People trying to guard civilisation against catastrophe usually focus on one specific kind of catastrophe at a time. This can be useful for building concrete knowledge with some certainty in order for others to build on it. However, there are disadvantages to this catastrophe-specific approach:

1. Catastrophe researchers (including Anders Sandberg and Nick Bostrom) think that there are substantial risks from catastrophes that have not yet been anticipated. Resilience-boosting measures may mitigate risks that have not yet been investigated.

2. Thinking about resilience measures in general may suggest new mitigation ideas that were missed by the catastrophe-specific approach.

One analogy for this is that an intrusion (or hack) to a software system can arise from a combination of many minor security failures, each of which might appear innocuous in isolation. You can decrease the chance of an intrusion by adding extra security measures, even without a specific idea of what kind of hacking would be performed. Things like being being able to power down and reboot a system, storing a backup and being able to run it in a "safe" offline mode are all standard resilience measures for software systems. These measures aren't necessarily the first thing that would come to mind if you were trying to model a specific risk like a password getting stolen, or a hacker subverting administrative privileges, although they would be very useful in those cases. So mitigating risk doesn't necessarily require a precise idea of the risk to be mitigated. Sometimes it can be done instead by thinking about the principles required for proper operation of a system - in the case of its software, preservation of its clean code - and the avenues through which it is vulnerable - such as the internet.

So what would be good robustness measures for human civilisation? I have a bunch of proposals:

Disaster forecasting

Disaster research

* Build research labs to survey and study catastrophic risks (like the Future of Humanity Institute, the Open Philanthropy Project and others)

Disaster prediction

* Prediction contests (like IARPA's Aggregative Contingent Estimation "ACE" program)

* Expert aggregation and elicitation

Disaster prevention

General prevention measures

* Build a culture of prudence in groups that run risky scientific experiments

* Lobby for these mitigation measures

* Improving the foresight and clear-thinking of policymakers and other relevant decision-makers

* Build research labs to plan more risk-mitigation measures (including the Centre for Study of Existential Risk)

Preventing intentional violence

* Improve focused surveillance of people who might commit large-scale terrorism (this is controversial because excessive surveillance itself poses some risk)

* Improve cooperation between nations and large institutions

Preventing catastrophic errors

* Legislating for individuals to be held more accountable for large-scale catastrophic errors that they may make (including by requiring insurance premiums for any risky activities)

Disaster response

* Improve political systems to respond to new risks

* Improved vaccine development, quarantine and other pandemic response measures

* Building systems for disaster notification

Disaster recovery

Shelters

* Build underground bomb shelters

* Provide a sheltered place for people to live with air and water

* Provide (or store) food and farming technologies (cf Dave Denkenberger's *Feeding Everyone No Matter What*

* Store energy and energy-generators

* Store reproductive technologies (which could include IVF, artificial wombs or measures for increasing genetic diversity)

* Store information about building the above

* Store information about building a stable political system, and about mitigating future catastrophes

* Store other useful information about science and technology (e.g. reading and writing)

* Store some of the above in submarines

* (maybe) store biodiversity

Space Travel

* Grow (or replicate) the international space station

* Improve humanity's capacity to travel to the Moon and Mars

* Build sustainable settlements on the Moon and Mars

Of course, some caveats are in order. 

To begin with, one could argue that surveilling terrorists is a measure specifically designed to reduce the risk from terrorism. But there are a number of different scenarios and methods through which a malicious actor could try to inflict major damage on civilisation, and so I still regard this as a general robustness measure, granted that there is some subjectivity to all of this. If you know absolutely nothing about the risks that you might face, and the structures in society that are to be preserved, then the exercise is futile. So some of the measures on this list will mitigate a smaller subset of risks than others, and that's just how it is, though I think the list is pretty different from the one people think of by using a risk-specific paradigm, which is the reason for the exercise.

Additionally, I'll disclaim that some of these measures are already well invested, and yet others will not be able to be done cheaply or effectively. But many seem to me to be worth thinking more about.

Additional suggestions for this list are welcome in the comments, as are proposals for their implementation.

Related readings

https://www.academia.edu/7266845/Existential_Risks_Exploring_a_Robust_Risk_Reduction_Strategy

http://www.nickbostrom.com/existential/risks.pdf

http://users.physics.harvard.edu/~wilson/pmpmta/Mahoney_extinction.pdf

http://gcrinstitute.org/aftermath

http://sethbaum.com/ac/2015_Food.html

http://the-knowledge.org

http://lesswrong.com/lw/ma8/roadmap_plan_of_action_to_prevent_human/

This is part of a semi-monthly reading group on Eliezer Yudkowsky's ebook, Rationality: From AI to Zombies. For more information about the group, see the announcement post.

Welcome to the Rationality reading group. This fortnight we discuss Part Z: The Craft and the Community (pp. 1651-1750). This post summarizes each article of the sequence, linking to the original LessWrong post where available.

Z. The Craft and the Community

312. Raising the Sanity Waterline - Behind every particular failure of social rationality is a larger and more general failure of social rationality; even if all religious content were deleted tomorrow from all human minds, the larger failures that permit religion would still be present. Religion may serve the function of an asphyxiated canary in a coal mine - getting rid of the canary doesn't get rid of the gas. Even a complete social victory for atheism would only be the beginning of the real work of rationalists. What could you teach people without ever explicitly mentioning religion, that would raise their general epistemic waterline to the point that religion went underwater?

313. A Sense That More Is Possible - The art of human rationality may have not been much developed because its practitioners lack a sense that vastly more is possible. The level of expertise that most rationalists strive to develop is not on a par with the skills of a professional mathematician - more like that of a strong casual amateur. Self-proclaimed "rationalists" don't seem to get huge amounts of personal mileage out of their craft, and no one sees a problem with this. Yet rationalists get less systematic training in a less systematic context than a first-dan black belt gets in hitting people.

314. Epistemic Viciousness - An essay by Gillian Russell on "Epistemic Viciousness in the Martial Arts" generalizes amazingly to possible and actual problems with building a community around rationality. Most notably the extreme dangers associated with "data poverty" - the difficulty of testing the skills in the real world. But also such factors as the sacredness of the dojo, the investment in teachings long-practiced, the difficulty of book learning that leads into the need to trust a teacher, deference to historical masters, and above all, living in data poverty while continuing to act as if the luxury of trust is possible.

315. Schools Proliferating Without Evidence - The branching schools of "psychotherapy", another domain in which experimental verification was weak (nonexistent, actually), show that an aspiring craft lives or dies by the degree to which it can be tested in the real world. In the absence of that testing, one becomes prestigious by inventing yet another school and having students, rather than excelling at any visible performance criterion. The field of hedonic psychology (happiness studies) began, to some extent, with the realization that you could measure happiness - that there was a family of measures that by golly did validate well against each other. The act of creating a new measurement creates new science; if it's a good measurement, you get good science.

316. Three Levels of Rationality Verification - How far the craft of rationality can be taken, depends largely on what methods can be invented for verifying it. Tests seem usefully stratifiable into reputational, experimental, andorganizational. A "reputational" test is some real-world problem that tests the ability of a teacher or a school (like running a hedge fund, say) - "keeping it real", but without being able to break down exactly what was responsible for success. An "experimental" test is one that can be run on each of a hundred students (such as a well-validated survey). An "organizational" test is one that can be used to preserve the integrity of organizations by validating individuals or small groups, even in the face of strong incentives to game the test. The strength of solution invented at each level will determine how far the craft of rationality can go in the real world.

317. Why Our Kind Can't Cooperate - The atheist/libertarian/technophile/sf-fan/early-adopter/programmer/etc crowd, aka "the nonconformist cluster", seems to be stunningly bad at coordinating group projects. There are a number of reasons for this, but one of them is that people are as reluctant to speak agreement out loud, as they are eager to voice disagreements - the exact opposite of the situation that obtains in more cohesive and powerful communities. This is not rational either! It is dangerous to be half a rationalist (in general), and this also applies to teaching only disagreement but not agreement, or only lonely defiance but not coordination. The pseudo-rationalist taboo against expressing strong feelings probably doesn't help either.

318. Tolerate Tolerance - One of the likely characteristics of someone who sets out to be a "rationalist" is a lower-than-usual tolerance for flawed thinking. This makes it very important to tolerate other people's tolerance - to avoid rejecting them because they tolerate people you wouldn't - since otherwise we must all have exactly the same standards of tolerance in order to work together, which is unlikely. Even if someone has a nice word to say about complete lunatics and crackpots - so long as they don't literally believe the same ideas themselves - try to be nice to them? Intolerance of tolerance corresponds to punishment of non-punishers, a very dangerous game-theoretic idiom that can lock completely arbitrary systems in place even when they benefit no one at all.

319. Your Price for Joining - The game-theoretical puzzle of the Ultimatum game has its reflection in a real-world dilemma: How much do you demand that an existing group adjust toward you, before you will adjust toward it? Our hunter-gatherer instincts will be tuned to groups of 40 with very minimal administrative demands and equal participation, meaning that we underestimate the inertia of larger and more specialized groups and demand too much before joining them. In other groups this resistance can be overcome by affective death spirals and conformity, but rationalists think themselves too good for this - with the result that people in the nonconformist cluster often set their joining prices way way way too high, like an 50-way split with each player demanding 20% of the money. Nonconformists need to move in the direction of joining groups more easily, even in the face of annoyances and apparent unresponsiveness. If an issue isn't worth personally fixing by however much effort it takes, it's not worth a refusal to contribute.

320. Can Humanism Match Religion's Output? - Anyone with a simple and obvious charitable project - responding with food and shelter to a tidal wave in Thailand, say - would be better off by far pleading with the Pope to mobilize the Catholics, rather than with Richard Dawkins to mobilize the atheists. For so long as this is true, any increase in atheism at the expense of Catholicism will be something of a hollow victory, regardless of all other benefits. Can no rationalist match the motivation that comes from the irrational fear of Hell? Or does the real story have more to do with the motivating power of physically meeting others who share your cause, and group norms of participating?

321. Church vs. Taskforce - Churches serve a role of providing community - but they aren't explicitly optimized for this, because their nominal role is different. If we desire community without church, can we go one better in the course of deleting religion? There's a great deal of work to be done in the world; rationalist communities might potentially organize themselves around good causes, while explicitly optimizing for community.

322. Rationality: Common Interest of Many Causes - Many causes benefit particularly from the spread of rationality - because it takes a little more rationality than usual to see their case, as a supporter, or even just a supportive bystander. Not just the obvious causes like atheism, but things like marijuana legalization. In the case of my own work this effect was strong enough that after years of bogging down I threw up my hands and explicitly recursed on creating rationalists. If such causes can come to terms with not individually capturing all the rationalists they create, then they can mutually benefit from mutual effort on creating rationalists. This cooperation may require learning to shut up about disagreements between such causes, and not fight over priorities, except in specialized venues clearly marked.

323. Helpless Individuals - When you consider that our grouping instincts are optimized for 50-person hunter-gatherer bands where everyone knows everyone else, it begins to seem miraculous that modern-day large institutions survive at all. And in fact, the vast majority of large modern-day institutions simply fail to exist in the first place. This is why funding of Science is largely through money thrown at Science rather than donations from individuals - research isn't a good emotional fit for the rare problems that individuals can manage to coordinate on. In fact very few things are, which is why e.g. 200 million adult Americans have such tremendous trouble supervising the 535 members of Congress. Modern humanity manages to put forth very little in the way of coordinated individual effort to serve our collective individual interests.

324. Money: The Unit of Caring - Omohundro's resource balance principle implies that the inside of any approximately rational system has a common currency of expected utilons. In our world, this common currency is called "money" and it is the unit of how much society cares about something - a brutal yet obvious point. Many people, seeing a good cause, would prefer to help it by donating a few volunteer hours. But this avoids the tremendous gains of comparative advantage, professional specialization, and economies of scale - the reason we're not still in caves, the only way anything ever gets done in this world, the tools grownups use when anyone really cares. Donating hours worked within a professional specialty and paying-customer priority, whether directly, or by donating the money earned to hire other professional specialists, is far more effective than volunteering unskilled hours.

325. Purchase Fuzzies and Utilons Separately - Wealthy philanthropists typically make the mistake of trying to purchase warm fuzzy feelings, status among friends, and actual utilitarian gains, simultaneously; this results in vague pushes along all three dimensions and a mediocre final result. It should be far more effective to spend some money/effort on buying altruistic fuzzies at maximum optimized efficiency (e.g. by helping people in person and seeing the results in person), buying status at maximum efficiency (e.g. by donating to something sexy that you can brag about, regardless of effectiveness), and spending most of your money on expected utilons (chosen through sheer cold-blooded shut-up-and-multiply calculation, without worrying about status or fuzzies).

326. Bystander Apathy - The bystander effect is when groups of people are less likely to take action than an individual. There are a few explanations for why this might be the case.

327. Collective Apathy and the Internet - The causes of bystander apathy are even worse on the Internet. There may be an opportunity here for a startup to deliberately try to avert bystander apathy in online group coordination.

328. Incremental Progress and the Valley - The optimality theorems for probability theory and decision theory, are for perfect probability theory and decision theory. There is no theorem that incremental changes toward the ideal, starting from a flawed initial form, must yield incremental progress at each step along the way. Since perfection is unattainable, why dare to try for improvement? But my limited experience with specialized applications suggests that given enough progress, one can achieve huge improvements over baseline - it just takes a lot of progress to get there.

329. Bayesians vs. Barbarians - Suppose that a country of rationalists is attacked by a country of Evil Barbarians who know nothing of probability theory or decision theory. There's a certain concept of "rationality" which says that the rationalists inevitably lose, because the Barbarians believe in a heavenly afterlife if they die in battle, while the rationalists would all individually prefer to stay out of harm's way. So the rationalist civilization is doomed; it is too elegant and civilized to fight the savage Barbarians... And then there's the idea that rationalists should be able to (a) solve group coordination problems, (b) care a lot about other people and (c) win...

330. Beware of Other-Optimizing - Aspiring rationalists often vastly overestimate their own ability to optimize other people's lives. They read nineteen webpages offering productivity advice that doesn't work for them... and then encounter the twentieth page, or invent a new method themselves, and wow, it really works - they've discovered the true method. Actually, they've just discovered the one method in twenty that works for them, and their confident advice is no better than randomly selecting one of the twenty blog posts. Other-Optimizing is exceptionally dangerous when you have power over the other person - for then you'll just believe that they aren't trying hard enough.

331. Practical Advice Backed by Deep Theories - Practical advice is genuinely much, much more useful when it's backed up by concrete experimental results, causal models that are actually true, or valid math that is validly interpreted. (Listed in increasing order of difficulty.) Stripping out the theories and giving the mere advice alone wouldn't have nearly the same impact or even the same message; and oddly enough, translating experiments and math into practical advice seems to be a rare niche activity relative to academia. If there's a distinctive LW style, this is it.

332. The Sin of Underconfidence - When subjects know about a bias or are warned about a bias, overcorrection is not unheard of as an experimental result. That's what makes a lot of cognitive subtasks so troublesome - you know you're biased but you're not sure how much, and if you keep tweaking you may overcorrect. The danger of underconfidence (overcorrecting for overconfidence) is that you pass up opportunities on which you could have been successful; not challenging difficult enough problems; losing forward momentum and adopting defensive postures; refusing to put the hypothesis of your inability to the test; losing enough hope of triumph to try hard enough to win. You should ask yourself "Does this way of thinking make me stronger, or weaker?"

333. Go Forth and Create the Art! - I've developed primarily the art of epistemic rationality, in particular, the arts required for advanced cognitive reductionism... arts like distinguishing fake explanations from real ones and avoiding affective death spirals. There is much else that needs developing to create a craft of rationality - fighting akrasia; coordinating groups; teaching, training, verification, and becoming a proper experimental science; developing better introductory literature... And yet it seems to me that there is a beginning barrier to surpass before you can start creating high-quality craft of rationality, having to do with virtually everyone who tries to think lofty thoughts going instantly astray, or indeed even realizing that a craft of rationality exists and that you ought to be studying cognitive science literature to create it. It's my hope that my writings, as partial as they are, will serve to surpass this initial barrier. The rest I leave to you.

This has been a collection of notes on the assigned sequence for this fortnight. The most important part of the reading group though is discussion, which is in the comments section. Please remember that this group contains a variety of levels of expertise: if a line of discussion seems too basic or too incomprehensible, look around for one that suits you better!

This is the end, beautiful friend!

This is a follow-up to last year's report. Here, I will talk about my successes and failures using Spaced Repetition Software (SRS) in the classroom for a second year. The year's not over yet, but I have reasons for reporting early that should become clear in a subsequent post. A third post will then follow, and together these will constitute a small sequence exploring classroom SRS and the adjacent ideas that bubble up when I think deeply about teaching.

Summary

I experienced net negative progress this year in my efforts to improve classroom instruction via spaced repetition software. While this is mostly attributable to shifts in my personal priorities, I have also identified a number of additional failure modes for classroom SRS, as well as additional shortcomings of Anki for this use case. My experiences also showcase some fundamental challenges to teaching-in-general that SRS depressingly spotlights without being any less susceptible to. Regardless, I am more bullish than ever about the potential for classroom SRS, and will lay out a detailed vision for what it can be in the next post.

Motte and bailey is a technique by which one protects an interesting but hard-to-defend view by making it similar to a less interesting but more defensible position. Whenever the more interesting position - the bailey - is attacked - one retreats to the more defensible one - the motte -, but when the attackers are gone, one expands again to the bailey. 

In that case, one and the same person switches between two interpretations of the original claim. Here, I rather want to focus on situations where different people make different interpretations of the original claim. The originator of the claim adds a number of caveats and hedges to their claim, which makes it more defensible, but less striking and sometimes also less interesting.* When others refer to the same claim, the caveats and hedges gradually disappear, however, making it more and more motte-like.

A salient example of this is that scientific claims (particularly in messy fields like psychology and economics) often come with a number of caveats and hedges, which tend to get lost when re-told. This is especially so when media writes about these claims, but even other scientists often fail to properly transmit all the hedges and caveats that come with them.

Since this happens over and over again, people probably do expect their hedges to drift to some extent. Indeed, it would not surprise me if some people actually want hedge drift to occur. Such a strategy effectively amounts to a more effective, because less observable, version of the motte-and-bailey-strategy. Rather than switching back and forth between the motte and the bailey - something which is at least moderately observable, and also usually relies on some amount of vagueness, which is undesirable - you let others spread the bailey version of your claim, whilst you sit safe in the motte. This way, you get what you want - the spread of the bailey version - in a much safer way.

Even when people don't use this strategy intentionally, you could argue that they should expect hedge drift, and that omitting to take action against it is, if not ouright intellectually dishonest, then at least approaching that. This argument would rest on the consequentialist notion that if you have strong reasons to believe that some negative event will occur, and you could prevent it from happening by fairly simple means, then you have an obligation to do so. I certainly do think that scientists should do more to prevent their views from being garbled via hedge drift. 

Another way of expressing all this is by saying that when including hedging or caveats, scientists often seem to seek plausible deniability ("I included these hedges; it's not my fault if they were misinterpreted"). They don't actually try to prevent their claims from being misunderstood. 

What concrete steps could one then take to prevent hedge-drift? Here are some suggestions. I am sure there are many more.

1. Many authors use eye-catching, hedge-free titles and/or abstracts, and then only include hedges in the paper itself. This is a recipe for hedge-drift and should be avoided.
2. Make abundantly clear, preferably in the abstract, just how dependent the conclusions are on keys and assumptions. Say this not in a way that enables you to claim plausible deniability in case someone misinterprets you, but in a way that actually reduces the risk of hedge-drift as much as possible. 
3. Explicitly caution against hedge drift, using that term or a similar one, in the abstract of the paper.

* Edited 2/5 2016. By hedges and caveats I mean terms like "somewhat" ("x reduces y somewhat"), "slightly", etc, as well as modelling assumptions without which the conclusions don't follow and qualifications regarding domains in which the thesis don't hold.

Hi everyone! This is an experimental thread to relax and enjoy the company of other aspiring rationalists. Special rules for communication and voting apply here. Please play along!

(If for whatever reason you cannot or don't want to follow the rules, please don't post in this thread. However, feel free to voice your opinion in the corresponding meta thread.)

Here is the spirit of the rules:

• be nice
• be cheerful
• don't go meta

And here are the details:

On the scale from negative (hostility, complaints, passive aggression) through neutral (bare facts) to positive (happiness, fun, love), please only post comments from the "neutral to positive" half. Preferably at least slightly positive; but don't push yourself too far if you don't feel so. The goal is to make both yourself and your audience feel comfortable.

If you disagree with someone, please consider whether the issue is important enough to disagree openly. If it isn't, you also have an option to simply skip the comment. You can send the author a private message. Or you can post your disagreement in the meta thread (and then send them the link in a private message). If you still believe it is better to disagree here, please do it politely and friendly.

Avoid inherently controversial topics, such as politics, religion, or interpretations of quantum physics.

Feel free to post stuff that normally doesn't get posted on LessWrong. Feel free to be silly, as long as it harms no one. Emoticons are allowed. Note: This website supports Unicode. ◕‿◕

Upvote the stuff you like. :)

Downvote only the stuff that breaks the rules. :( In this thread, the proper reaction to a comment that you don't like, but doesn't break the rules, is to ignore it.

Please don't downvote a comment below zero, unless you believe that the breaking of rules was intentional.

(Note: There is one user permanently banned from this website. Any comment posted from any of this user's new accounts is considered an intentional breaking of the rules, regardless of its content.)

Don't go meta in this thread. If you want to discuss whether the rules here should be different, or whether a specific comment did or didn't break the rules, or something like that, please use the meta thread.

Don't abuse the rules. I already know that you are clever, and that you could easily break the spirit of the rules while following the letter. Just don't, please.

Even if you notice or suspect that other people are breaking some of the rules, please continue following all the rules. Don't let one uncooperative person start an avalanche of defection. That includes if you notice that people are not voting according to the rules. If necessary, complain in the meta thread.

Okay, that's enough rules for today. Have fun! I love you! ❤ ❤ ❤ ٩(⁎❛ᴗ❛⁎)۶

EDIT: Oops, I forgot the most important part. LOL! The topic is "anything that makes you happy" (basically Open Thread / Bragging Thread / etc., but only the positive things).

This is part of a semi-monthly reading group on Eliezer Yudkowsky's ebook, Rationality: From AI to Zombies. For more information about the group, see the announcement post.

Welcome to the Rationality reading group. This fortnight we discuss Part V: Value Theory (pp. 1359-1450). This post summarizes each article of the sequence, linking to the original LessWrong post where available.

V. Value Theory

264. Where Recursive Justification Hits Bottom - Ultimately, when you reflect on how your mind operates, and consider questions like "why does Occam's Razor work?" and "why do I expect the future to be like the past?", you have no other option but to use your own mind. There is no way to jump to an ideal state of pure emptiness and evaluate these claims without using your existing mind.

265. My Kind of Reflection - A few key differences between Eliezer Yudkowsky's ideas on reflection and the ideas of other philosophers.

266. No Universally Compelling Arguments - Because minds are physical processes, it is theoretically possible to specify a mind which draws any conclusion in response to any argument. There is no argument that will convince every possible mind.

267. Created Already in Motion - There is no computer program so persuasive that you can run it on a rock. A mind, in order to be a mind, needs some sort of dynamic rules of inference or action. A mind has to be created already in motion.

268. Sorting Pebbles into Correct Heaps - A parable about an imaginary society that has arbitrary, alien values.

269. 2-Place and 1-Place Words - It is possible to talk about "sexiness" as a property of an observer and a subject. It is also equally possible to talk about "sexiness" as a property of a subject, as long as each observer can have a different process to determine how sexy someone is. Failing to do either of these will cause you trouble.

270. What Would You Do Without Morality? - If your own theory of morality was disproved, and you were persuaded that there was no morality, that everything was permissible and nothing was forbidden, what would you do? Would you still tip cabdrivers?

271. Changing Your Metaethics - Discusses the various lines of retreat that have been set up in the discussion on metaethics.

272. Could Anything Be Right? - You do know quite a bit about morality. It's not perfect information, surely, or absolutely reliable, but you have someplace to start. If you didn't, you'd have a much harder time thinking about morality than you do.

273. Morality as Fixed Computation - A clarification about Yudkowsky's metaethics.

274. Magical Categories - We underestimate the complexity of our own unnatural categories. This doesn't work when you're trying to build a FAI.

275. The True Prisoner's Dilemma - The standard visualization for the Prisoner's Dilemma doesn't really work on humans. We can't pretend we're completely selfish.

276. Sympathetic Minds - Mirror neurons are neurons that fire both when performing an action oneself, and watching someone else perform the same action - for example, a neuron that fires when you raise your hand or watch someone else raise theirs. We predictively model other minds by putting ourselves in their shoes, which is empathy. But some of our desire to help relatives and friends, or be concerned with the feelings of allies, is expressed as sympathy, feeling what (we believe) they feel. Like "boredom", the human form of sympathy would not be expected to arise in an arbitrary expected-utility-maximizing AI. Most such agents would regard any agents in its environment as a special case of complex systems to be modeled or optimized; it would not feel what they feel.

277. High Challenge - Life should not always be made easier for the same reason that video games should not always be made easier. Think in terms of eliminating low-quality work to make way for high-quality work, rather than eliminating all challenge. One needs games that are fun to play and not just fun to win. Life's utility function is over 4D trajectories, not just 3D outcomes. Values can legitimately be over the subjective experience, the objective result, and the challenging process by which it is achieved - the traveller, the destination and the journey.

278. Serious Stories - Stories and lives are optimized according to rather different criteria. Advice on how to write fiction will tell you that "stories are about people's pain" and "every scene must end in disaster". I once assumed that it was not possible to write any story about a successful Singularity because the inhabitants would not be in any pain; but something about the final conclusion that the post-Singularity world would contain no stories worth telling seemed alarming. Stories in which nothing ever goes wrong, are painful to read; would a life of endless success have the same painful quality? If so, should we simply eliminate that revulsion via neural rewiring? Pleasure probably does retain its meaning in the absence of pain to contrast it; they are different neural systems. The present world has an imbalance between pain and pleasure; it is much easier to produce severe pain than correspondingly intense pleasure. One path would be to address the imbalance and create a world with more pleasures, and free of the more grindingly destructive and pointless sorts of pain. Another approach would be to eliminate pain entirely. I feel like I prefer the former approach, but I don't know if it can last in the long run.

279. Value is Fragile - An interesting universe, that would be incomprehensible to the universe today, is what the future looks like if things go right. There are a lot of things that humans value that if you did everything else right, when building an AI, but left out that one thing, the future would wind up looking dull, flat, pointless, or empty. Any Future not shaped by a goal system with detailed reliable inheritance from human morals and metamorals, will contain almost nothing of worth.

280. The Gift We Give to Tomorrow - How did love ever come into the universe? How did that happen, and how special was it, really?

This has been a collection of notes on the assigned sequence for this fortnight. The most important part of the reading group though is discussion, which is in the comments section. Please remember that this group contains a variety of levels of expertise: if a line of discussion seems too basic or too incomprehensible, look around for one that suits you better!

The next reading will cover Part W: Quantified Humanism (pp. 1453-1514) and Interlude: The Twelve Virtues of Rationality (pp. 1516-1521). The discussion will go live on Wednesday, 23 March 2016, right here on the discussion forum of LessWrong.

There have been a couple of brief discussions of this in the Open Thread, but it seems likely to generate more so here's a place for it.

The original paper in Nature about AlphaGo.

Google Asia Pacific blog, where results will be posted. DeepMind's YouTube channel, where the games are being live-streamed.

Discussion on Hacker News after AlphaGo's win of the first game.