follow CCP

Recent blog entries
popular papers

What Is the "Science of Science Communication"?

Climate-Science Communication and the Measurement Problem

Ideology, Motivated Cognition, and Cognitive Reflection: An Experimental Study

'Ideology' or 'Situation Sense'? An Experimental Investigation of Motivated Reasoning and Professional Judgment

A Risky Science Communication Environment for Vaccines

Motivated Numeracy and Enlightened Self-Government

Ideology, Motivated Cognition, and Cognitive Reflection: An Experimental Study

Making Climate Science Communication Evidence-based—All the Way Down 

Neutral Principles, Motivated Cognition, and Some Problems for Constitutional Law 

Cultural Cognition of Scientific Consensus
 

The Tragedy of the Risk-Perception Commons: Science Literacy and Climate Change

"They Saw a Protest": Cognitive Illiberalism and the Speech-Conduct Distinction 

Geoengineering and the Science Communication Environment: a Cross-Cultural Experiment

Fixing the Communications Failure

Why We Are Poles Apart on Climate Change

The Cognitively Illiberal State 

Who Fears the HPV Vaccine, Who Doesn't, and Why? An Experimental Study

Cultural Cognition of the Risks and Benefits of Nanotechnology

Whose Eyes Are You Going to Believe? An Empirical Examination of Scott v. Harris

Cultural Cognition and Public Policy

Culture, Cognition, and Consent: Who Perceives What, and Why, in "Acquaintance Rape" Cases

Culture and Identity-Protective Cognition: Explaining the White Male Effect

Fear of Democracy: A Cultural Evaluation of Sunstein on Risk

Cultural Cognition as a Conception of the Cultural Theory of Risk

« *Now* where am I? Oklahoma City! | Main | Hurry up & get your copy of "Expressive rationality of inaccurate perceptions" before sold out! »
Tuesday
Apr042017

Science of Science Communication seminar: Session 8 reading list (climate change 2)

Feel free to comment if you are playing along at home . . . .

PrintView Printer Friendly Version

EmailEmail Article to Friend

Reader Comments (14)

On the Cook, Lewandowsky article on Rational Irrationality: Bayesian networks offer the advantage of conputational ease but are sadly unable to model underlying data in higher dimensions. Tensor calculus has to be used instead. (For those unfamiliar, tensors are multidimensional vectors, i.e.provide a generalization of a scalar (tensor rank 0), a vector (rank 1) and a NxN matrix (rank 2) and are used in fluid dynamics). Nobody knew that better than their inventor, Pearl, who was careful to describe what they could accomplish:

http://amturing.acm.org/award_winners/pearl_2658896.cfm
...Pearl believed that sound probabilistic analysis of a problem would give intuitively correct results, even in those cases where rule-based systems behaved incorrectly. One such case had to do with the ability to reason both causally (from cause to effect) and diagnostically (from effect to cause). “If you used diagnostic rules, you could not do prediction, and if you used predictive rules you could not reason diagnostically, and if you used both, you ran into positive-feedback instabilities, something we never encountered in probability theory.” Another case concerned the “explaining-away” phenomenon, whereby the degree of belief in any cause of a given effect is increased when the effect is observed, but then decreases when some other cause is found to be responsible for the observed effect. Rule-based systems could not exhibit the explaining-away phenomenon, whereas it happens automatically in probabilistic analysis....

Fluid dynamics are chock-full of positive feedback loops. And modeling "degrees of belief" - never mind their derivatives! - is mathematically equivalent to constructing a fusion reactor: nature's solution, involving a whole lot of mass, isn't accessible to us yet. We did manage to solve uncontrolled fusion, so there is hope, unless of course our solution is deployed on a massive scale to combat global warming by creating a nuclear winter - not beyond the realm of possibility given the prevailing climate change hysteria.

April 9, 2017 | Unregistered CommenterEcoute Sauvage

"Bayesian networks offer the advantage of conputational ease but are sadly unable to model underlying data in higher dimensions. Tensor calculus has to be used instead."

What are you talking about?

What data, why can't Bayesian networks model higher dimensions, and what problem are you claiming needs tensor calculus to solve?

(I'm very familiar with tensor calculus, and I've implemented Bayesian networks, so don't feel the need to explain that. But I've no idea what problem you're trying to apply them to.)

April 9, 2017 | Unregistered CommenterNiV

NiV - I will answer you, and answer you in toto, even though this unfortunately involves length I'm not used to (for good reason, given my literary abilities). I must start by telling you HOW I arrived at this conclusion; that involves several steps any of all of which might be wrong, in addition to the the risk of the conclusion itself being wrong.

Many, many, years ago I had to calculate economic odds of survivability of limited nuclear wars. The information cascade involved first a military person (whose name I never learned, but who was locally known as Gold Braid) would tell a bunch of physicists the presumed targets, then they would all work out magnitudes of explosions, then atmospheric scientists, geologists etc got involved to calculate prevailing winds, fallouts, and the like, then engineers, architects, urban planners etc would calculate damage from blast and so on, then others would calculate casualties, and finally I was handed the results of those simulations and asked to figure out such questions as - if Atlanta, a major rail hub, is lost, can railroads be rerouted to keep connections to the West Coast, or if Miami is lost can Galveston pick up sea traffic, and so on.

The first thing I realized was that the CEP (circular error probable) targeting experts were not really sure of their results - by which I mean not at all sure. Anything targeted anywhere might drop someplace else, especially over polar orbits. And let's not even mention countermeasures, like tinfoil shreds, sent up by the "missile interceptors". Then of course there was additional uncertainty (all of which I hardly need add is multiplicative) involving temperatures, winds, blast effects, and on and on. By the time I had to provide a provisional report I timidly informed the august assembly I could not come up with any kind of functioning economic framework because one often unstated condition of any economic system is trust, and the panic that was sure to have descended in any of those scenarios would be most probably such that no trust in a system of any kind would survive, excluding maybe in businesses specializing in cockroach protein bars. "Who brought in that skirt?" was the muttered but clearly audible comment of Gold Braid - I was wearing jeans at the time but was undeniably the only woman around so I figured that was no compliment. A physicist quickly said the project was too far along to replace me now, so I was dispatched to repeat my calculations using tensors rather than the simplistic if-then Bayesian analysis. Complexity, I was sternly informed, could be tamed.

Now of course however dire the situation panics eventually subside. As long as there are survivors some new path forward will be found. But in social and economic problems proving that a solution exists - as in Nash equilibria, for instance - isn't at all the same thing as computing it. But there's no throwing up your hands and saying this damn problem is intractable; you can always come up with some scenarios where order of sorts and workable economies are restored, and tensors allow for relaxation of constraints more easily than Bayesian networks. Anyway, that's why I prefer the former to the latter - a completely idiosyncratic reason for which I can certainly not provide mathematical proof :)

April 9, 2017 | Unregistered CommenterEcoute Sauvage

"... and finally I was handed the results of those simulations and asked to figure out such questions as ..."

Hmm. Sounds to me like your colleagues might have been playing 'status' games. I'm not sure - you will no doubt be in a better position to judge - but generally where I come from when a young new analyst comes along they start off being closely supervised and taught the ropes, their output is reviewed until we're happy with it, and they only get put up in front of the brass to present it after a few years, when we're confident they can cope. You woul;dn't let anyone present results without reviewing them internally first. And you definitely wouldn't set someone up to fall like that - in front of a senior officer no less! - unless they were especially cocky, or unless you wanted to make a particular point to them about the value of teamwork. If you're the only woman in a department of men, back in the old days when that sort of thing was a novelty, I can just imagine 'the boys' in the department doing that. It can also happen with bad management, if they assume you know what you're doing and you don't ask for help, which is quite common with youngsters trying to prove themselves in a new environment. I don't know. Either way, you have my sympathy.

On the subject in question - Yes, you're quite right that targeting models are largely guesswork with lots of uncertainty, it means that the damage to the infrastructure is essentially random. That's good, because it means the damage is going to be a lot less than if they were able to target it precisely on the most critical bits. But it means that the answer will be a probability distribution. What they're usually interested in is the distribution of risk (= impact x probability); what's the typical value you can expect, and what are the "worst case" consequences. It sounds like you were asked to assess the impact of blowing random holes in the civil logistics infrastructure. The targeting guys then told you the probabilities, so you could get the risk.

Economies adapt, even to war conditions. Sustained panic only occurs when people in a survival situation believe there is not enough of some essential resource, and race to grab it first. It only lasts until supply catches up with demand. London's economy carried on through the Blitz during World War II, and there are generally active economies in war zones around the world. The first thing you would do is study those - see how they worked, how they changed from their pre-conflict precursors, and how quickly. Economies normally evolve logistic networks with precisely enough robustness to cope with the expected level of failures, because robustness generally costs extra. The equilibrium is where the price of extra robustness just equals the cost saved of the failures prevented. Increase the failure rate (by dropping bombs on it), and the optimum point shifts. Looked at on a high level, it's about systems of partial differential equations.

I'm personally not convinced that doing detailed modelling at the level of individual ports and rail junctions is going to be very useful; things at that level change too fast in a real economy. It smacks of the central planning of command economies. The Communists tried that, and it was a disaster. But it might be reasonable for exploring the sorts of things that might happen in various scenarios.

If you fit a function relating transport capacity to price at each edge (one that rockets upwards at the maximum capacity of the edge), plus supply/demand price curves at each node, add constraints relating utilised capacity at each edge to its immediate neighbours, and then you can find the equilibrium by minimising costs. Then look at the sensitivity of the minimum to removing sets of edges/nodes (or equivalently, constraining the minimisation to the subspace where the capacity used on those edges is zero). That will tell you whether you've got the capacity to route round them, and what the increase in prices and consumption will be. Add in the costs of developing more capacity along each edge, and see how things will change in the longer run.

It shouldn't require tensor analysis; although the cost function is certainly multi-dimensional of a high order, it's just a scalar field. So I'm still not sure why you would want to use tensor analysis - or, in this case, Bayesian networks, either. But if you don't have the time or inclination to explain further, don't worry about it. I was just curious.

April 10, 2017 | Unregistered CommenterNiV

Thank you - and since we're guests on Dan's thread, to his topic: a thorough understanding of cesium-137, strontium-90, iodine-131 and related isotopes will ensure perpetual indifference to increasing atmospheric CO2. Doubt that Al Gore knows the difference between any of the above and any of the carbon isotopes, but he's made $100 million investing in taxpayer-funded windmills and solar panels, so probably he knows something I don't - on subsidies, not isotopes.

April 10, 2017 | Unregistered CommenterEcoute Sauvage

Quick PS about those edges: you may be familiar with the work of Andy Lo, who models manias, panics, and crashes in financial markets using functions imported from biology (that old exploding regressor problem). See e.g. second part here (free registration required) for an interview on his new book on adaptive markets: https://ftalphaville.ft.com/2017/03/24/2186452/podcast-adaptive-markets-and-the-lessons-of-an-infamous-call-to-sell/

April 11, 2017 | Unregistered CommenterEcoute Sauvage

PPS Andy Lo (excerpt from intro to his new book) provides a variant of the Cultural Cognition hypothesis:

"....Some debates are merely academic. Th is one isn’t. If you believe that
people are rational and markets are effi cient, this will largely determine
your views on gun control (unnecessary), consumer protection laws (caveat
emptor), welfare programs (too many unintended consequences),
derivatives regulation (let a thousand fl owers bloom), whether you
should invest in passive index funds or hyperactive hedge funds (index
funds only), the causes of fi nancial crises (too much government intervention
in housing and mortgage markets), and how the government
should or shouldn’t respond to them (the primary fi nancial role for government
should be producing and verifying information so that it can
be incorporated into market prices). [....]

[...] Our behavior adapts to new environments— it has to because of
evolution— but it adapts in the short term as well as across evolutionary
time, and it doesn’t always adapt in financially beneficial ways. Financial
behavior that may seem irrational now is really behavior that hasn’t
had sufficient time to adapt to modern contexts. An obvious example
from nature is the great white shark, a near- perfect predator that moves
through the water with fearsome grace and efficiency, thanks to 400
million years of adaptation. But take that shark out of the water and
drop it onto a sandy beach, and its flailing undulations will look silly
and irrational. It’s perfectly adapted to the depths of the ocean, not to
dry land.

Irrational financial behavior is similar to the shark’s distress: human
behavior taken out of its proper evolutionary context. The diff erence between
the irrational investor and the shark on the beach is the shorter
length of time the investor has had to adapt to the financial environment,
and the much faster speed with which that environment is
changing. Economic expansions and contractions are the consequences
of individuals and institutions adapting to changing fi nancial environments,
and bubbles and crashes are the result when the change occurs
too quickly....."

April 11, 2017 | Unregistered CommenterEcoute Sauvage

"Irrational financial behavior is similar to the shark’s distress: human behavior taken out of its proper evolutionary context."

Could be. Or it may just be evolution in action.

Consider those sharks. The food supply varies from year to year, for external reasons. When there's plenty of food, sharks multiply, and increase in number. When the food supply drops the sharks starve, the population collapses, and only the fittest of them survive. Sound familiar?

Even if they knew the crash was coming, it still makes sense for sharks to multiply in the good times. The more descendants they have compared to other sharks, the more chance there is of some surviving. The good times give sharks the slack to explore new niches and new ways of life. The crashes apply the natural selection that made them the supreme survival machines that they are. They are all descendants of the survivors of a long chain of crashes, each crash making the species stronger.

Despite the recent financial crashes, humanity is still more prosperous and powerful today than it has ever been in virtually the whole 200,000 years of its history. The catallaxy humans have evolved, with unparalleled and only a century ago unimaginable capabilities was not, and could not have been consciously designed. It can only evolve, evolution proceeds only by natural selection where the weakest are periodically eliminated, the survivors of each earthquake building ever stronger edifices on the rubble of the old. And we have developed far faster and further than the sharks not because human financial behaviour hasn't had time to adapt, but precisely because it adapts at incredible speed. Culture can evolve far faster and more precisely than our genes - it doesn't even have to do so blindly: we can think about ways to solve our problems. Human financial behaviour, I suggest, is far more finely evolved and adapted to current circumstances than sharks are. That's why we operate 'shark fishing' businesses, and they don't.

Evolution is a wonderful way of creating amazing capabilities, but it needs the practical experience of failure/survival to guide it. An island without predators evolves dodos.

April 11, 2017 | Unregistered CommenterNiV

NiV - not all those who see the dazzling beauty of your partial differential equations necessarily share your apparent confidence that the equations have acceptable policy solutions. In brief: I know no biology, but I know probability functions and I'm rooting for the sharks. If it were up to me I would zero in on those shark-finning, plankton-sucking Chinese trawlers and blow them out of the water. Bioconcentation is on my mind, CO2 is not.

And what's the point of sending the newest antibiotics to India - only to compound the problem? Southern Africa (several countries) has anti-retroviral drugs, populations that can read, and even so has half a million people dying annually of AIDS - might stopping the drugs bring down the infection rate? And that's not to mention the grotesque overpopulation rates which only pick up whenever there is a famine, since that's when we send food.

Finally, a word on my former colleagues: everyone on that team was of either European or Japanese ancestry, and I've not the least doubt they would have treated any mathematical modeler in the exact same way they treated me - so I may have elicited your sympathy on the basis of elliptical expression. Culture evolves, but I see no evidence that the evolution proceeds uniformly across all of "humanity", so the next question is providing a definition of "we", in case we do have to fight this out. Blood matters. If and when intelligent bots - unaffected by cultural constraints - take over, they are sure to provide rational solutions to the equations, so the race as I see it now it to ensure these are "our" bots. Not politically correct, but true.

April 12, 2017 | Unregistered CommenterEcoute Sauvage

"NiV - not all those who see the dazzling beauty of your partial differential equations necessarily share your apparent confidence that the equations have acceptable policy solutions."

"Acceptable" to who? And what makes you think the differential equations care?

"If it were up to me I would zero in on those shark-finning, plankton-sucking Chinese trawlers and blow them out of the water."

That's an evolutionary strategy that everyone can play at. If it was up to certain other people, they would zero in on the Greenpeace boats and blow them out of the water, too! The reason they don't is the same reason you don't.

"And what's the point of sending the newest antibiotics to India - only to compound the problem?"

What problem? The problem as I see it is that there are a lot of people dying unnecessarily of infections in India. But since it's obvious that sending antibiotics reduces that problem, I assume you're talking about a different one?

"And that's not to mention the grotesque overpopulation rates"

There is no overpopulation problem. That was a scare story generated back in the late 1960s by the socialists of the Club of Rome and Paul Ehrlich and so on. It was one of the same series of 'global crises' as climate change, used to justify the imposition of an authoritarian global government by progressive environmentalists. They predicted back then that overpopulation would have ended civilisation before the year 2000, unless they took authoritarian control of society to impose population control. When that didn't happen, they just invented another, then another, and now that climate change has failed to bring about the governmental changes they wanted and is fading politically, they're probably starting to think about what's next.

India and Africa are not overpopulated, They're just poor. And the modern economic revolution is in the process of fixing that.

"Finally, a word on my former colleagues: everyone on that team was of either European or Japanese ancestry, ..."

Err... what?!! What's that got to do with anything?

"If and when intelligent bots - unaffected by cultural constraints - take over, they are sure to provide rational solutions to the equations, so the race as I see it now it to ensure these are "our" bots."

Intelligent bots will undoubtedly be affected by cultural constraints, as culture is a perfectly rational solution to the equations. That's why humans have succeeded the way they have.

And of course the concept of them being culturally "our" bots implies that.

April 14, 2017 | Unregistered CommenterNiV

>>> "Err... what?!! What's that got to do with anything?" <<<<<<<<<<<

"Overpopulated" sounds better than the accurate "filthy", but you're right it is statistically incorrect. Here is a map:
https://ehp.niehs.nih.gov/wp-content/uploads/2014/11/ehp.122-A298.g003.jpg
source: https://ehp.niehs.nih.gov/122-a298/
Countries sharing half an island - Haiti and Papua New Guinea - are illustrative of the concept.

As to your "..err...." cultural concepts of "clean" and "trustworthy" overlap with the map. There are variations by location - I remember for instance the old South Africa where Japanese had to stand at customs until they were officially classified as "honorary whites", and I also remember myself working in Saudi Arabia where I had to hover by the dinner table until my host pronounced some magical incantation turning me into a man for purposes of sitting down - their own women were not subject to that rule so never appeared at dinners.

Finally I'm less certain than you on the extent to which cultural concepts can be transferred to bots - just look at the algorithm of United, booting out as necessary those whose ticket values (adjusted for frequent miles) were lowest to the airline without adjustment for "political correctness"; it will probably now be amended to include something along the lines of "skip if the designated person is a minority and move to the next lowest value".

April 14, 2017 | Unregistered CommenterEcoute Sauvage

Good article on how to teach PC to bots:

http://science.sciencemag.org/content/356/6334/183.full

"..Our findings suggest that if we build an intelligent system that learns enough about the properties of language to be able to understand and produce it, in the process it will also acquire historical cultural associations, some of which can be objectionable. Already, popular online translation systems incorporate some of the biases we study [..]. Further concerns may arise as AI is given agency in our society.[…]. We recommend addressing this through the explicit characterization of acceptable behavior. One such approach is seen in the nascent field of fairness in machine learning, which specifies and enforces mathematical formulations of nondiscrimination in decision-making. Another approach can be found in modular AI architectures, such as cognitive systems, in which implicit learning of statistical regularities can be compartmentalized and augmented with explicit instruction of rules of appropriate conduct..."

April 15, 2017 | Unregistered CommenterEcoute Sauvage

"Here is a map"

A map of what? There's no label telling me what variable is being plotted.

"I remember for instance the old South Africa where Japanese had to stand at customs until they were officially classified as "honorary whites", and I also remember myself working in Saudi Arabia where I had to hover by the dinner table until my host pronounced some magical incantation turning me into a man for purposes of sitting down - their own women were not subject to that rule so never appeared at dinners."

*Every* culture has its own rules and taboos, but we only notice those from other cultures as being notable or strange.

"Finally I'm less certain than you on the extent to which cultural concepts can be transferred to bots"

"Culture" just means information and knowledge transferred from one individual in a social group to another by communication. "Moral rules" are the constraints individuals follow living together in societies to reduce and mediate conflict, and as they are communicated/negotiated between individuals are part of the culture. Moral rules work a lot like languages. The details of what particular rules are used are arbitrary, like the specific sequence of sounds any language uses to represent a concept. But the methods by which the group jointly negotiates them, and the way they change gradually over time are common across all humanity.

Bots would most certainly have a culture - to not transfer information between individuals is just too big an advantage to give up. And they would also swiftly evolve moral rules if they are to interact frequently in a communicative community where there is a potential for conflict. If two bits of software try to use the same resource at the same time, there is a protocol to decide which gets access and which waits. The rules they might use for deciding priority could well be arbitrary. They might for example give priority to whichever has the higher network address serial number. And likewise, they would probably develop rules about not shutting off the power to other bots, on the understanding that nobody is allowed to shut theirs off.

Bot culture would not necessarily be the same as any particular human culture, although if they interact a lot with humans we would no doubt evolve some sort of joint culture - at least about those aspects of the world they share with us. But any sort of repeated competition for resources constitutes a game in which the most successful strategies involve constraining one's own behaviour to avoid conflict - wars are a lot more expensive than the costs of politeness. Bots will likely speak different languages to humans - different vocabulary and syntax - but they will undoubtedly have language. In the same way, they will undoubtedly have culturally-defined moral rules - although they might have arbitrary elements and be unlike anything a human would come up with.

"Our findings suggest that if we build an intelligent system that learns enough about the properties of language to be able to understand and produce it, in the process it will also acquire historical cultural associations, some of which can be objectionable."

That's part of the conflict and negotiation between two human cultures. One culture allows/expects certain beliefs to be expressed while another (more recent) one forbids them. The newer culture is practicing 'cultural imperialism' in trying to eliminate the older one, and recruiting the bots people use to help them in their effort. I think you can predict how humans of the older culture are likely to respond... :-)

But yes, bots that interact with humans will need to know about human cultural restrictions, and will tailor their output depending on the culture of the human they're talking to. Like Google searches in China filter out material the Chinese government considers objectionable. They call it the "filter bubble", and it's already a thing.

April 15, 2017 | Unregistered CommenterNiV

Graph unfortunately does not explain axes, but posted source does - percentages refer to stunted children, color adjusted accordingly.

April 15, 2017 | Unregistered CommenterEcoute Sauvage

PostPost a New Comment

Enter your information below to add a new comment.

My response is on my own website »
Author Email (optional):
Author URL (optional):
Post:
 
Some HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>