, you need to get the performance measures.</p>
<h1 id="designing-performance-measures">Designing Performance Measures</h1>
<p>We can imagine a representative sample of problems - a sort of practical exam. Your performance here should probably correlate with your actual skill level.</p>
<p>Or, we could have unobtrusive measures about your daily activities. Motivation could measure the percentage of times you procrastinated on everyday stuff - if you are getting better, you should be procrastinating less by some meaningful measure. Ditto for Truth and Skill.</p>
<p>First, though, what are the actual problems we’re trying to solve? Finding the Truth about what, specifically? Generating Motivation for what? Increasing Skill at what?</p>
<p>Solving hard problems in areas that matter for human life and/or will make a lot of money. These seem to be primarily math problems. Choosing decisions that generate more wealth.</p>
<p>Let’s just say it outright. Solving important research problems. Creating solutions to problems that can scale like hell - startup. Rationality Research - solving Epistemic Rationality and Instrumental Rationality problems.</p>
<hr />
<p>Having a formal test benchmark would be a good thing. But there are other ways.</p>
<p>Mainly, <a href>Rapid Prototyping</a>! Only after the Technique Trap insight did I realize that a major contribution of Rapid Prototyping is the empirical evidence you get at each step. Every part of your essay or program or product is there because it is shown to work, when combined with the other parts.</p>
<p>I could take a hybrid approach - using Rapid Prototyping on some actual problems to get a feel for which patterns of behaviour aka techniques give me the results I want. I can isolate them and practice them intensely. I can use the original problem as a performance measure for the technique.</p>
<blockquote>
<p>Skill = Pattern of behaviour that increases your performance empirically.</p>
</blockquote>
<p>In other words, I can use Rapid Prototyping on some actual problems to discover the techniques that will help me increase performance. Then, I can use a version of those problems as part of my performance measure. Using that performance measure, I can practice those techniques in isolation and kick ass.</p>
<p>Also, I need to make sure that the Skills are actually valuable. They must lead to significant benefits in the domain.</p>
<h1 id="aim">Aim</h1>
<p>Judge yourself <em>only</em> by your empirical achievements. No Affective Death Spirals.</p>
<p>Discover important empirical techniques. Then, practice them like hell.</p>
<p>Take up small projects just to test out your “techniques”. If your MySTiC techniques are going to be of use to people everywhere (Epistemic and Instrumental Rationality), then you need to show real-world benefits right away. These techniques need not be related to the final, huge problems you need to solve. They can just be problems along the way.</p>
<p>Yes, I’m taking up projects just to show that my techniques work. That’s the point. So far, I haven’t shown any empirical results. I have nothing. I need to Rapid Prototype my “techniques” till they give <em>some</em> real-world benefit at all.</p>
<p>In other words, I’m going to use these small projects as the performance measure for my techniques. If my “techniques” don’t give good results on these projects, then I modify them until they do or I ditch them.</p>
<p>In fact, start off with trivial projects. Set the bar as low as you can. Get some results before you move towards real-world projects.</p>
<p>Maybe even take the “techniques” you think are promising and think of projects on which they can deliver good results. Then, test them out. List out all the great “techniques” you have heard of and think up projects where they can give results. If you find it hard to do so, well, you have your results right there.</p>
<p>If your techniques really are so great, you should be able to get empirical results from them very easily.</p>
<p>Techniques: Naive Realism, Taboo, Scientific Method, Information Theory, Hard Work vs Hard-to-do Work, etc.</p>
<p>Brain Techniques: Chunking, Recognizing Impasses, One Conscious Process at a time, etc.</p>
<p>Possible Projects:</p>
<ul>
<li>Learn stuff faster - take a textbook and master it and score well on the exam.</li>
<li>Scientific Method - Design the right hypothesis tests and gain Predictive Power about some area</li>
<li>Category Theory - Write a large program simply and easily.</li>
<li>TME and Gamification - Be able to do even the most tedious and challenging tasks.</li>
<li>Power of Vulnerability and Alain de Botton wisdom - reduce anxiety, frustration, anger, etc.</li>
<li>Deep Practice - increase skill in some area a lot</li>
<li>Rewrite Sequences on my own</li>
</ul>
<p>Constraint: They should be doable without too much fuss or startup cost. Should take maximum a week.</p>
<h1 id="actual-problems">Actual Problems</h1>
<p>So, all we need to get started are the actual problems we need to solve. Based on those, we can start the Rapid Prototyping process and build up our performance measure. With that, we can create a practice regimen and also a research program for new techniques.</p>
<p>I need a few sample actual problems from the three domains. Plus, Complexity too.</p>
<p>Let’s start easy.</p>
<p>Maybe look at ideas from LessWrong - Epistemic and Instrumental Rationality.</p>
<p>What problems do people need solved? After all, I think solving these Big Problems will help people a lot. How exactly will it help them? Specify the concrete problems and use some of them here.</p>
<ul>
<li><p>Motivation</p>
<p>Do a 100 household tasks. Do one task from each of the categories.</p>
<p>How about this - Sticking to a schedule for two weeks?</p></li>
<li><p>Truth</p></li>
<li><p>Skill</p>
<p>??</p></li>
<li><p>Complexity and Creativity</p>
<p>Maybe write a medium-sized program (1k lines? in Haskell?). Write a couple of long essays, maybe? Distill a complex book, maybe. Maybe create a product (maybe as a program) and package it well and publish it.</p></li>
</ul>
<h1 id="notes">Notes</h1>
<ul>
<li><p>If you’re trying to increase your skill, design your practice regimen using <em>actual</em> Deliberate Practice principles.</p></li>
<li><p>I need to reconsider techniques I gave up on a long time ago, after my Technique Trap insight. I need to test empirically whether they work or not.</p></li>
<li><p>Aim: Counter Naive Realism and Taboo the Big Problems</p></li>
<li><p>Link to Eliezer’s Problems ahead essay.</p></li>
</ul>
<div class="footnotes">
<hr />
<ol>
<li id="fn1"><p>Terence Tao’s <a href>essay</a>.<a href="#fnref1">↩</a></p></li>
</ol>
</div>
<div class="info">Created: May 23, 2015</div>
<div class="info">Last modified: September 28, 2019</div>
<div class="info">Status: finished</div>
<div class="info"><b>Tags</b>: big problems</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 = '/The-Big-Problems-Reloaded.html';
var disqus_title = 'The Big Problems: Reloaded';
/* * * 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>