Help us Optimize the Contents of the Sequences eBook
MIRI’s ongoing effort to publish the sequences as an eBook has given us the opportunity to update their contents and organization.
We’re looking for suggested posts to reorder, add, or remove.
To help with this, here is a breakdown of the current planned contents of the eBook and any currently planned modifications. Following that is a list of the most popular links within the sequences to posts that are not included therein.
Now’s a good time to suggested changes or improvements!
———
Map and Territory
Added …What’s a Bias Again? because it’s meant to immediately follow Why Truth, And….
Mysterious Answers to Mysterious Questions
No changes.
A Human’s Guide to Words
No changes.
How to Actually Change Your Mind
Politics is the Mind-Killer
Removed The Robbers Cave Experiment because it already appears in Death Spirals and the Cult Attractor, and there in the original chronological order which flows better.
Death Spirals and the Cult Attractor
Removed The Litany Against Gurus because it already appears in Politics is the Mind-killer.
Seeing with Fresh Eyes
Removed Asch’s Conformity Experiment and Lonely Dissent because they both appear at the end of Death Spirals. Removed The Genetic Fallacy because it’s in the Metaethics sequence: that’s where it falls chronologically and it fits better there with the surrounding posts.
Noticing Confusion
Removed this entire subsequence because it is entirely contained within Mysterious Answers to Mysterious Questions.
Against Rationalization
Added Pascal’s Mugging (before Torture vs Dust Specks) because it explains the 3^^^3 notation. Added Torture vs Dust Specks before A Case Study of Motivated Continuation because A Case Study refers to it frequently.
Against Doublethink
No changes.
Overly Convenient Excuses
Removed How to Convince Me that 2+2=3 because it’s already in Map & Territory.
Letting Go
No change.
The Simple Math of Evolution
Added Evolutionary Psychology because it fits nicely at the end and it’s referred to by other posts many times.
Challenging the Difficult
No change.
Yudkowsky’s Coming of Age
No change.
Reductionism
No change. (Includes the Zombies subsequence.)
Quantum Physics
No change. Doesn’t include any “Preliminaries” posts, since they’d all be duplicates
Metaethics
No change.
Fun Theory
No change.
The Craft and the Community
No change.
Appendix
Includes:
———
Here are the most-frequently-referenced links within the sequences to posts outside of the sequences (with a count of three or more). This may help you notice posts that you think should be included in the sequences eBook.
The Second Law of Thermodynamics, and Engines of Cognition ⇒ 22
Burdensome Details ⇒ 16
Thou Art Godshatter ⇒ 14
Scope Insensitivity ⇒ 13
The Ultimate Source ⇒ 13
Think Like Reality ⇒ 10
Lost Purposes ⇒ 9
A Priori ⇒ 8
Availability ⇒ 7
I Defy the Data! ⇒ 4
Bystander Apathy ⇒ 3
Optimization ⇒ 3
Fake Morality ⇒ 3
Suggestions?
I tend to think of “Diseased thinking: dissolving questions about disease ” as belonging into the “A Human’s Guide to Words” sequence. As I remember, the example of “what is a desease” is much more relevant, motivating and enlightening than any example in Yudkowsky’s sequence.
Will the QM sequence be checked over for mathematical errors? These have been used in practice to quickly dismiss the sequence.
You can build arbitarily-phase-shifting optical components. There’s no reason one couldn’t make half-silvered mirrors with a coating that makes it act like Eliezer’s… and any physicist ought to know this. Plus, the real issue is the total difference in phase across the two paths, and you can tweak that however you like by adjusting the path lengths.
SO, either fix it numerically or include a note to that effect, because there’s no reason this needs to fall to a silly nitpick.
To be clear, is the criticism just wrong, or should the sequence be adjusted? What exactly needs to be done to fix it numerically?
Both. The sequence is technically off on that detail, but someone who knows their stuff should know better than to complain about it. The simplest fix is to just say that this is a special custom half-silvered mirror that has these phase shifts.
I was going to say that the cleanest fix would be to make the math right, but looking at the math, it seems that the real math is so much messier and harder to explain (half-silvered mirrors invert or not based on which side you half-reflected from!) that getting it right would muddy the waters far more than using your power of arbitrary setups to make an idealized apparatus.
Motivated reasoning made its masterpiece there—the guy does know his stuff, but he basically found one tiny nit and went “THIS IS CLEARLY ALL BOGUS, AS AN EXPERT I IMPLORE YOU ALL TO IGNORE EVERY WORD”. That was why I posted to Physics Exchange about it, to demonstrate to a friend that when he says “trust me, I’m an expert” you can’t trust him.
Not directly related, but just wanted to mention that unqualified statements like
are way too strong, and unnecessarily so.
A single photon is not a part of reality, it’s just a simplified convenient model of electromagnetic emission and absorption, which happens to work well in certain circumstances. Consider the question “When was the photon emitted?”. Unless you measure the recoil of the emitter (a classical measurement), you don’t know. You can reconstruct this time if you use a sensitive enough photomultiplier in the detector, which is again a classical measurement.
A somewhat more detailed QED-based model describes spontaneous emission as unitary evolution of an excited atom into a superposition of the multitude of the states of the form (a recoiled ground state of atom * a corresponding excited state of the EM field). This is not a well-defined single photon, but rather a superposition of all possible photons emitted at all possible times in all possible directions. What makes it into a single photon is the post-selection process (selecting a single possible world out of infinitely many ones, in the MWI picture).
So “A photon heading toward A” is not the territory, it’s a useful simplification of a more accurate model. Which only underscores the point that “distinct configurations are not distinct particles”.
Would you be satisfied if he pointed out that there’s no way to achieve such a well-specified initial condition in real life?
I would be happy if he did not claim that a specific model is the reality.
That’s a different point you raised elsewhere, yes. I meant on the point you raised above.
Would you be satisfied on this other front if he restricted himself to saying that whatever it ends up being, it’s not going to be an objective collapse theory? A distressingly large number of people haven’t gotten that memo.
Oh, I don’t disagree that claiming that some form of objective collapse is the “reality” is not very smart. While QM can be simulated this way, and usually is, there is no reason to expect that this ad hoc rule is as deep as it gets. Unless we are in a poorly written simulation.
A concrete example of a paper using the add-i-to-reflected-part type of beam splitter is the “Quantum Cheshire Cats” paper:
The figure from the paper:
I also translated the optical system into a similar quantum logic circuit:
Note that I also included the left-path detector they talk about later in the paper, and some read-outs that show (among other things) that the conditional probability of the left-path detector having gone off, given that D1 went off, is indeed 100%. (The circuit editor I fiddle with is here.)
It’s notable that my recreation uses gates with different global phase factors (the beam splitter is 1/2-i/2 and 1/2+i/2 instead of 1/sqrt(2) and i/sqrt(2)). It also ignores the mirrors that appear once on both paths. The effect is the same because global phase factors don’t matter.
edit My ability to make sign errors upon sign errors is legendary and hopefully fixed.
I agree with moridinamael...nix “The Simple Truth”. I know it’s not really from the original sequences, but I think “The Useful Idea of Truth” should be used instead.
I think this adding “The Useful Idea of Truth” is the most important change. The first post is the most important, and even if you want to call it “Sequences 2006-2009,” you should still open with a post designed not to be informative as much as to catch readers. I think the “The Useful Idea of Truth” is the current best option. What would be even better would be if Eliezer wrote a new intro.
I liked The Simple Truth as a cute little story… I still have no idea what the intended message was supposed to be. It’s not even clear if the guy who decided to experimentally and dramatically test his model of the world survives through quantum immortality.
The intended message was that a lot of the dead ends in epistemology are pretty silly when considered explicitly.
What was the dead end?
every other sentence markos makes is a reference to some famous epistemological claim.
I really like the cute little story as you say, but agree that it isn’t effective where it is. Maybe include it in the end as a sort of appendix?
I can think of two possible primary goals for this project:
A prettier format for LWers to review the Sequences in, allowing easy full-text-search. A refresher and trophy.
A tidier, better-organized, more approachable update of the Sequences to introduce entirely new people to the ideas therein.
If the latter, brevity is a virtue. Every 50 pages longer the eBook gets probably takes a significant chunk out of how many people read any of it at all. So there’s a lot of reason to excise everything unnecessary to an Appendix. And to shunt the Appendix off into a separate eBook of its own. (“The Out-Of-Sequences.”)
An alternative (or complementary) solution is to divide the Sequences into several separate volumes, each no more than 250 pages long. Then you can be given the option either of downloading a single eBook that collects these volumes, or of downloading the separate eBooks and choosing a favorite one to send to a friend. Makes for a drastically less intimidating Christmas present.
We may later decide to release particular sequences as ebooks, especially the Quantum Physics sequence.
That’s the one.
This is outside the scope of this project.
They’d be more easily bought as books then too.
fake fake utility functions is an important subsequence, IMO, as is the one around thermodynamics and engines of cognition.
In fact I’d feel a bit uneasy cutting out any well referenced post. IMO the whole thing should be in chronological order, or an order strongly influenced by chronology, with just the worst stuff cut out.
I really would like a chronological order.
Expecting Short Inferential Distances may be a good idea, since it encapsulates the reason there are so many words/posts in the sequences.
I thought Religion’s Claim to be Non-Disprovable is an amazing post, one of the most memorable, and it’d be a shame not to include it.
Likewise, The Hidden Complexity of Wishes was an incredibly intuitive way to explain that an AI doesn’t have to hate you for it to destroy the world, it simply has to not have your values.
No Highly Advanced Epistemology 101 for Beginners?
Luke has said it will be in a different ebook.
Do you have a list of blog posts that won’t be in the book as a result?
On the one hand, I want to say “it’s an ebook, stick them all in an appendix”. On the other hand, the more slowly the progress bar at the bottom of the page moves the more readers are likely to give up. What can be left out should be.
That’s what the list at the bottom of the OP is for: posts that are currently not slotted to be in the book, but are linked to by sequence posts.
I lean this way as well.
D’oh! Thanks.
This seems to imply that the listed posts aren’t part of the sequences, but several of the linked posts say they are part of a sequence, for example Decoherence is Falsifiable and Testable says it it part of the QM sequence.
Opps, you’re right, there are a few that are already in the sequences that accidentally made it onto this list (I count three). All other links are to posts that are not in the sequences, but are linked to posts within the sequences. The list is auto-generated.
I think it was just an auto-generated list.
Luke, Alex: After reading through most of Eliezer’s 2006-2009 posts over the last 3 days, I suggest including in the eBook all the out-of-sequence articles you listed, except:
The Cartoon Guide to Löb’s Theorem
Conjunction Fallacy (summarized adequately in lots of other articles, especially Burdensome details)
Cynical About Cynicism
Fake Fake Utility Functions
For The People Who Are Still Alive
Locate the hypothesis
The Martial Art of Rationality
The Modesty Argument
No One Knows What Science Doesn’t Know
Optimization
Passing the Recursive Buck
Possibility and Could-ness
Priors as Mathematical Objects
Rational evidence
Rationality is Systematized Winning (don’t need a whole article for this; to the extent it isn’t just a recap of the Newcomb’s Problem article, the point can be made in one sentence in the Introduction, or in a slightly edited What do I mean by rationality? and/or Incremental progress and the valley)
Surface Analogies and Deep Causes
The Two-Party Swindle (interesting but not essential, and copy-pastes content from other pages)
The Ultimate Source
The Unfinished Mystery of the Shangri-La Diet (this and Akrasia and Shangri-La are both too specific to be good core sequence candidates; their message is better covered in Beware of other-optimizing)
We Don’t Really Want Your Participation
Why is the Future So Absurd? (use Stranger than history instead; showing is more powerful than telling)
You Only Live Twice (already in the FOOM eBook)
Zen and the Art of Rationality
From what I can tell, all the links to those pages within the Sequences are minor passing mentions, and their contents aren’t particularly valuable for understanding the core content of the Sequences. Understanding Free Will, for instance, isn’t needed for the rest of the Sequences, and EY actively discourages just copy-pasting his answer here.
I note that about 25 of the posts from the Quantum Mechanics / Timeless Physics sequence also aren’t necessary for understanding the rest of the Sequences, differ in tone in interesting ways (they remind me of 2010-2013 Eliezer), and make the QM section look a heck of a lot more intimidating than it needs to. They also need new images to replace the quick pencil drawings, so I suggest leaving out the more technical stuff and then consolidating all the QM posts in a future eBook, once the images are polished up.
I can also suggest places to integrate the new additions that will make it possible to read the entire Sequences eBook from front to back without doing any hopping around. It doesn’t take very much reshuffling to make the dependencies work sequentially in the Sequences.
I would rather the QM sequence was shortened to the low-controversy subset Eliezer described in An Intuitive Explanation of Quantum Mechanics and checked for technical accuracy. The pure MWI advocacy part belongs in some appendix, and the outright nonsense like “a Bayesian can become as smart as Einstein” should be chucked and never mentioned again.
+1
Citation needed. I believe shminux made this up.
He’s not making it up out of whole cloth, though he’s being significantly uncharitable.
More precisely, I think this is a reference to Einstein’s Speed and related Sequence posts, where EY argues that Einstein’s unusual success at understanding physical law was significantly due to updating on all available Bayesian evidence rather than just the subset of such evidence that non-Bayesian scientists use.
That said, it’s of course a huge jump from “Einstein would not have been as successful had he not been a Bayesian” to “any Bayesian can be as successful as Einstein,” and I don’t recall EY (or anyone else) ever making the latter claim.
It seems to me that accusing someone of saying “outright nonsense” like this when they in fact did not say something like that and said only something vaguely related is an act that I would like to see discouraged when detected. Straw men are not welcome on lesswrong!
More precisely, the author ‘made this up’ because he believes it is acceptable to distort reality to that degree when arguing in this environment. Reception of the claim at the time I replied to it indicated that this belief is correct. I would prefer it if this were not so.
I share your preference for discouraging straw men and uncharitable readings.
I haven’t found anything Eliezer’s written about Einstein to not be useful. Could you explain why you don’t like it (and/or specify what it is you dislike), or link me to an explanation?
If we want to shorten the QM stuff and explain MWI without belaboring its truth, I don’t think it would be out of the question to commission a specialist like Amit Hagar or David Albert to write a short explanation of what the QM-interpretation fuss is all about, insert that right before the more important QM implications philosophy-of-science stuff (Think like reality, When science can’t help, etc.), and then put Eliezer’s technical explanations in an appendix. That would do a lot to mitigate the criticism of the Sequences for uncredentialed nonstandard physics espousal, and it would lose fewer readers whose math or physics backgrounds are weak.
What has been proven wrong is the idea that explicit Bayesian thinking gives a physicist a significant rather than a marginal advantage. I don’t know of any physicist who learned Bayesian thinking and suddenly became much more productive/successful/famous. You are likely to do better than without it, but you will never be as good as a noticeably smarter not-explicitly-Bayesian physicist, let alone Einstein.
Eliezer’s waxing poetic about Barbour, who is a fringe scientist with intriguing ideas but without many notable achievements, is high on pathos, but not very convincing.
Are there any updates on when this will be released?
Has anyone considered A/B testing different versions of proposed changes or orderings?
I don’t think it’s practical to A/B test a million word book.
It’s perfectly possible to A/B test it, it just depends on what you’re varying and how it’s being delivered. My website is well over a million words, and I A/B test changes to its appearance all the time.
You can A/B test whether doubling the length with appendices affects download completion rates.
That’s not necessarily a bad idea, but how do you A/B test different versions of a PDF? What is the response that is being measured?
I’d test one chapter (or one sub-chapter) at a time. Reader-reported assessments of quality and ease of understanding would be the obvious thing to measure for any instructional book… but also, isn’t CFAR trying to come up with various “rationality tests”? Accuracy of responses to test questions would be a great metric to try and optimize, and most instructional material benefits from having end-of-chapter questions anyway if only to help readers verify that they’ve gained actual understanding and not just an illusion thereof.
You can embed arbitrary javascript in PDFs, so what about including “phone-home” text-boxes for marginalia in rather the same way that online editions of Real World Haskell and other programming books have comment threads for each paragraph? I’d think other metrics like time-on-page could be measured as well. This would need to be disclosed, of course, and not every reader can be expected to comment, but the “how” seems tractable at first impression.
I don’t have any useful insights on what response to measure.
This doesn’t seem usefully true. Some googling for search queries like ‘a/b testing PDFs’ or ‘PDF phone home’ show no one discussing the topic of A/B testing different versions of PDFs, and Wikipedia indicates that only Adobe supports JS and even it produces a popup when you try to phonehome. So any A/B test is going to work on only a fraction of users (how many LWers still use Adobe Acrobat to read PDFs?), and it will alarm the ones it does work on (‘is this PDF spyware‽’).
Foxit Reader supports javascript, and libpoppler (which powers evince and okular, among others) does as well.
Without something to measure, though, that’s really just a technical curiosity.
I don’t know how the security model works in various PDF readers, but wouldn’t the javascript code be sandboxed, hopefully? Sane security practices shouldn’t allow arbitrary code in PDFs to talk to random ’net addresses...
If the PDF is signed by a certificate the user has manually installed, it can embed what Adobe calls “high privilege” javascript, which includes the ability to launch any URL. That’s an extra step, which would discourage some users, but on the plus side it addresses the “who’s given informed consent?” problem.
Momentarily donning a slightly darker hat: it is also possible for a PDF to launch an arbitrary executable (see pp. 30-34 of Julia Wolf’s OMG WTF PDF, video). AIUI this requires no additional privileges.
My estimate for the value of that “some” is 95%+
Not to mention that most of the people who can be easily persuaded to manually install a cert on their PC probably already have a dozen toolbars in their browser… :-D
One metric that could be calculated automatically: how many articles in the book contain hyperlinks to the articles that appear later in the book, or don’t appear in the book?
I guess we should try to make that number as small as possible, for convenient linear reading, but of course it has to be balanced against other concerns (such as keeping articles with similar topics together, so we can make chapters with a unified topic).
Another idea: identify the main topics of the book (they roughly correspond to the chapters you have now: “map and territory”, “words”, “politics”, “reductionism”, “quantum physics”, “metaethics”, “community”, etc.), and mark each article with the corresponding tags.
The goal of this is to reduce the cognitive load while sorting the chapters. Instead of thinking “what was the Fake Morality article about, again?” (and trying to keep all those articles in your head), reduce all chapters to a short information: article name (or identifiers), article tags, names of referenced articles. And then only use this information to sort the articles. -- Perhaps by describing every articles in one line of a text file, easily moving them around in the text editor, and calculating the “non-backwards hyperlinks” metric automatically.
I also wanted a list of forward references to assess this ordering against.
I like this idea!
That’s the list at the end of the OP.
Should it open with “The Simple Truth”, since it’s a precursor to Map and Territory?
It could be that The Simple Truth may work better after (some of the many) concepts it illustrates have already been conveyed. I like it because it demonstrates in a memorable way a sane response to various philosophical confusions. It may be less helpful for those who are less inclined to appreciate satirical allegory and who need to be persuaded by more sombre explanation. Comprehend then memorize is common advice to those using spaced repetition and probably generalises to other means of facilitating long term retention.
Exactly what I was thinking! Also, I often send people ‘The Simple Truth’ when I start recommending them to read the LessWrong Sequences.
I’ve mentioned this elsewhere, but, The Simple Truth is stylistically totally different from the rest of the Sequences, extremely long, and rather meandering. I think it’s enjoyable if you’ve already bought in to the memeplex but utterly confusing if you haven’t. Specifically, I tried to get my wife to read the Seauences and The Simple Truth was the article that she never got past. I know, n=1 and all, but I stand by my other points.
Plus, more generally, if I were putting the book together I’d want to shuttle the reader from page 1 to Mysterious Answers to Mysterious Questions as soon as I could, since MAMQ is where the Sequences really pulls up the landing gear and takes off. Putting “The Simple Truth” at the start puts another 7,000-odd words between the reader and that take-off. (And I agree that it’s tortuous & opaque, which just makes it an even bigger roadblock. I’d hide it an appendix too.)
I agree with this. I’d start with some bits and pieces of the Map/Territory sequence, followed by a crash course on assorted cognitive biases (because it won’t make sense that Map/Territory talks at length about ‘human irrationality’ and ‘bias’ unless the reader is swiftly provided with examples), followed immediately by a (slightly modified) Mysterious Answers. Then How To Actually Change Your Mind.
It’s a story you either love or hate. I happen to love it, but that’s probably because (a) I already agree with the ideas, and (b) I have met people expressing the silly ideas in real life, so I know what it is about. -- In other words, it is a bit like an applause light for the believers. Not a good introduction.
I also happen to love the “twelve virtues”… so I am curious whether people have mixed reactions to that one, too. Unlike the simple truth, the twelve virtues is short. (More precisely, the individual twelve points are short; but the total length is still acceptable.)
I neither love nor hate it. It’s pretty good. The concept is great; the execution is a bit too injokey and unstructured. Perhaps the main problem is that you need to be pretty savvy and experienced regarding pop philosophy to know what the point of the allegory is (without expending a fair amount of effort), but the sorts of people most likely to make the crude errors The Simple Truth is correcting aren’t likely to be particularly savvy.
I do love The Twelve Virtues, though I think they could easily be compressed to something cleaner and easier to remember (e.g., with less built-in redundancy). For instance, Relinquishment and Humility seem to just be special cases of Lightness and Evenness (with a dash of cog-sci Empiricism/Scholarship, mayhap).
If I were redoing the list, I’d go with Ten Virtues, something more like: Curiosity, Daring, Attentiveness, Lightness, Evenness, Simplicity, Precision, Rigor, Community, The Void.
(‘Argument’ breaks up into Daring and Community, ‘Empiricism’ breaks up into Daring and Attentiveness and Simplicity and pretty much all of the other virtues, ‘Scholarship’ breaks up into Rigor and Community, etc. I’d still want virtues like Scholarship explicitly talked about, but as more complicated practices that come out of the primary-color virtues.)
Yeah, I usually make it explicitly clear that the style of The Simple Truth is nothing like the style of the sequences. I also point out that some people are not fond of it when recommending it.
The closing parenthesis is missing on the link to “Map and Territory (sequence)”. If you compose in Markdown, you have to escape this with a backslash.
I have a comment that takes no issue with the central argument of the ‘Mind Projection Fallacy’ sequence post, it’s just about an embellishment that is, I think, false and a little cringe-inducing:
From “Mind Projection Fallacy”:
This should be removed or altered. Kant’s declaration that space is by its very nature flat is a declaration about the conditions of experience, not the world as such. He’s explicit that space and its flatness are not features of the territory (so to speak), but necessary features of the map. He may be wrong about this, but it’s not a case of the mind projection fallacy.
As to the Hume comment, this is a bit of a non-sequitor as phrased: a definition of a philosophical term can’t be a case of the mind projection fallacy.
I’ve been working on an introductory set of posts for a (religious) friend and Hpmor fan. I’ve been trying to find the posts that explain the various parts of the LW philosophy best, and I’ve come up with this.
I’d just like to comment a little on which posts I selected, and then recommend for the ebook.
I think zen-like little pieces are nice to start with, like the Twelve Virtues and The Simple Truth, but I think that the Simple Truth requires as either a pre-req or an immediate following by ‘The Useful Idea of Truth’, to make people see the meaning of the former, otherwise it can be lost entirely. I think f Useful was edited slightly it could fit into the ebook well enough on its own. Also the Map and Territory and Appreciating Cognitive Agorothms from HAE101FB I think are great to introduce core ideas.
I think the inferential distances should be in, so the reader realises things can take time to explain. Also, posts on Curiosity are very useful near the beginning, to give the reader a care for the truth after they later perhaps read something they disagree with.
I’d like to see all the ones that have been referenced over six times added, and also Conjunction Fallacy, Lawful Creativity and Illusion of Transparency.
Cheers. Hope there’s an opportunity to get it as two or three physical volumes.
I’d drop A Case Study of Motivated Continuation from Politics rather than trying to add two such unrelated prereqs.
Like other commenters here, I’d like to see the Twelve Virtues make the cut.
I’m really excited about this book!