Jump to ratings and reviews
Rate this book

Anthropic Bias: Observation Selection Effects in Science and Philosophy

Rate this book
Anthropic Bias explores how to reason when you suspect that your evidence is biased by "observation selection effects"--that is, evidence that has been filtered by the precondition that there be some suitably positioned observer to "have" the evidence. This conundrum--sometimes alluded to as "the anthropic principle," "self-locating belief," or "indexical information"--turns out to be a surprisingly perplexing and intellectually stimulating challenge, one abounding with important implications for many areas in science and philosophy. There are the philosophical thought experiments and the Doomsday Argument; Sleeping Beauty; the Presumptuous Philosopher; Adam & Eve; the Absent-Minded Driver; the Shooting Room. And there are the applications in contemporary cosmology ("How many universes are there?", "Why does the universe appear fine-tuned for life?"); evolutionary theory ("How improbable was the evolution of intelligent life on our planet?"); the problem of time's arrow ("Can it be given a thermodynamic explanation?"); quantum physics ("How can the many-worlds theory be tested?"); game-theory problems with imperfect recall ("How to model them?"); even traffic analysis ("Why is the 'next lane' faster?"). Anthropic Bias argues that the same principles are at work across all these domains. And it offers a a mathematically explicit theory of observation selection effects that attempts to meet scientific needs while steering clear of philosophical paradox.

240 pages, Hardcover

First published January 1, 2002

Loading interface...
Loading interface...

About the author

Nick Bostrom

25 books1,446 followers
Nick Bostrom is Professor at Oxford University, where he is the founding Director of the Future of Humanity Institute. He also directs the Strategic Artificial Intelligence Research Center. He is the author of some 200 publications, including Anthropic Bias (Routledge, 2002), Global Catastrophic Risks (ed., OUP, 2008), Human Enhancement (ed., OUP, 2009), and Superintelligence: Paths, Dangers, Strategies (OUP, 2014), a New York Times bestseller.

Bostrom holds bachelor degrees in artificial intelligence, philosophy, mathematics and logic followed by master’s degrees in philosophy, physics and computational neuroscience. In 2000, he was awarded a PhD in Philosophy from the London School of Economics.
He is recipient of a Eugene R. Gannon Award (one person selected annually worldwide from the fields of philosophy, mathematics, the arts and other humanities, and the natural sciences). He has been listed on Foreign Policy's Top 100 Global Thinkers list twice; and he was included on Prospect magazine's World Thinkers list, the youngest person in the top 15 from all fields and the highest-ranked analytic philosopher. His writings have been translated into 24 languages. There have been more than 100 translations and reprints of his works. During his time in London, Bostrom also did some turns on London’s stand-up comedy circuit.

Nick is best known for his work on existential risk, the anthropic principle, human enhancement ethics, the simulation argument, artificial intelligence risks, the reversal test, and practical implications of consequentialism. The bestseller Superintelligence, and FHI’s work on AI, has changed the global conversation on the future of machine intelligence, helping to stimulate the emergence of a new field of technical research on scalable AI control.

More: https://nickbostrom.com

Ratings & Reviews

What do you think?
Rate this book

Friends & Following

Create a free account to discover what your friends think of this book!

Community Reviews

5 stars
35 (34%)
4 stars
39 (38%)
3 stars
19 (18%)
2 stars
5 (4%)
1 star
4 (3%)
Displaying 1 - 17 of 17 reviews
Profile Image for Nick Wellings.
77 reviews80 followers
September 1, 2013
Nick Bostrom's ' Anthropic Bias' is a clear sighted and rigorous look at an inherent bias in our reasoning about the universe around us.

This 'anthropic bias' loosely states that because we are observers in a well ordered cosmos, our viewpoint is privileged and automatically skewed towards confirmation of that unconscious bias. So, as the gist of the argument traditionally has it: we are in a universe conducive to life and are able to know this because if the universe wasn't hospitable we'd not be here to observe. Although this is tautologically true, it contains definite bias and also natural language inimicable to saying anything more useful about our world. As the book blurb says too, this bias is not just limited to cosmological questions (although the literature surrounding that is large and varied.)

Not often does a work do such a good job as this of clearing away the underbrush from around previous attempts to state and explore its problem and import .

Chapters 1 and 2 give background to our problem. What is meant by the anthropic principle? By fine tuning? Once we know our for we can work to 'defeat' it (or indeed bargain or reconcile with it!)

Bostrom examines this 'hodgepodge' of various anthropic statements in chapter 2-3, and finds them wanting.

At the end of chapter 3 a solution emerges: Bostrom's lancet is what he calls the Self Sampling Assumption, which becomes a powerful heuristic.

Chapter 4 moves into a discussion on this more useful (and mathematically rigorourised) Self Sampling Assumption, which in turn is later nuanced (after about 6 more chapters!) until we have an 'Observer Equation'. This is an equation which uses Bayesian reasoning to allow for any 'observation selection' bias in our reasoning. In other words I think, it tries to allow for our contingency of experience - our observations - by acknowledging it and working around it. It tries to do away with bias.

To read as Bostrom defines and almost axiomises away this bias was for me was pretty monumental and ingenious. On our journey he makes extensive use of Bayes' theorem, statistics, cute gedanken and probability and (some) game theory . One gets to see an incisive mind at work in quest for honing down then rebuilding his theory til it seems best fit.

Thus defined, Bostrom says the Observer Equation can be called upon to do explanatory work. No longer do we have to reason under generalised 'anthropic principles'. We have OE to help.

By the close of the book we have therefore been given a framework to investigate and understand a number of issues: eg, why we find ourselves in a universe conducive to life, how probable life and a 'multiverse' is and why traffic really might move faster in the other lane and how we might reason more ably by factoring in the time and place in which we do so.

* * *

I thoroughly enjoyed 'Anthropic Bias'. I learned about Boltzmann brains and freak observers, statistically probable chances that given an infinite universe a fully fledged brain or observer might pop out of a black hole! Also there's a full chapter devoted to the Doomsday Argument (http://en.m.wikipedia.org/wiki/Doomsd...) which was pretty mind boggling for me. (Let's just say that its slightly more advanced than traditionally Malthusian arguments! ) It seems this has become his chief concern since 'Anthropic Bias' was published, as he now lectures and publishes about 'existential risk' in his position as Director of Oxford University's Institute of Future Humanity.

But leaving such soteriological fears behind and to return to the book at hand: If nothing else, it seems that Bostrom paves the way for a pavilion of further exploration to be erected upon the groundwork of his thought. Indeed, his explicit wish in the closing paragraph is the hope 'that others will see more clearly than I have and will be able to advance further into this fascinating land of thought'. It is a sentiment I share, and I wonder how his humble wish has fared since publication of this deeply thought provoking book.

* * *
Interested readers may find the entire text of 'Anthropic Bias' for free in HTML or PDF here:

http://www.anthropic-principle.com/?q...

(well worth a glance!)
Profile Image for Paul Adkin.
Author 10 books22 followers
March 8, 2016
I have now read two books by Nick Bostrom: this one “Anthropic Bias” and “Superintelligence”. As my reaction to these works is quite similar then take these comments to apply to both.
I read these books because of the importance of the subject matter and together they deal with some of the biggest questions facing humanity today. Nevertheless, the content of these books is hardly satisfying for several reasons.
A. They are not well written. The arguments are not well expressed. He would do well to study a course on creating allegories -- the thought experiments that Bostrom loves to give us tend to obscure rather clarify things.
B. Bostrom wrestles with tautologies and he gets wrapped up in those tautologies. Most of what he says runs around and around itself, stating what seems obvious without drawing any particularly interesting conclusions about them.
Bostrom is a leader of the Transhumanist movement, which in itself makes these books tempting morsels to dip into. However, I have the sense that he himself is a transhumanist creation, or perhaps he has extra-terrestrial DNA. He certainly seems out of touch with authentic human experiences, and because of that his ideas are cold, disinterested and pedantic. In order to make the task of reading these books more manageable, I kept the image of Sheldon Cooper, from The Big Bang Theory TV series, sprouting the narrative. Yes, both books are something that Sheldon Cooper could have written.
As far as Transhumanism goes, Bostrom, who is hardly human anymore misses some vital points:
A) The human species exists in a state of fragmentations (e.g. cultural, social, economic and racial) and it is in an internecine process of competition. This process, combined with the effects of rampant consumerism, make it very likely that the species will go extinct before reaching the "posthuman" stage; (B) any posthuman civilisation created from the fragmented human species we have at the moment would, despite its revolutionary transformation of the species, intensify the effects of that fragmentation; and from A and B; (C) the only way to combat the dystopian effects of a transhumanist leap would be to create an authentically, human-interested humanity first.
Profile Image for Paige McLoughlin.
231 reviews73 followers
March 1, 2021
I have been reading Bostrom first on the internet in the early aughts and picked up this academic book soon after it was published. It touches on some of my intellectual kinks, Multiverse (or very large universe), and statistical observations in a (near) infinite ensemble and philosophical guidelines for drawing statistical conclusions from such sets of data. Bostrom proposes to uses as his basis the Self-sampling assumption and later suggests the Strong Self-Sampling assumption (SSA and SSSA respectively) as opposed to its challenger the Self-Indication assumption (SIA). There are subtle differences between the two positions but in some cases lead to different statistical outcomes when applied to various data. In a nutshell, SSA argues we should reason that we are a typical set of observers from the "actual" set of observers we see. While the SIA says we should draw our analysis from the set of all "possible" observers. In some cases, the SIA comes up with more optimistic conclusions about the number of observers but faces certain paradoxes like the presumptuous philosopher the SSA and SSSA have the doomsday argument paradox. Bostrom explores this and the sleeping beauty problem which gives two different probable outcomes SSA 1/2 and SIA 1/3 which demonstrates how the assumptions change the outcome. A google search will explain them. I will leave links. Basically, the methods can be used in cosmology especially since we can't see the multiverse (if it is indeed real) we can base our assumptions from data of being observers in our patch of the universe. It can also be applied in more earthly matters in something like traffic jams and other areas. Worth a look if that is your kink.

SSA and SSSA
https://en.wikipedia.org/wiki/Anthrop...

Sleeping Beauty problem
https://en.wikipedia.org/wiki/Sleepin...

Anthropic Reasoning
https://en.wikipedia.org/wiki/Anthrop...

Doomsday Argument
https://en.wikipedia.org/wiki/Doomsda...

Presumptuous Philosopher
https://www.scottaaronson.com/democri...

Arcane topic but may have consequences for you dear reader.
https://www.lesswrong.com/posts/zh2vW...

and the world
https://www.google.com/search?q=why+e...
Profile Image for Hom Sack.
507 reviews11 followers
April 30, 2015
This book is one of the suggested readings in preparation for tonight's Philosophy Cafe meeting to discuss the question: "What is the role of anthropocentric reasoning in science and philosophy?" As such, it fails as far as philosophy is concern and I would say for science as well despite that the book is mostly about the latter. Even the author offers one of the criticism of the Anthropic Principle as "mere tautologies (and therefore incapable of doing any interesting explanatory work whatever) and for being widely speculative (and lacking any empirical support)".
Profile Image for Reed Caron.
6 reviews1 follower
January 8, 2018
An absolute tour de force in understanding the self-sampling function. Only 200 pages but took me several months (and an incredible amount of caffeine) to digest. For people who are intrigued when Elon Musk starts spouting off about living in a simulation, this book provides the philosophical foundation of the argument. A strong grasp of bayesian probability helps a lot (which I did not have at first).
30 reviews10 followers
November 3, 2016
A fascinating (and compelling) bit of philosophy
Profile Image for Paige McLoughlin.
597 reviews32 followers
May 10, 2021
I have been reading Bostrom first on the internet in the early aughts and picked up this academic book soon after it was published. It touches on some of my intellectual kinks, Multiverse (or very large universe), and statistical observations in a (near) infinite ensemble and philosophical guidelines for drawing statistical conclusions from such sets of data. Bostrom proposes to uses as his basis the Self-sampling assumption and later suggests the Strong Self-Sampling assumption (SSA and SSSA respectively) as opposed to its challenger the Self-Indication assumption (SIA). There are subtle differences between the two positions but in some cases lead to different statistical outcomes when applied to various data. In a nutshell, SSA argues we should reason that we are a typical set of observers from the "actual" set of observers we see. While the SIA says we should draw our analysis from the set of all "possible" observers. In some cases, the SIA comes up with more optimistic conclusions about the number of observers but faces certain paradoxes like the presumptuous philosopher the SSA and SSSA have the doomsday argument paradox. Bostrom explores this and the sleeping beauty problem which gives two different probable outcomes SSA 1/2 and SIA 1/3 which demonstrates how the assumptions change the outcome. A google search will explain them. I will leave links. Basically, the methods can be used in cosmology especially since we can't see the multiverse (if it is indeed real) we can base our assumptions from data of being observers in our patch of the universe. It can also be applied in more earthly matters in something like traffic jams and other areas. Worth a look if that is your kink. SSA and SSSA
I have been reading Bostrom first on the internet in the early aughts and picked up this academic book soon after it was published. It touches on some of my intellectual kinks, Multiverse (or very large universe), and statistical observations in a (near) infinite ensemble and philosophical guidelines for drawing statistical conclusions from such sets of data. Bostrom proposes to uses as his basis the Self-sampling assumption and later suggests the Strong Self-Sampling assumption (SSA and SSSA respectively) as opposed to its challenger the Self-Indication assumption (SIA). There are subtle differences between the two positions but in some cases lead to different statistical outcomes when applied to various data. In a nutshell, SSA argues we should reason that we are a typical set of observers from the "actual" set of observers we see. While the SIA says we should draw our analysis from the set of all "possible" observers. In some cases, the SIA comes up with more optimistic conclusions about the number of observers but faces certain paradoxes like the presumptuous philosopher the SSA and SSSA have the doomsday argument paradox. Bostrom explores this and the sleeping beauty problem which gives two different probable outcomes SSA 1/2 and SIA 1/3 which demonstrates how the assumptions change the outcome. A google search will explain them. I will leave links. Basically, the methods can be used in cosmology especially since we can't see the multiverse (if it is indeed real) we can base our assumptions from data of being observers in our patch of the universe. It can also be applied in more earthly matters in something like traffic jams and other areas. Worth a look if that is your kink. SSA and SSSA

https://en.wikipedia.org/wiki/Anthrop... Sleeping Beauty problem

https://en.wikipedia.org/wiki/Sleepin... Anthropic Reasoning

https://en.wikipedia.org/wiki/Anthrop... Doomsday Argument

https://en.wikipedia.org/wiki/Doomsda... Presumptuous Philosopher

https://www.scottaaronson.com/democri... Arcane topic but may have consequences for you dear reader.

https://www.lesswrong.com/posts/zh2vW... and the world

https://www.theatlantic.com/science/a...

https://youtu.be/EffcY6tnbcA
Profile Image for Ives Parr.
13 reviews
January 13, 2024
Observation Selection Effects is an important and under-considered topic. This early work built off of observations from cosmologists and tried to defend a certain perspective about the right assumption for thinking about observation selection effects.

Bostrom defends the SSA (self-sampling assumption), before he kicks away the ladder in favor of the SSSA (strong self-sampling assumption). The SSA says: "All other things equal, an observer should reason as if they are randomly selected from the set of all actually existent observers (past, present and future) in their reference class." The SSSA replaces observers with observation moments, and in doing so, appears to resolve some paradoxes. The SSA carries with it a difficult problem of the appropriate reference class (who is my reference class? me? people like me? Are aliens included?), which Bostrom thinks could be defined in many different ways reasonably. So...it seems challenging to actually reach a strong conclusion about things.

The SSA and SSSA constrasts the SIA which means you should reason as if you're drawn from possible moments. Bostrom's knock-down argument is that this entails the presumptious philosopher;

"The Presumptuous Philosopher It is the year 2100 and physicists have narrowed down the search for a theory of everything to only two remaining plausible candidate theories, T1 and T2 (using considerations from super-duper symmetry). According to T1 the world is very, very big but finite and there are a total of a trillion trillion observers in the cosmos. According to T2, the world is very, very, very big but finite and there are a trillion trillion trillion observers. The super-duper symmetry considerations are indifferent between these two theories. Physicists are preparing a simple experiment that will falsify one of the theories. Enter the presumptuous philosopher: “Hey guys, it is completely unnecessary for you to do the experiment, because I can already show to you that T2 is about a trillion times more likely to be true than T1! (whereupon the philosopher runs the Incubator thought experiment and explains Model 3).”"

In a four part series, Joe Carlsmith actually defends SIA: https://forum.effectivealtruism.org/p...

I didn't feel like this was adequate. There are several other arguments against SSA (Adam and Eve, Quantum Joe, UN++), and Bostrom refuted them but...I don't understand it very well.

Actually, this is one of my main issues. This book is very challenging to read. It is hard to give the text the time it probably deserves. The topic is fascinating but it felt pretty boring at times because it was so challenging. That is partially my fault, but still. Bostrom wrote this in a way that made it just difficult to follow. It's dense and he's verbose with complex sentence structures. The topic is already very difficult, so it's just especially difficult.

Perhaps I should've read it more slowly, but I felt ready to finish. But I feel like I walk away without a solid grasp of observer selection effects, nor a particular belief that SSSA is right, nor how to apply it very well if I wanted to, nor how exactly to pick my reference class. I feel somewhat sympathetic to SIA still, and I think I'll read Carlsmith again. Overall, I'm just confused. I wanted to like this but these aspects made it challenging. Perhaps I am partially to blame, but this was very tough. It may be because it was adapted from his dissertation, I believe.
Profile Image for Hamid.
3 reviews
February 3, 2020
*Very interesting topic.
*Reader needs basic knowledge of probability theory as well as logic.
*Some parts contain very lengthy arguments to prove propositions that look obvious from the beginning. On the other hand I couldn't follow some other arguments. Perhaps my lack of training in probability and logic is to blame for these.
Profile Image for Matthew Adelstein.
45 reviews7 followers
November 17, 2023
Very good yet totally wrong. Bostrom rejects SIA too quickly based on the presumptuous philosopher, before biting lots of bullets associated with SSA. But a scenario like the presumptuous philosopher is everyone's problem. Bostrom's view implies the equally presumptuous archeolgist result.
https://benthams.substack.com/p/sia-i...
Profile Image for Richie.
36 reviews
February 20, 2023
Interesting and opinionated primer on the topic.

Also a rare demonstration of Bostrom being confused. See SIA >> SSA by Joe Carlsmith for less confusion (a blog post series that is heavily influenced by this work).
Profile Image for Tony.
280 reviews1 follower
June 9, 2016
Fascinating read on a fascinating (albeit under-studied) branch of philosophical thought. Much of it went far over my head, but I question whether that's entirely my fault. Bostrom could have invested more effort elucidating the meanings of his equations if he were preparing to pitch this to a larger audience (as he did with Superintelligence). His folksy, whimsical disposition and writing style would lend themselves well to a less technical handling of the topic, and I would be delighted to read one should he ever find the time to revisit the topic.
356 reviews29 followers
October 19, 2021
Fascinating and very clearly written book about how to reason in the presence of observation selection effects. I got through 90% of this book, but never finished it since it takes quite a bit of thought to appreciate fully. I keep meaning to pick it back up.

I don’t recommend the Kindle book, as it had major errors that hurt my ability to understand the material (e.g. mistakes in equations). I bought the Kindle book first, but then returned it and bought a physical copy.
Displaying 1 - 17 of 17 reviews

Can't find what you're looking for?

Get help and learn more about the design.