Monthly Archives: August 2009

Singularity PR Dupes?

I’m to speak at a $500-per-attendee Singularity Summit in New York in early October. “Singularity” is associated with many claims, but most are controversial. They say:

The Singularity represents an “event horizon” in the predictability of human technological development past which present models of the future may cease to give reliable answers, following the creation of strong AI or the enhancement of human intelligence.

(They also list related definitions.)  An awful lot of folks, perhaps even most, consider these ideas silly and/or crazy.  They also say:

The Singularity Summit is the world’s leading dialog on the Singularity, bringing together scientists, technologists, skeptics, and enthusiasts alike.

But looking over their program, I noticed that while many speakers are distinguished, those folks won’t directly address the controversial claims; they will instead talk on their usual topics.  A few will talk on how they are trying to design general machine intelligence, but only Kurzweil, Yudkowsky, and Salamon will speak directly to the main controversial issues, and they will take “pro” sides.  As far as I can tell, only I will take a somewhat con side (explained below), but only on some claims, and only tangentially to my brief talk.

It seems as if the organizers plan to gain credibility for their claims by having credible people speak at an event where some speakers make such claims, even if those credible speakers do not address those claims.  Such organizers even expect to gain credit for promoting a “dialog.”  How common is this strategy?  How effective?  How fair?  How much does agreeing to speak at such an event make it seem that you agree with its theme claims?   How many of the summit’s distinguished speakers do agree with those claims?

Those who followed my debate here at OB with Eliezer Yudkowsky last year (e.g., here, here) will be familiar with all this, but let me review.  Here are some of the more controversial claims associated with “singularity”:

  1. Progress is accelerating rapidly across a wide range of techs.
  2. Smarter than human machines are likely in a few decades.
  3. Such machines will induce dramatic and rapid social change.
  4. This change is impossible to foresee; don’t even try.
  5. A single localized super-smart machine or a cabal of them is likely to take over everything.
  6. That cabal’s values determine everything, but via self-modification could become anything.
  7. So everything depends on finding a way to give such machines stable values we like.
  8. No one should try to make super smart machines before knowing how to do this.

I disagree with many but hardly all of these:

  • No, overall neither econ nor tech progress is much accelerating lately.
  • Yes, smarter than human machines are likely in roughly a half century to a century or two, but most likely because whole brain emulations will first induce an important era of near human level machines.
  • Yes, this em era will bring huge rapid social changes, but we can and should use social science to foresee these changes.
  • Yes, this em era may well end via super smart machines, and yes it is hard to constrain the values of the distant future, but a single local machine or cabal taking over everything and then immediately evolving out of value control seems extremely unlikely.  It runs counter to most of our econ and tech innovation experience, and the theories we use to make sense of that experience.
  • Yes, a few powerful-enough mind-design insights could conceivably allow one brash team to leap this far ahead of the world, and some folks should think about how to give machines stable values we like, but most futurists should focus on more likely scenarios.
GD Star Rating
Tagged as: , ,

Blogger Pots Call Senator Kettles Black

Matt Yglesias:

Senators who genuinely do believe that carbon dioxide emissions are contributing to a global climate crisis seem to think nothing of nevertheless taking actions that endanger the welfare of billions of people on the grounds that acting otherwise would be politically problematic in their state. In other words, they don’t want to do the right thing because their self-interest points them toward doing something bad. …

It’s not even clear that voting “the wrong way” poses particularly serious threats to one’s re-election. But even if it did, one might assume that people who bother to dedicating their lives to securing vast political power did so because they actually wanted to accomplish something and get in the history books.

Tyler Cowen:

It’s stunning how true this is.  …  Many people — especially those who become politicians — really do want fame and power and it is amazing what they will talk themselves into to get there and to stay there.  They don’t even want fame in the sense of being recognized, in the longer run, for having done the right thing.  They want more personal influence and power now.

Are Matt and Tyler really surprised here?  Senators are roughly our top hundred politicians, after all, and such critiques should most apply to our most elite politicians, who have been most selected for putting winning above other considerations.

A similar critique applies to our top columnists and bloggers. The topics on which they write show a strong correlation with topics that their readers (or patrons) are likely to find engaging, because those topics are “in the news,” or popular among other pundits, or perennially popular like sex or scandal, and so on.  Meanwhile, a vast cloud of good ideas remain neglected because they are less engaging, or even offensive, to readers (and patrons).

Yes many pundits do sometimes sacrifice their popularity to promote ideas deserving more attention, but the tendency is weak and weakest among the most elite pundits.  Pundits might similarly gain fame in the history books as having been among the first to give serious attention to some neglected idea, but they also show little interest in this prospect.

GD Star Rating
Tagged as: ,

Downturns Are Not Existential Risks

Epidemics, wars, and quakes are distributed with long tails, so that most of the expected harm from such events are in the largest possible events.  Most who are expected to die in epidemics die in events that kill much of the population.  These long tail disasters plausibly embody existential risk – a risk to the existence of civilization and humanity.

A new paper on economic downturns suggests that such events do not have long tails, and so are not existential concerns:

In the rare-disasters setting, a key determinant of the equity premium is the size distribution of macroeconomic disasters, gauged by proportionate declines in per capita consumption or GDP. The long-term national-accounts data for up to 36 countries provide a large sample of disaster events of magnitude 10% or more. For this sample, a power-law density provides a good fit to the distribution of the ratio of normal to disaster consumption or GDP. The key parameter of the size distribution is the upper-tail exponent, alpha, estimated to be near 5, with a 95% confidence interval between 3-1/2 and 7. …

We work with the transformed disaster size z ≡ 1/(1-b), which is the ratio of normal to disaster consumption or GDP. … We start with a familiar (single) power law, which specifies the [probability] density function as f(z) = Az.

For a long tail, alpha needs to be two or less.

GD Star Rating
Tagged as: ,

Who Are US Policy Elites?

It is ten hour drive each way from my home in Virginia to the GENCON gaming convention I attended last week.  Those hours driving were more fun than the convention, because I spent them talking to the smart polymath Bill Dickens.  I’d love more long car trips like that.

Here are two related things Bill told me:

  1. Bill learned to fly planes from the CEO of a well-known successful tech company, a man who was forced out by VCs when his company went public.  This wasn’t a tech person replaced by a management person; this CEO had been doing a fine job managing employees, suppliers, customers, etc.  The VCs told him he was pushed out to make room for someone “better known to Wall Street.”  It seems they wanted a CEO with more social connections to the investment bankers that the company needed to impress.  Such bankers would induce less investment in a company managed by someone who wasn’t in their social circle.
  2. For the last few decades academic macroeconomic journals have been dominated by models that do not support the actual US policies used to deal with this latest economic downturn.  Alex Tabarrok confirms this, citing John Cochrane: Continue reading "Who Are US Policy Elites?" »
GD Star Rating
Tagged as: , ,

A Test of Moral Progress

We treat each other differently than our distant ancestors treated each other.  In particular we are “nicer”, at least by our lights, to a wider circle of associates.  For example, we enslave, rape, and murder each other less.  The two main explanations offered for this change are:

  1. Moral Progress – We have long wanted to act morally, and some of us have long pondered moral issues.  Relative to what we had thought, these experts have slowly discovered via reason that morality demands that we be nicer to a wider circle of others.  Experts told others, who believed them and wanted to act morally.   So we now more do what reason demands.
  2. Conditional Morality – Our evolved moral intuitions are context dependent.  We are built to be nicer to each other when times are good, to invest in an attractive reputation.  We are also built to form alliances with some in order to counter threats by others; the further in social distance are the threats we perceive, the wider a circle of allies we collect in response.  Since we are now richer and have interactions with more distant others, we are nicer to a wider range of allies.

These theories make different predictions about futures where we become poorer and our interactions become more local.  In their simplest forms, the moral progress theory predicts that we would continue to be as nice to as wide a circle of creatures in this situation, while the conditional morality theory predicts that the social circle to whom we are nice would narrow to the range of our ancestors with similar poverty and interaction locality.  Of course we might expect some inertial or momentum in moral attitudes; so it might take several generations of poverty and local interactions to really see the predicted difference.

My best guess is that cultural selection has produced real progress in institutions for keeping the peace, and perhaps in cultures to promote cooperation, but that any changes in our personal moral intuitions are due primarily due to an inherited conditional morality.   I expect we will actually see a future of much lower per-capita wealth, after the em transition, but it is hard to see a narrowing circle of interactions until there is substantial space colonization.

GD Star Rating
Tagged as: ,

Me in USA Today on SETI

But researchers such as Robin Hanson of George Mason University in Fairfax, Va., wonder whether the big picture really looks so promising when it comes to advanced life. Hanson supports SETI but finds it telling that humans haven’t come across anything yet.  “It has been remarkable and somewhat discouraging,” Hanson says, “that the universe is so damn big and so damn dead.”

More here, but little you don’t already know.  The reporter and I discussed lots of interesting issues, such as burning the cosmic commons, but this is what made the cut.  Other folks were cut entirely; I guess it pays to swear to reporters.

GD Star Rating
Tagged as: ,

Klyde the Barbarian

Stories set in the distant past often describe extremely gendered characters, like Conan the barbarian, or Raquel Welch in One Million Years B.C. The prototypical ripped caveman supposedly dragged his buxom woman to his cave by her hair.  It seems that rich folks today assume that civilization has forced them to sacrifice gender distinctiveness in order to become peaceful productive members of society.

But this is a myth; in fact poor cultures can’t afford very distinct genders; survival demands a similar mix cooperativeness, toughness, and so on from both men and women.  The typical barbarian was more a Klyde than a Conan. The data:

Sex differences in personality traits are larger in prosperous, healthy, and egalitarian cultures in which women have more opportunities equal with those of men. … Women reported higher levels of neuroticism, extraversion, agreeableness, and conscientiousness than did men across most nations. … Overall, higher levels of human development — including long and healthy life, equal access to knowledge and education, and economic wealth — were the main nation-level predictors of larger sex differences in personality. Changes in men’s personality traits appeared to be the primary cause of sex difference variation across cultures. It is proposed that heightened levels of sexual dimorphism result from personality traits of men and women being less constrained and more able to naturally diverge in developed nations. In less fortunate social and economic conditions, innate personality differences between men and women may be attenuated.

Thus one of the main way rich societies spend their wealth is to make their genders more distinct, especially via more extremely male men.  Men in rich cultures today are probably more distinctively male than at any time in history.  This fits with what I said about the recent rise of unwed moms:

Women free to pick a dad without expecting him to stay as a long term helper probably pick sexier men.  This should create more inequality in male access to women for sex and kids, and give men more free time to compete to be the few super-sexy super-dads.

An interesting related finding:

Manhood, in contrast to womanhood, is seen as a precarious state requiring continual social proof and validation. Because of this precariousness … men feel especially threatened by challenges to their masculinity.

GD Star Rating
Tagged as: ,

Tolstoy on Medicine

From the best novel ever, War and Peace:

“Natasha’s illness was so serious that, fortunately for her and for her parents, the consideration of all that had caused the illness, her conduct and the breaking off of her engagement, receded into the background. She was so ill that it was impossible for them to consider in how far she was to blame for what had happened. She could not eat or sleep, grew visibly thinner, coughed, and, as the doctors made them feel, was in danger. They could not think of anything but how to help her. Doctors came to see her singly and in consultation, talked much in French, German, and Latin, blamed one another, and prescribed a great variety of medicines for all the diseases known to them, but the simple idea never occurred to any of them that they could not know the disease Natasha was suffering from, as no disease suffered by a live man can be known, for every living person has his own peculiarities and always has his own peculiar, personal, novel, complicated disease, unknown to medicine—not a disease of the lungs, liver, skin, heart, nerves, and so on mentioned in medical books, but a disease consisting of one of the innumerable combinations of the maladies of those organs. Continue reading "Tolstoy on Medicine" »

GD Star Rating
Tagged as: ,

Newer Best Game Theory

Two of the four members of my PhD committee were McKelvey and Palfrey, originators of the “quantual response equilbrium” game theory concept, which for a while seemed to fit experimental data better than other game theories.  A year ago I wrote:

The latest American Economic Review says lab experiments have crowned a new best game theory … impulse balance equilibrium … The king is dead – long live the king!

Today I report a new tentative winner: equity-driven quantal response equilibrium. This is ordinary quantal response, but using modified payoffs: a player’s cash payment minus penalties for differences between player cash payments.  The new kind is dead – long live the new king.  The new results:

Substantial evidence has accumulated in recent empirical works on the limited ability of the Nash equilibrium to rationalize observed behavior in many classes of games played by experimental subjects. This realization has led to several attempts aimed at finding tractable equilibrium concepts which perform better empirically; one such example is the impulse balance equilibrium (Selten, Chmura, 2008), which introduces a psychological reference point to which players compare the available payoff allocations. This paper is concerned with advancing two new, empirically sound, concepts: equity-driven impulse balance equilibrium (EIBE) and equity-driven quantal response equilibrium (EQRE): both introduce a distributive reference point to the corresponding established stationary concepts known as impulse balance equilibrium (IBE) and quantal response equilibrium (QRE). The explanatory power of the considered models leads to the! following ranking, starting with the most successful in terms of fit to the experimental data: EQRE, IBE, EIBE, QRE and Nash equilibrium.

GD Star Rating
Tagged as:

The Confidence Heuristic

In Moore’s experiment, volunteers were given cash for correctly guessing the weight of people from their photographs. In each of the eight rounds of the study, the guessers bought advice from one of four other volunteers. The guessers could see in advance how confident each of these advisers was (see table), but not which weights they had opted for.

From the start, the more confident advisers found more buyers for their advice, and this caused the advisers to give answers that were more and more precise as the game progressed. This escalation in precision disappeared when guessers simply had to choose whether or not to buy the advice of a single adviser. In the later rounds, guessers tended to avoid advisers who had been wrong previously, but this effect was more than outweighed by the bias towards confidence. …

Moore said that following the advice of the most confident person often makes sense, as there is evidence that precision and expertise do tend to go hand in hand. For example, people give a narrower range of answers when asked about subjects with which they are more familiar.

There are times, however, when this link breaks down. With complex but politicised subjects such as global warming, for example, scientific experts who stress uncertainties lose out to activists or lobbyists with a more emphatic message.

That is from New Scientist back in June.  So the key hard question is: when does confidence credibly signal expertise, and when it is just empty cheap talk?  Hat tip to Toby Ord.

GD Star Rating
Tagged as: