= 27d567bc2d48d9f40e9ccc20ea370b15)
[LINK] Wait But Why - The AI Revolution Part 2
Part 1 was previously posted and it seemed that people likd it, so I figured that I should post part 2 - http://waitbutwhy.com/2015/01/artificial-intelligence-revolution-2.html
Slides online from "The Future of AI: Opportunities and Challenges"
In the first weekend of this year, the Future of Life institute hosted a landmark conference in Puerto Rico: "The Future of AI: Opportunities and Challenges". The conference was unusual in that it was not made public until it was over, and the discussions were under Chatham House rules. The slides from the conference are now available. The list of attenders includes a great many famous names as well as lots of names familiar to those of us on Less Wrong: Elon Musk, Sam Harris, Margaret Boden, Thomas Dietterich, all three DeepMind founders, and many more.
This is shaping up to be another extraordinary year for AI risk concerns going mainstream!
Questions of Reasoning under Logical Uncertainty
I'm pleased to announce a new paper from MIRI: Questions of Reasoning Under Logical Uncertainty.
Abstract:
A logically uncertain reasoner would be able to reason as if they know both a programming language and a program, without knowing what the program outputs. Most practical reasoning involves some logical uncertainty, but no satisfactory theory of reasoning under logical uncertainty yet exists. A better theory of reasoning under logical uncertainty is needed in order to develop the tools necessary to construct highly reliable artificial reasoners. This paper introduces the topic, discusses a number of historical results, and describes a number of open problems.
Following Corrigibility and Toward Idealized Decision Theory, this is the third in a series of six papers motivating MIRI's technical research agenda. This paper mostly motivates and summarizes the state of the field, and contains one very minor new technical result. Readers looking for more technical meat can find it in Paul Christiano's paper Non-Omniscience, Probabilistic Inference, and Metamathematics, published mid-2014. This paper is instead intended to motivate the study of logical uncertainty as relevant to the design of highly reliable smarter-than-human systems. The introduction runs as follows:
Formalizing Two Problems of Realistic World Models
I'm pleased to announce a new paper from MIRI: Formalizing Two Problems of Realistic World Models.
Abstract:
An intelligent agent embedded within the real world must reason about an environment which is larger than the agent, and learn how to achieve goals in that environment. We discuss attempts to formalize two problems: one of induction, where an agent must use sensory data to infer a universe which embeds (and computes) the agent, and one of interaction, where an agent must learn to achieve complex goals in the universe. We review related problems formalized by Solomonoff and Hutter, and explore challenges that arise when attempting to formalize analogous problems in a setting where the agent is embedded within the environment.
This is the fifth of six papers discussing active research topics that we've been looking into at MIRI. It discusses a few difficulties that arise when attempting to formalize problems of induction and evaluation in settings where an agent is attempting to learn about (and act upon) a universe from within. These problems have been much discussed on LessWrong; for further reading, see the links below. This paper is intended to better introduce the topic, and motivate it as relevant to FAI research.
- Intelligence Metrics with Naturalized Induction using UDT
- Building Phenomenological Bridges
- Failures of an Embodied AIXI
- The Naturalized Induction wiki page
The (rather short) introduction to the paper is reproduced below.
Vingean Reflection: Reliable Reasoning for Self-Improving Agents
I'm pleased to announce a new paper from MIRI: Vingean Reflection: Reliable Reasoning for Self-Improving Agents.
Abstract:
Today, human-level machine intelligence is in the domain of futurism, but there is every reason to expect that it will be developed eventually. Once artificial agents become able to improve themselves further, they may far surpass human intelligence, making it vitally important to ensure that the result of an "intelligence explosion" is aligned with human interests. In this paper, we discuss one aspect of this challenge: ensuring that the initial agent's reasoning about its future versions is reliable, even if these future versions are far more intelligent than the current reasoner. We refer to reasoning of this sort as Vingean Reflection.
A self-improving agent must reason about the behavior of its smarter successors in abstract terms, since if it could predict their actions in detail, it would already be as smart as them. This is called the Vingean principle, and we argue that theoretical work on Vingean reflection should focus on formal models that reflect this principle. However, the framework of expected utility maximization, commonly used to model rational agents, fails to do so. We review a body of work which instead investigates agents that use formal proofs to reason about their successors. While it is unlikely that real-world agents would base their behavior entirely on formal proofs, this appears to be the best currently available formal model of abstract reasoning, and work in this setting may lead to insights applicable to more realistic approaches to Vingean reflection.
This is the fourth in a series of six papers discussing various components of MIRI's technical research agenda. It motivates the field of Vingean reflection, which studies methods by which agents can reason reliably about agents that are more intelligent than themselves. Toy models used to study this problem in the past include the "tiling agent" models that have been discussed on LessWrong in the past. The introduction to the paper runs as follows:
Apptimize -- rationalist startup hiring engineers
Apptimize is a 2-year old startup closely connected with the rationalist community, one of the first founded by CFAR alumni. We make “lean” possible for mobile apps -- our software lets mobile developers update or A/B test their apps in minutes, without submitting to the App Store. Our customers include big companies such as Nook and Ebay, as well as Top 10 apps such as Flipagram. When companies evaluate our product against competitors, they’ve chosen us every time.
We work incredibly hard, and we’re striving to build the strongest engineering team in the Bay Area. If you’re a good developer, we have a lot to offer.
Team
-
Our team of 14 includes 7 MIT alumni, 3 ex-Googlers, 1 Wharton MBA, 1 CMU CS alum, 1 Stanford alum, 2 MIT Masters, 1 MIT Ph. D. candidate, and 1 “20 Under 20” Thiel Fellow. Our CEO was also just named to the Forbes “30 Under 30”
-
David Salamon, Anna Salamon’s brother, built much of our early product
-
Our CEO is Nancy Hua, while our Android lead is "20 under 20" Thiel Fellow James Koppel. They met after James spoke at the Singularity Summit
-
HP:MoR is required reading for the entire company
-
We evaluate candidates on curiosity even before evaluating them technically
-
Seriously, our team is badass. Just look
Self Improvement
-
You will have huge autonomy and ownership over your part of the product. You can set up new infrastructure and tools, expense business products and services, and even subcontract some of your tasks if you think it's a good idea
-
You will learn to be a more goal-driven agent, and understand the impact of everything you do on the rest of the business
-
Access to our library of over 50 books and audiobooks, and the freedom to purchase more
-
Everyone shares insights they’ve had every week
-
Self-improvement is so important to us that we only hire people committed to it. When we say that it’s a company value, we mean it
The Job
-
Our mobile engineers dive into the dark, undocumented corners of iOS and Android, while our backend crunches data from billions of requests per day
-
Engineers get giant monitors, a top-of-the-line MacBook pro, and we’ll pay for whatever else is needed to get the job done
-
We don’t demand prior experience, but we do demand the fearlessness to jump outside your comfort zone and job description. That said, our website uses AngularJS, jQuery, and nginx, while our backend uses AWS, Java (the good parts), and PostgreSQL
-
We don’t have gratuitous perks, but we have what counts: Free snacks and catered meals, an excellent health and dental plan, and free membership to a gym across the street
-
Seriously, working here is awesome. As one engineer puts it, “we’re like a family bent on taking over the world”
If you’re interested, send some Bayesian evidence that you’re a good match to jobs@apptimize.com
2014 Survey Results
Thanks to everyone who took the 2014 Less Wrong Census/Survey. Extra thanks to Ozy, who did a lot of the number crunching work.
This year's results are below. Some of them may make more sense in the context of the original survey questions, which can be seen here. Please do not try to take the survey as it is over and your results will not be counted.
I. Population
There were 1503 respondents over 27 days. The last survey got 1636 people over 40 days. The last four full days of the survey saw nineteen, six, and four responses, for an average of about ten. If we assume the next thirteen days had also gotten an average of ten responses - which is generous, since responses tend to trail off with time - then we would have gotten about as many people as the last survey. There is no good evidence here of a decline in population, although it is perhaps compatible with a very small decline.
II. Demographics
Sex
Female: 179, 11.9%
Male: 1311, 87.2%
Gender
F (cisgender): 150, 10.0%
F (transgender MtF): 24, 1.6%
M (cisgender): 1245, 82.8%
M (transgender FtM): 5, 0.3%
Other: 64, 4.3%
Sexual Orientation
Asexual: 59, 3.9%
Bisexual: 216, 14.4%
Heterosexual: 1133, 75.4%
Homosexual: 47, 3.1%
Other: 35, 2.3%
[This question was poorly worded and should have acknowledged that people can both be asexual and have a specific orientation; as a result it probably vastly undercounted our asexual readers]
Relationship Style
Prefer monogamous: 778, 51.8%
Prefer polyamorous: 227, 15.1%
Uncertain/no preference: 464, 30.9%
Other: 23, 1.5%
Number of Partners
0: 738, 49.1%
1: 674, 44.8%
2: 51, 3.4%
3: 17, 1.1%
4: 7, 0.5%
5: 1, 0.1%
Lots and lots: 3, 0.2%
Relationship Goals
Currently not looking for new partners: 648, 43.1%
Open to new partners: 467, 31.1%
Seeking more partners: 370, 24.6%
[22.2% of people who don’t have a partner aren’t looking for one.]
Relationship Status
Married: 274, 18.2%
Relationship: 424, 28.2%
Single: 788, 52.4%
[6.9% of single people have at least one partner; 1.8% have more than one.]
Living With
Alone: 345, 23.0%
With parents and/or guardians: 303, 20.2%
With partner and/or children: 411, 27.3%
With roommates: 428, 28.5%
Children
0: 1317, 81.6%
1: 66, 4.4%
2: 78, 5.2%
3: 17, 1.1%
4: 6, 0.4%
5: 3, 0.2%
6: 1, 0.1%
Lots and lots: 1, 0.1%
Want More Children?
Yes: 549, 36.1%
Uncertain: 426, 28.3%
No: 516, 34.3%
[418 of the people who don’t have children don’t want any, suggesting that the LW community is 27.8% childfree.]
Country
United States, 822, 54.7%
United Kingdom, 116, 7.7%
Canada, 88, 5.9%
Australia: 83, 5.5%
Germany, 62, 4.1%
Russia, 26, 1.7%
Finland, 20, 1.3%
New Zealand, 20, 1.3%
India, 17, 1.1%
Brazil: 15, 1.0%
France, 15, 1.0%
Israel, 15, 1.0%
Lesswrongers Per Capita
Finland: 1/271,950
New Zealand: 1/223,550
Australia: 1/278,674
United States: 1/358,390
Canada: 1/399,545
Israel: 1/537,266
United Kingdom: 1/552,586
Germany: 1/1,290,323
France: 1/ 4,402,000
Russia: 1/ 5,519,231
Brazil: 1/ 13,360,000
India: 1/ 73,647,058
Race
Asian (East Asian): 59. 3.9%
Asian (Indian subcontinent): 33, 2.2%
Black: 12. 0.8%
Hispanic: 32, 2.1%
Middle Eastern: 9, 0.6%
Other: 50, 3.3%
White (non-Hispanic): 1294, 86.1%
Work Status
Academic (teaching): 86, 5.7%
For-profit work: 492, 32.7%
Government work: 59, 3.9%
Homemaker: 8, 0.5%
Independently wealthy: 9, 0.6%
Nonprofit work: 58, 3.9%
Self-employed: 122, 5.8%
Student: 553, 36.8%
Unemployed: 103, 6.9%
Profession
Art: 22, 1.5%
Biology: 29, 1.9%
Business: 35, 4.0%
Computers (AI): 42, 2.8%
Computers (other academic): 106, 7.1%
Computers (practical): 477, 31.7%
Engineering: 104, 6.1%
Finance/Economics: 71, 4.7%
Law: 38, 2.5%
Mathematics: 121, 8.1%
Medicine: 32, 2.1%
Neuroscience: 18, 1.2%
Philosophy: 36, 2.4%
Physics: 65, 4.3%
Psychology: 31, 2.1%
Other: 157, 10.2%
Other “hard science”: 25, 1.7%
Other “social science”: 34, 2.3%
Degree
None: 74, 4.9%
High school: 347, 23.1%
2 year degree: 64, 4.3%
Bachelors: 555, 36.9%
Masters: 278, 18.5%
JD/MD/other professional degree: 44, 2.9%
PhD: 105, 7.0%
Other: 24, 1.4%
III. Mental Illness
535 answer “no” to all the mental illness questions. Upper bound: 64.4% of the LW population is mentally ill.
393 answer “yes” to at least one mental illness question. Lower bound: 26.1% of the LW population is mentally ill. Gosh, we have a lot of self-diagnosers.
Depression
Yes, I was formally diagnosed: 273, 18.2%
Yes, I self-diagnosed: 383, 25.5%
No: 759, 50.5%
OCD
Yes, I was formally diagnosed: 30, 2.0%
Yes, I self-diagnosed: 76, 5.1%
No: 1306, 86.9%
Autism spectrum
Yes, I was formally diagnosed: 98, 6.5%
Yes, I self-diagnosed: 168, 11.2%
No: 1143, 76.0%
Bipolar
Yes, I was formally diagnosed: 33, 2.2%
Yes, I self-diagnosed: 49, 3.3%
No: 1327, 88.3%
Anxiety disorder
Yes, I was formally diagnosed: 139, 9.2%
Yes, I self-diagnosed: 237, 15.8%
No: 1033, 68.7%
BPD
Yes, I was formally diagnosed: 5, 0.3%
Yes, I self-diagnosed: 19, 1.3%
No: 1389, 92.4%
[Ozy says: RATIONALIST BPDERS COME BE MY FRIEND]
Schizophrenia
Yes, I was formally diagnosed: 7, 0.5%
Yes, I self-diagnosed: 7, 0.5%
No: 1397, 92.9%
IV. Politics, Religion, Ethics
Politics
Communist: 9, 0.6%
Conservative: 67, 4.5%
Liberal: 416, 27.7%
Libertarian: 379, 25.2%
Social Democratic: 585, 38.9%
[The big change this year was that we changed "Socialist" to "Social Democratic". Even though the description stayed the same, about eight points worth of Liberals switched to Social Democrats, apparently more willing to accept that label than "Socialist". The overall supergroups Libertarian vs. (Liberal, Social Democratic) vs. Conservative remain mostly unchanged.]
Politics (longform)
Anarchist: 40, 2.7%
Communist: 9, 0.6%
Conservative: 23, 1.9%
Futarchist: 41, 2.7%
Left-Libertarian: 192, 12.8%
Libertarian: 164, 10.9%
Moderate: 56, 3.7%
Neoreactionary: 29, 1.9%
Social Democrat: 162, 10.8%
Socialist: 89, 5.9%
[Amusing politics answers include anti-incumbentist, having-well-founded-opinions-is-hard-but-I’ve-come-to-recognize-the-pragmatism-of-socialism-I-don’t-know-ask-me-again-next-year, pirate, progressive social democratic environmental liberal isolationist freedom-fries loving pinko commie piece of shit, republic-ist aka read the federalist papers, romantic reconstructionist, social liberal fiscal agnostic, technoutopian anarchosocialist (with moderate snark), whatever it is that Scott is, and WHY ISN’T THERE AN OPTION FOR NONE SO I CAN SIGNAL MY OBVIOUS OBJECTIVITY WITH MINIMAL EFFORT. Ozy would like to point out to the authors of manifestos that no one will actually read their manifestos except zir, and they might want to consider posting them to their own blogs.]
American Parties
Democratic Party: 221, 14.7%
Republican Party: 55, 3.7%
Libertarian Party: 26, 1.7%
Other party: 16, 1.1%
No party: 415, 27.6%
Non-Americans who really like clicking buttons: 415, 27.6%
Voting
Yes: 881, 58.6%
No: 444, 29.5%
My country doesn’t hold elections: 5, 0.3%
Religion
Atheist and not spiritual: 1054, 70.1%
Atheist and spiritual: 150, 10.0%
Agnostic: 156, 10.4%
Lukewarm theist: 44, 2.9%
Deist/pantheist/etc.: 22,, 1.5%
Committed theist: 60, 4.0%
Religious Denomination
Christian (Protestant): 53, 3.5%
Mixed/Other: 32, 2.1%
Jewish: 31, 2.0%
Buddhist: 30, 2.0%
Christian (Catholic): 24, 1.6%
Unitarian Universalist or similar: 23, 1.5%
[Amusing denominations include anti-Molochist, CelestAI, cosmic engineers, Laziness, Thelema, Resimulation Theology, and Pythagorean. The Cultus Deorum Romanorum practitioner still needs to contact Ozy so they can be friends.]
Family Religion
Atheist and not spiritual: 213, 14.2%
Atheist and spiritual: 74, 4.9%
Agnostic: 154. 10.2%
Lukewarm theist: 541, 36.0%
Deist/Pantheist/etc.: 28, 1.9%
Committed theist: 388, 25.8%
Religious Background
Christian (Protestant): 580, 38.6%
Christian (Catholic): 378, 25.1%
Jewish: 141, 9.4%
Christian (other non-protestant): 88, 5.9%
Mixed/Other: 68, 4.5%
Unitarian Universalism or similar: 29, 1.9%
Christian (Mormon): 28, 1.9%
Hindu: 23, 1.5%’
Moral Views
Accept/lean towards consequentialism: 901, 60.0%
Accept/lean towards deontology: 50, 3.3%
Accept/lean towards natural law: 48, 3.2%
Accept/lean towards virtue ethics: 150, 10.0%
Accept/lean towards contractualism: 79, 5.3%
Other/no answer: 239, 15.9%
Meta-ethics
Constructivism: 474, 31.5%
Error theory: 60, 4.0%
Non-cognitivism: 129, 8.6%
Subjectivism: 324, 21.6%
Substantive realism: 209, 13.9%
V. Community Participation
Less Wrong Use
Lurker: 528, 35.1%
I’ve registered an account: 221, 14.7%
I’ve posted a comment: 419, 27.9%
I’ve posted in Discussion: 207, 13.8%
I’ve posted in Main: 102, 6.8%
Sequences
Never knew they existed until this moment: 106, 7.1%
Knew they existed, but never looked at them: 42, 2.8%
Some, but less than 25%: 270, 18.0%
About 25%: 181, 12.0%
About 50%: 209, 13.9%
About 75%: 242, 16.1%
All or almost all: 427, 28.4%
Meetups
Yes, regularly: 154, 10.2%
Yes, once or a few times: 325, 21.6%
No: 989, 65.8%
Community
Yes, all the time: 112, 7.5%
Yes, sometimes: 191, 12.7%
No: 1163, 77.4%
Romance
Yes: 82, 5.5%
I didn’t meet them through the community but they’re part of the community now: 79, 5.3%
No: 1310, 87.2%
CFAR Events
Yes, in 2014: 45, 3.0%
Yes, in 2013: 60, 4.0%
Both: 42, 2.8%
No: 1321, 87.9%
CFAR Workshop
Yes: 109, 7.3%
No: 1311, 87.2%
[A couple percent more people answered 'yes' to each of meetups, physical interactions, CFAR attendance, and romance this time around, suggesting the community is very very gradually becoming more IRL. In particular, the number of people meeting romantic partners through the community increased by almost 50% over last year.]
HPMOR
Yes: 897, 59.7%
Started but not finished: 224, 14.9%
No: 254, 16.9%
Referrals
Referred by a link: 464, 30.9%
HPMOR: 385, 25.6%
Been here since the Overcoming Bias days: 210, 14.0%
Referred by a friend: 199, 13.2%
Referred by a search engine: 114, 7.6%
Referred by other fiction: 17, 1.1%
[Amusing responses include “a rationalist that I follow on Tumblr”, “I’m a student of tribal cultishness”, and “It is difficult to recall details from the Before Time. Things were brighter, simpler, as in childhood or a dream. There has been much growth, change since then. But also loss. I can't remember where I found the link, is what I'm saying.”]
Blog Referrals
Slate Star Codex: 40, 2.6%
Reddit: 25, 1.6%
Common Sense Atheism: 21, 1.3%
Hacker News: 20, 1.3%
Gwern: 13, 1.0%
VI. Other Categorical Data
Cryonics Status
Don’t understand/never thought about it: 62, 4.1%
Don’t want to: 361, 24.0%
Considering it: 551, 36.7%
Haven’t gotten around to it: 272, 18.1%
Unavailable in my area: 126, 8.4%
Yes: 64, 4.3%
Type of Global Catastrophic Risk
Asteroid strike: 64, 4.3%
Economic/political collapse: 151, 10.0%
Environmental collapse: 218, 14.5%
Nanotech/grey goo: 47, 3.1%
Nuclear war: 239, 15.8%
Pandemic (bioengineered): 310, 20.6%
Pandemic (natural): 113. 7.5%
Unfriendly AI: 244, 16.2%
[Amusing answers include ennui/eaten by Internet, Friendly AI, “Greens so weaken the rich countries that barbarians conquer us”, and Tumblr.]
Effective Altruism (do you self-identify)
Yes: 422, 28.1%
No: 758, 50.4%
[Despite some impressive outreach by the EA community, numbers are largely the same as last year]
Effective Altruism (do you participate in community)
Yes: 191, 12.7%
No: 987, 65.7%
Vegetarian
Vegan: 31, 2.1%
Vegetarian: 114, 7.6%
Other meat restriction: 252, 16.8%
Omnivore: 848, 56.4%
Paleo Diet
Yes: 33, 2.2%
Sometimes: 209, 13.9%
No: 1111, 73.9%
Food Substitutes
Most of my calories: 8. 0.5%
Sometimes: 101, 6.7%
Tried: 196, 13.0%
No: 1052, 70.0%
Gender Default
I only identify with my birth gender by default: 681, 45.3%
I strongly identify with my birth gender: 586, 39.0%
Books
<5: 198, 13.2%
5 - 10: 384, 25.5%
10 - 20: 328, 21.8%
20 - 50: 264, 17.6%
50 - 100: 105, 7.0%
> 100: 49, 3.3%
Birth Month
Jan: 109, 7.3%
Feb: 90, 6.0%
Mar: 123, 8.2%
Apr: 126, 8.4%
Jun: 107, 7.1%
Jul: 109, 7.3%
Aug: 120, 8.0%
Sep: 94, 6.3%
Oct: 111, 7.4%
Nov: 102, 6.8%
Dec: 106, 7.1%
[Despite my hope of something turning up here, these results don't deviate from chance]
Handedness
Right: 1170, 77.8%
Left: 143, 9.5%
Ambidextrous: 37, 2.5%
Unsure: 12, 0.8%
Previous Surveys
Yes: 757, 50.7%
No: 598, 39.8%
Favorite Less Wrong Posts (all > 5 listed)
An Alien God: 11
Joy In The Merely Real: 7
Dissolving Questions About Disease: 7
Politics Is The Mind Killer: 6
That Alien Message: 6
A Fable Of Science And Politics: 6
Belief In Belief: 5
Generalizing From One Example: 5
Schelling Fences On Slippery Slopes: 5
Tsuyoku Naritai: 5
VII. Numeric Data
Age: 27.67 + 8.679 (22, 26, 31) [1490]
IQ: 138.25 + 15.936 (130.25, 139, 146) [472]
SAT out of 1600: 1470.74 + 113.114 (1410, 1490, 1560) [395]
SAT out of 2400: 2210.75 + 188.94 (2140, 2250, 2320) [310]
ACT out of 36: 32.56 + 2.483 (31, 33, 35) [244]
Time in Community: 2010.97 + 2.174 (2010, 2011, 2013) [1317]
Time on LW: 15.73 + 95.75 (2, 5, 15) [1366]
Karma Score: 555.73 + 2181.791 (0, 0, 155) [1335]
P Many Worlds: 47.64 + 30.132 (20, 50, 75) [1261]
P Aliens: 71.52 + 34.364 (50, 90, 99) [1393]
P Aliens (Galaxy): 41.2 + 38.405 (2, 30, 80) [1379]
P Supernatural: 6.68 + 20.271 (0, 0, 1) [1386]
P God: 8.26 + 21.088 (0, 0.01, 3) [1376]
P Religion: 4.99 + 18.068 (0, 0, 0.5) [1384]
P Cryonics: 22.34 + 27.274 (2, 10, 30) [1399]
P Anti-Agathics: 24.63 + 29.569 (1, 10, 40) [1390]
P Simulation 24.31 + 28.2 (1, 10, 50) [1320]
P Warming 81.73 + 24.224 (80, 90, 98) [1394]
P Global Catastrophic Risk 72.14 + 25.620 (55, 80, 90) [1394]
Singularity: 2143.44 + 356.643 (2060, 2090, 2150) [1177]
[The mean for this question is almost entirely dependent on which stupid responses we choose to delete as outliers; the median practically never changes]
Abortion: 4.38 + 1.032 (4, 5, 5) [1341]
Immigration: 4 + 1.078 (3, 4, 5) [1310]
Taxes : 3.14 + 1.212 (2, 3, 4) [1410] (from 1 - should be lower to 5 - should be higher)
Minimum Wage: 3.21 + 1.359 (2, 3, 4) [1298] (from 1 - should be lower to 5 - should be higher)
Feminism: 3.67 + 1.221 (3, 4, 5) [1332]
Social Justice: 3.15 + 1.385 (2, 3, 4) [1309]
Human Biodiversity: 2.93 + 1.201 (2, 3, 4) [1321]
Basic Income: 3.94 + 1.087 (3, 4, 5) [1314]
Great Stagnation: 2.33 + .959 (2, 2, 3) [1302]
MIRI Mission: 3.90 + 1.062 (3, 4, 5) [1412]
MIRI Effectiveness: 3.23 + .897 (3, 3, 4) [1336]
[Remember, all of these are asking you to rate your belief in/agreement with the concept on a scale of 1 (bad) to 5 (great)]
Income: 54129.37 + 66818.904 (10,000, 30,800, 80,000) [923]
Charity: 1996.76 + 9492.71 (0, 100, 800) [1009]
MIRI/CFAR: 511.61 + 5516.608 (0, 0, 0) [1011]
XRisk: 62.50 + 575.260 (0, 0, 0) [980]
Older siblings: 0.51 + .914 (0, 0, 1) [1332]
Younger siblings: 1.08 + 1.127 (0, 1, 1) [1349]
Height: 178.06 + 11.767 (173, 179, 184) [1236]
Hours Online: 43.44 + 25.452 (25, 40, 60) [1221]
Bem Sex Role Masculinity: 42.54 + 9.670 (36, 42, 49) [1032]
Bem Sex Role Femininity: 42.68 + 9.754 (36, 43, 50) [1031]
Right Hand: .97 + 0.67 (.94, .97, 1.00)
Left Hand: .97 + .048 (.94, .97, 1.00)
VIII. Fishing Expeditions
[correlations, in descending order]
SAT Scores out of 1600/SAT Scores out of 2400 .844 (59)
P Supernatural/P God .697 (1365)
Feminism/Social Justice .671 (1299)
P God/P Religion .669 (1367)
P Supernatural/P Religion .631 (1372)
Charity Donations/MIRI and CFAR Donations .619 (985)
P Aliens/P Aliens 2 .607 (1376)
Taxes/Minimum Wage .587 (1287)
SAT Score out of 2400/ACT Score .575 (89)
Age/Number of Children .506 (1480)
P Cryonics/P Anti-Agathics .484 (1385)
SAT Score out of 1600/ACT Score .480 (81)
Minimum Wage/Social Justice .456 (1267)
Taxes/Social Justice .427 (1281)
Taxes/Feminism .414 (1299)
MIRI Mission/MIRI Effectiveness .395 (1331)
P Warming/Taxes .385 (1261)
Taxes/Basic Income .383 (1285)
Minimum Wage/Feminism .378 (1286)
P God/Abortion -.378 (1266)
Immigration/Feminism .365 (1296)
P Supernatural/Abortion -.362 (1276)
Feminism/Human Biodiversity -.360 (1306)
MIRI and CFAR Donations/Other XRisk Charity Donations .345 (973)
Social Justice/Human Biodiversity -.341 (1288)
P Religion/Abortion -.326 (1275)
P Warming/Minimum Wage .324 (1248)
Minimum Wage/Basic Income .312 (1276)
P Warming/Basic Income .306 (1260)
Immigration/Social Justice .294 (1278)
P Anti-Agathics/MIRI Mission .293 (1351)
P Warming/Feminism .285 (1281)
P Many Worlds/P Anti-Agathics .276 (1245)
Social Justice/Femininity .267 (990)
Minimum Wage/Human Biodiversity -.264 (1274)
Immigration/Human Biodiversity -.263 (1286)
P Many Worlds/MIRI Mission .263 (1233)
P Aliens/P Warming .262 (1365)
P Warming/Social Justice .257 (1262)
Taxes/Human Biodiversity -.252 (1291)
Social Justice/Basic Income .251 (1281)
Feminism/Femininity .250 (1003)
Older Siblings/Younger Siblings -.243 (1321)
Charity Donations/Other XRisk Charity Donations .240 (957
P Anti-Agathics/P Simulation .238 (1312)
Abortion/Minimum Wage .229 (1293)
Feminism/Basic Income .227 (1297)
Abortion/Feminism .226 (1321)
P Cryonics/MIRI Mission .223 (1360)
Immigration/Basic Income .208 (1279)
P Many Worlds/P Cryonics .202 (1251)
Number of Current Partners/Femininity: .202 (1029)
P Warming/Immigration .202 (1260)
P Warming/Abortion .201 (1289)
Abortion/Taxes .198 (1304)
Age/P Simulation .197 (1313)
Political Interest/Masculinity .194 (1011)
P Cryonics/MIRI Effectiveness .191 (1285)
Abortion/Social Justice .191 (1301)
P Simulation/MIRI Mission .188 (1290)
P Many Worlds/P Warming .188 (1240)
Age/Number of Current Partners .184 (1480)
P Anti-Agathics/MIRI Effectiveness .183 (1277)
P Many Worlds/P Simulation .181 (1211)
Abortion/Immigration .181 (1304)
Number of Current Partners/Number of Children .180 (1484)
P Cryonics/P Simulation .174 (1315)
P Global Catastrophic Risk/MIRI Mission -.174 (1359)
Minimum Wage/Femininity .171 (981)
Abortion/Basic Income .170 (1302)
Age/P Cryonics -.165 (1391)
Immigration/Taxes .165 (1293)
P Warming/Human Biodiversity -.163 (1271)
P Aliens 2/Warming .160 (1353)
Abortion/Younger Siblings -.155 (1292)
P Religion/Meditate .155 (1189)
Feminism/Masculinity -.155 (1004)
Immigration/Femininity .155 (988)
P Supernatural/Basic Income -.153 (1246)
P Supernatural/P Warming -.152 (1361)
Number of Current Partners/Karma Score .152 (1332)
P Many Worlds/MIRI Effectiveness .152 (1181)
Age/MIRI Mission -.150 (1404)
P Religion/P Warming -.150 (1358)
P Religion/Basic Income -.146 (1245)
P God/Basic Income -.146 (1237)
Human Biodiversity/Femininity -.145 (999)
P God/P Warming -.144 (1351)
Taxes/Femininity .142 (987)
Number of Children/Younger Siblings .138 (1343)
Number of Current Partners/Masculinity: .137 (1030)
P Many Worlds/P God -.137 (1232)
Age/Charity Donations .133 (1002)
P Anti-Agathics/P Global Catastrophic Risk -.132 (1373)
P Warming/Masculinity -.132 (992)
P Global Catastrophic Risk/MIRI and CFAR Donations -.132 (982)
P Supernatural/Singularity .131 (1148)
God/Taxes -.130 (1240)
Age/P Anti-Agathics -.128 (1382)
P Aliens/Taxes .127(1258)
Feminism/Great Stagnation -.127 (1287)
P Many Worlds/P Supernatural -.127 (1241)
P Aliens/Abortion .126 (1284)
P Anti-Agathics/Great Stagnation -.126 (1248)
P Anti-Agathics/P Warming .125 (1370)
Age/P Aliens .124 (1386)
P Aliens/Minimum Wage .124 (1245)
P Aliens/P Global Catastrophic Risk .122 (1363)
Age/MIRI Effectiveness -.122 (1328)
Age/P Supernatural .120 (1370)
P Supernatural/MIRI Mission -.119 (1345)
P Many Worlds/P Religion -.119 (1238)
P Religion/MIRI Mission -.118 (1344)
Political Interest/Social Justice .118 (1304)
P Anti-Agathics/MIRI and CFAR Donations .118 (976)
Human Biodiversity/Basic Income -.115 (1262)
P Many Worlds/Abortion .115 (1166)
Age/Karma Score .114 (1327)
P Aliens/Feminism .114 (1277)
P Many Worlds/P Global Catastrophic Risk -.114 (1243)
Political Interest/Femininity .113 (1010)
Number of Children/P Simulation -.112 (1317)
P Religion/Younger Siblings .112 (1275)
P Supernatural/Taxes -.112 (1248)
Age/Masculinity .112 (1027)
Political Interest/Taxes .111 (1305)
P God/P Simulation .110 (1296)
P Many Worlds/Basic Income .110 (1139)
P Supernatural/Younger Siblings .109 (1274)
P Simulation/Basic Income .109 (1195)
Age/P Aliens 2 .107 (1371)
MIRI Mission/Basic Income .107 (1279)
Age/Great Stagnation .107 (1295)
P Many Worlds/P Aliens .107 (1253)
Number of Current Partners/Social Justice .106 (1304)
Human Biodiversity/Great Stagnation .105 (1285)
Number of Children/Abortion -.104 (1337)
Number of Current Partners/P Cryonics -.102 (1396)
MIRI Mission/Abortion .102 (1305)
Immigration/Great Stagnation -.101 (1269)
Age/Political Interest .100 (1339)
P Global Catastrophic Risk/Political Interest .099 (1295)
P Aliens/P Religion -.099 (1357)
P God/MIRI Mission -.098 (1335)
P Aliens/P Simulation .098 (1308)
Number of Current Partners/Immigration .098 (1305)
P God/Political Interest .098 (1274)
P Warming/P Global Catastrophic Risk .096 (1377)
In addition to the Left/Right factor we had last year, this data seems to me to have an Agrees with the Sequences Factor-- the same people tend to believe in many-worlds, cryo, atheism, simulationism, MIRI’s mission and effectiveness, anti-agathics, etc. Weirdly, belief in global catastrophic risk is negatively correlated with most of the Agrees with Sequences things. Someone who actually knows how to do statistics should run a factor analysis on this data.
IX. Digit Ratios
After sanitizing the digit ratio numbers, the following correlations came up:
Digit ratio R hand was correlated with masculinity at a level of -0.180 p < 0.01
Digit ratio L hand was correlated with masculinity at a level of -0.181 p < 0.01
Digit ratio R hand was slightly correlated with femininity at a level of +0.116 p < 0.05
Holy #@!$ the feminism thing ACTUALLY HELD UP. There is a 0.144 correlation between right-handed digit ratio and feminism, p < 0.01. And an 0.112 correlation between left-handed digit ratio and feminism, p < 0.05.
The only other political position that correlates with digit ratio is immigration. There is a 0.138 correlation between left-handed digit ratio and believe in open borders p < 0.01, and an 0.111 correlation between right-handed digit ratio and belief in open borders, p < 0.05.
No digit correlation with abortion, taxes, minimum wage, social justice, human biodiversity, basic income, or great stagnation.
Okay, need to rule out that this is all confounded by gender. I ran a few analyses on men and women separately.
On men alone, the connection to masculinity holds up. Restricting sample size to men, left-handed digit ratio corresponds to masculinity with at -0.157, p < 0.01. Left handed at -0.134, p < 0.05. Right-handed correlates with femininity at 0.120, p < 0.05. The feminism correlation holds up. Restricting sample size to men, right-handed digit ratio correlates with feminism at a level of 0.149, p < 0.01. Left handed just barely fails to correlate. Both right and left correlate with immigration at 0.135, p < 0.05.
On women alone, the Bem masculinity correlation is the highest correlation we're going to get in this entire study. Right hand is -0.433, p < 0.01. Left hand is -0.299, p < 0.05. Femininity trends toward significance but doesn't get there. The feminism correlation trends toward significance but doesn't get there. In general there was too small a sample size of women to pick up anything but the most whopping effects.
Since digit ratio is related to testosterone and testosterone sometimes affects risk-taking, I wondered if it would correlate with any of the calibration answers. I selected people who had answered Calibration Question 5 incorrectly and ran an analysis to see if digit ratio was correlated with tendency to be more confident in the incorrect answer. No effect was found.
Other things that didn't correlate with digit ratio: IQ, SAT, number of current partners, tendency to work in mathematical professions.
...I still can't believe this actually worked. The finger-length/feminism connection ACTUALLY WORKED. What a world. What a world. Someone may want to double-check these results before I get too excited.
X. Calibration
There were ten calibration questions on this year's survey. Along with answers, they were:
1. What is the largest bone in the body? Femur
2. What state was President Obama born in? Hawaii
3. Off the coast of what country was the battle of Trafalgar fought? Spain
4. What Norse God was called the All-Father? Odin
5. Who won the 1936 Nobel Prize for his work in quantum physics? Heisenberg
6. Which planet has the highest density? Earth
7. Which Bible character was married to Rachel and Leah? Jacob
8. What organelle is called "the powerhouse of the cell"? Mitochondria
9. What country has the fourth-highest population? Indonesia
10. What is the best-selling computer game? Minecraft
I ran calibration scores for everybody based on how well they did on the ten calibration questions. These failed to correlate with IQ, SAT, LW karma, or any of the things you might expect to be measures of either intelligence or previous training in calibration; they didn't differ by gender, correlates of community membership, or any mental illness [deleted section about correlating with MWI and MIRI, this was an artifact].
Your answers looked like this:
The red line represents perfect calibration. Where answers dip below the line, it means you were overconfident; when they go above, it means you were underconfident.
It looks to me like everyone was horrendously underconfident on all the easy questions, and horrendously overconfident on all the hard questions. To give an example of how horrendous, people who were 50% sure of their answers to question 10 got it right only 13% of the time; people who were 100% sure only got it right 44% of the time. Obviously those numbers should be 50% and 100% respectively.
This builds upon results from previous surveys in which your calibration was also horrible. This is not a human universal - people who put even a small amount of training into calibration can become very well calibrated very quickly. This is a sign that most Less Wrongers continue to neglect the very basics of rationality and are incapable of judging how much evidence they have on a given issue. Veterans of the site do no better than newbies on this measure.
XI. Wrapping Up
To show my appreciation for everyone completing this survey, including the arduous digit ratio measurements, I have randomly chosen a person to receive a $30 monetary prize. That person is...the person using the public key "The World Is Quiet Here". If that person tells me their private key, I will give them $30.
I have removed 73 people who wished to remain private, deleted the Private Keys, and sanitized a very small amount of data. Aside from that, here are the raw survey results for your viewing and analyzing pleasure:
(as Excel)
We Haven't Uploaded Worms
In theory you can upload someone's mind onto a computer, allowing them to live forever as a digital form of consciousness, just like in the Johnny Depp film Transcendence.
But it's not just science fiction. Sure, scientists aren't anywhere near close to achieving such feat with humans (and even if they could, the ethics would be pretty fraught), but now an international team of researchers have managed to do just that with the roundworm Caenorhabditis elegans.
—Science Alert
Uploading an animal, even one as simple as c. elegans would be very impressive. Unfortunately, we're not there yet. What the people working on Open Worm have done instead is to build a working robot based on the c. elegans and show that it can do some things that the worm can do.
The c. elegans nematode has only 302 neurons, and each nematode has the same fixed pattern. We've known this pattern, or connectome, since 1986. [1] In a simple model, each neuron has a threshold and will fire if the weighted sum of its inputs is greater than that threshold. Which means knowing the connections isn't enough: we also need to know the weights and thresholds. Unfortunately, we haven't figured out a way to read these values off of real worms. Suzuki et. al. (2005) [2] ran a genetic algorithm to learn values for these parameters that would give a somewhat realistic worm and showed various wormlike behaviors in software. The recent stories about the Open Worm project have been for them doing something similar in hardware. [3]
To see why this isn't enough, consider that nematodes are capable of learning. Sasakura and Mori (2013) [5] provide a reasonable overview. For example, nematodes can learn that a certain temperature indicates food, and then seek out that temperature. They don't do this by growing new neurons or connections, they have to be updating their connection weights. All the existing worm simulations treat weights as fixed, which means they can't learn. They also don't read weights off of any individual worm, which means we can't talk about any specific worm as being uploaded.
If this doesn't count as uploading a worm, however, what would? Consider an experiment where someone trains one group of worms to respond to stimulus one way and another group to respond the other way. Both groups are then scanned and simulated on the computer. If the simulated worms responded to simulated stimulus the same way their physical versions had, that would be good progress. Additionally you would want to demonstrate that similar learning was possible in the simulated environment.
(In a 2011 post on what progress with nematodes might tell us about uploading humans I looked at some of this research before. Since then not much has changed with nematode simulation. Moore's law looks to be doing much worse in 2014 than it did in 2011, however, which makes the prospects for whole brain emulation substantially worse.)
I also posted this on my blog.
[1] The Structure of the Nervous System of the Nematode Caenorhabditis elegans, White et. al. (1986).
[2] A Model of Motor Control of the Nematode C. Elegans With Neuronal Circuits, Suzuki et. al. (2005).
[3] It looks like instead of learning weights Busbice just set them all to +1 (excitatory) and -1 (inhibitory). It's not clear to me how they knew which connections were which; my best guess is that they're using the "what happens to work" details from [2]. Their full writeup is [4].
[4] The Robotic Worm, Busbice (2014).
[5] Behavioral Plasticity, Learning, and Memory in C. Elegans, Sasakura and Mori (2013).
You have a set amount of "weirdness points". Spend them wisely.
I've heard of the concept of "weirdness points" many times before, but after a bit of searching I can't find a definitive post describing the concept, so I've decided to make one. As a disclaimer, I don't think the evidence backing this post is all that strong and I am skeptical, but I do think it's strong enough to be worth considering, and I'm probably going to make some minor life changes based on it.
-
Chances are that if you're reading this post, you're probably a bit weird in some way.
No offense, of course. In fact, I actually mean it as a compliment. Weirdness is incredibly important. If people weren't willing to deviate from society and hold weird beliefs, we wouldn't have had the important social movements that ended slavery and pushed back against racism, that created democracy, that expanded social roles for women, and that made the world a better place in numerous other ways.
Many things we take for granted now as why our current society as great were once... weird.
Joseph Overton theorized that policy develops through six stages: unthinkable, then radical, then acceptable, then sensible, then popular, then actual policy. We could see this happen with many policies -- currently same-sex marriage is making its way from popular to actual policy, but not to long ago it was merely acceptable, and not too long before that it was pretty radical.
Some good ideas are currently in the radical range. Effective altruism itself is such a collection of beliefs typical people would consider pretty radical. Many people think donating 3% of their income is a lot, let alone the 10% demand that Giving What We Can places, or the 50%+ that some people in the community do.
And that's not all. Others would suggest that everyone become vegetarian, advocating for open borders and/or universal basic income, theabolishment of gendered language, having more resources into mitigating existential risk, focusing on research into Friendly AI, cryonicsand curing death, etc.
While many of these ideas might make the world a better place if made into policy, all of these ideas are pretty weird.
Weirdness, of course, is a drawback. People take weird opinions less seriously.
The absurdity heuristic is a real bias that people -- even you -- have. If an idea sounds weird to you, you're less likely to try and believe it,even if there's overwhelming evidence. And social proof matters -- if less people believe something, people will be less likely to believe it. Lastly, don't forget the halo effect -- if one part of you seems weird, the rest of you will seem weird too!
(Update: apparently this concept is, itself, already known to social psychology as idiosyncrasy credits. Thanks, Mr. Commenter!)
...But we can use this knowledge to our advantage. The halo effect can work in reverse -- if we're normal in many ways, our weird beliefs will seem more normal too. If we have a notion of weirdness as a kind of currency that we have a limited supply of, we can spend it wisely, without looking like a crank.
All of this leads to the following actionable principles:
Recognize you only have a few "weirdness points" to spend. Trying to convince all your friends to donate 50% of their income to MIRI, become a vegan, get a cryonics plan, and demand open borders will be met with a lot of resistance. But -- I hypothesize -- that if you pick one of these ideas and push it, you'll have a lot more success.
Spend your weirdness points effectively. Perhaps it's really important that people advocate for open borders. But, perhaps, getting people to donate to developing world health would overall do more good. In that case, I'd focus on moving donations to the developing world and leave open borders alone, even though it is really important. You should triage your weirdness effectively the same way you would triage your donations.
Clean up and look good. Lookism is a problem in society, and I wish people could look "weird" and still be socially acceptable. But if you're a guy wearing a dress in public, or some punk rocker vegan advocate, recognize that you're spending your weirdness points fighting lookism, which means less weirdness points to spend promoting veganism or something else.
Advocate for more "normal" policies that are almost as good. Of course, allocating your "weirdness points" on a few issues doesn't mean you have to stop advocating for other important issues -- just consider being less weird about it. Perhaps universal basic income truly would be a very effective policy to help the poor in the United States. But reforming the earned income tax credit and relaxing zoning laws would also both do a lot to help the poor in the US, and such suggestions aren't weird.
Use the foot-in-door technique and the door-in-face technique. The foot-in-door technique involves starting with a small ask and gradually building up the ask, such as suggesting people donate a little bit effectively, and then gradually get them to take the Giving What We Can Pledge. The door-in-face technique involves making a big ask (e.g., join Giving What We Can) and then substituting it for a smaller ask, like the Life You Can Save pledge or Try Out Giving.
Reconsider effective altruism's clustering of beliefs. Right now, effective altruism is associated strongly with donating a lot of money and donating effectively, less strongly with impact in career choice, veganism, and existential risk. Of course, I'm not saying that we should drop some of these memes completely. But maybe EA should disconnect a bit more and compartmentalize -- for example, leaving AI risk to MIRI, for example, and not talk about it much, say, on 80,000 Hours. And maybe instead of asking people to both give more AND give more effectively, we could focus more exclusively on asking people to donate what they already do more effectively.
Evaluate the above with more research. While I think the evidence base behind this is decent, it's not great and I haven't spent that much time developing it. I think we should look into this more with a review of the relevant literature and some careful, targeted, market research on the individual beliefs within effective altruism (how weird are they?) and how they should be connected or left disconnected. Maybe this has already been done some?
-
Also discussed on the EA Forum and EA Facebook group.
When the uncertainty about the model is higher than the uncertainty in the model
Most models attempting to estimate or predict some elements of the world, will come with their own estimates of uncertainty. It could be the Standard Model of physics predicting the mass of the Z boson as 91.1874 ± 0.0021 GeV, or the rather wider uncertainty ranges of economic predictions.
In many cases, though, the uncertainties in or about the model dwarf the estimated uncertainty in the model itself - especially for low probability events. This is a problem, because people working with models often try to use the in-model uncertainty and adjust it to get an estimate of the true uncertainty. They often realise the model is unreliable, but don't have a better one, and they have a measure of uncertainty already, so surely doubling and tripling this should do the trick? Surely...
The following three cases are going to be my go-to examples for showing what a mistake this can be; they cover three situations: extreme error, being in the domain of a hard science, and extreme negative impact.
= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
= f037147d6e6c911a85753b9abdedda8d){kind=logo}