Monthly Archives: April 2018

Summary of Nick Bostrom’s “Ethical Issues in Advanced AI”

Nick Bostrom (1973 – ) holds a Ph.D. from the London School of Economics (2000). He is a co-founder of the World Transhumanist Association (now called Humanity+) and co-founder of the Institute for Ethics and Emerging Technologies. He was on the faculty of Yale University until 2005, when he was appointed Director of the newly created Future of Humanity Institute at Oxford University. He is currently Professor, Faculty of Philosophy & Oxford Martin School; Director, Future of Humanity Institute; and Director, Program on the Impacts of Future Technology; all at Oxford University.

His recent book, Superintelligence: Paths, Dangers, Strategies, is the definitive work on superintelligence. A few of its main issues were discussed in his previous article, Ethical Issues in Advanced AI.” Here is a brief outline of that article. 

Introduction – “A superintelligence is any intellect that vastly outperforms the best human brains in practically every field, including scientific creativity, general wisdom, and social skills. This definition leaves open how the superintelligence is implemented – it could be in a digital computer, an ensemble of networked computers, cultured cortical tissue, or something else.”  B states that there is no reason to believe we won’t have SI in the lifetime of some persons alive today.

Superintelligence (SI) is different – And in ways, we can’t even imagine.

Moral Thinking of SI – If morality is a cognitive pursuit, then SI should be able to solve moral issues in ways previously undreamt of.

Importance of Initial Motivations – It is crucial to design SI to be friendly.

Should Development Be Delayed or Accelerated? – “It is hard to think of any problem that a superintelligence could not either solve or at least help us solve. Disease, poverty, environmental destruction, unnecessary suffering of all kinds: these are things that a superintelligence equipped with advanced nanotechnology would be capable of eliminating. Additionally, a superintelligence could give us indefinite lifespan, either by stopping and reversing the aging process.”

Given this promise, and considering B’s claim that SI will probably be developed anyway, we might as well do this asap. “If we get to superintelligence first, we may avoid this risk from nanotechnology and many others. If, on the other hand, we get nanotechnology first, we will have to face both the risks from nanotechnology and, if these risks are survived, also the risks from superintelligence.”

Reflection – I have made my views on this clear many times. Despite the risks, we need to develop superintelligence promptly if we are to have any chance of surviving.

The Solitary Life

Image result for Solitude free images
I recently I wrote about Kahlil Gibran and solitude. hile many sages and seers have claimed there is something preferable in the solitary life, I’m skeptical of overemphasizing the need for it. We are social beings and there is much evidence that loneliness is a major problem in the modern world.

Still, there has been a lot written lately about the benefits of the solitary life, or at least one with sufficient solitary time. (Note though that this has nothing whatsoever to do with the barbaric torture of solitary confinement.) Here are a few recent short articles on the topic:

The pursuit of loneliness: how I chose a life of solitude” (The Guardian)

The Virtues of Isolation“(The Atlantic)

Embracing a Life of Solitude“(The New York Times)

The Paradox of Dying Lonely and Living in Solitude” (HuffPost)

I am not endorsing any conclusions in these articles, simply referring them to those who might be interested.


Why Earth’s History Appears So Miraculous

A person holds a globe against a background of Earths hit by meteors, crumbling, and colliding with rockets.

Here is a brief recap of Peter Brannen’s recent piece in the Atlantic: “Why Earth’s History Appears So Miraculous: The strange, cosmic reason our evolutionary path will look ever luckier the longer we survive.

Brannen begins by introducing us to the observer selection effect. (An observation selection effect exists when some property of a thing is correlated with the observer existing in the first place. For example, if intelligence hadn’t evolved, we wouldn’t exist, and couldn’t evaluate the probability of intelligence evolving.) The question Brannen asks is whether this bias applies to our planet.

If you consider all the existential threats our planets faced: “It’s something of a miracle that life on our planet has been left to evolve without fatal interruption for billions of years.” Of course, if we had been wiped out we wouldn’t be here to contemplate our existence. Thus, as a result of the selection effect, we probably underestimate the observed frequencies of cataclysmic events. And this means that “our forecasts about the future could be blinded by our necessarily lucky past.”

As Anders Sandberg, a senior research fellow at Oxford’s Future of Humanity Institute says:

Maybe the universe is super dangerous and Earth-like planets are destroyed at a very high rate, but if the universe is big enough, then when observers do show up on some very, very rare planets, they’ll look at the record of meteor impacts and disasters and say, ‘The universe looks pretty safe!’ But the problem is, of course, that their existence depends on them being very, very lucky. They’re actually living in an unsafe universe and next Tuesday they might get a very nasty surprise.

Perhaps this explains the Fermi paradox too. The reason we find no evidence of alien life is because there are no (or few) aliens that have survived and we exist only due to unimaginable good luck.

Consider, for example, that so far we haven’t destroyed ourselves with nuclear weapons. This may lead us to believe that nuclear war is unlikely, but if we take observer selection effects into account we’ll realize that the fact that we’re still here tells us little about our chances for future survival. Nuclear annihilation may be virtually certain but in a big enough universe some civilizations avoid extinction for a long time. We may seem safe but more likely we’ve just been lucky. (In fact, we’ve come close to destroying ourselves with nuclear weapons multiple times. Our planet is just the one where Stanislav Petrov didn’t push the button.)

Extending this insight further, the observer selection effect may explain why the universe hasn’t succumbed to vacuum decay or some other catastrophe. We exist so we think such a scenario is unlikely, but that’s because we’re in a universe that has survived. The situation gets even stranger if you consider the “many worlds” multiverse. As the cosmologist, Anthony Aguirre states:

… suppose the many-worlds interpretation of quantum mechanics is right … So, one of the two versions of us ceases to exist, but do we actually notice that? So one of us keeps going on just as if nothing happened. Arguably, from moment to moment, I can’t rule out that five minutes ago the other version of us died. There’s no way for me to say that. So there’s an interesting, troubling question as to whether these things could be happening all the time and we just don’t even notice it.

All I can say is that the mind boggles at such suggestions. When confronted with possible implications of multiple realities my mind recoils.

Summary of “It’s Time to Fight Dirty: How Democrats Can Build a Lasting Majority in American Politics”

Political science professor David Faris has just published a new book: It’s Time to Fight Dirty: How Democrats Can Build a Lasting Majority in American Politics. The author begins by noting the many problems plaguing American politics today: we live in an era in which a tiny majority of folks in sparsely populated states have a wildly disproportionate impact on policy, when supreme court seats are stolen, when voting is suppressed, when a constitution built for the 18th century is considered sacred for the 21st, etc. It is an ugly time and Democrats must remain united until they can regain power. Then they must fight dirty as the Republicans have been doing for the last twenty years to avoid a total takeover of society by the plutocrats.

Here are a few of the perfectly legal steps that Faris suggests Democrats could take if they win Congress and the Presidency in 2020. They would require only that Congress pass laws and wouldn’t need any constitutional amendments.

  • Abolish the filibuster so any bill can pass the Senate with 50 votes and the veep
  • Make D.C. a state, which would add two black Democratic senators to the Senate
  • Make Puerto Rico a state, which would add two Latino Democrats to the Senate
  • Break California into multiple states to add yet more Democrats to the Senate
  • Expand the Supreme Court to 11 (or more) justices, and have the president appoint 40-year-olds

They could then reform voting rights by:

  • requiring a 2-week period of early voting from 6 a.m. to 10 p.m. every day
  • making make pre-registration of 17-year-olds legal nationwide
  • forbidding states from passing voter-ID laws or imposing other non-Constitutional voting requirements
  • making it a federal crime to intimidate voters
  • making Election Day a national holiday for federal employees (and others)
  • fighting gerrymandering by doubling the size of the House (and having multimember districts with proportional representation.)

While there is much to be said about this great new book let me bring my own background in game theory to the issue. For our purposes, we need only a brief understanding of the prisoner’s dilemma. A prisoner’s dilemma is an interactive situation in which it is better for all to cooperate rather than for no one to do so, yet it is best for each not to cooperate, regardless of what the others do. (I’ve explained this in detail below.)

What this means in layperson’s terms is that if your opponent plays dirty and you play nice you get steamrolled, you become a sucker. In this case, you have no choice but to fight fire with fire or you will be dominated. Ideally, you would all cooperate with each other. (You allow each other a Supreme Court seat when the President is of your party; you don’t threaten the world economy by not raising the debt ceiling, you don’t abuse the filibuster, etc.) But if the other side violates the norms of governing you either fight back or you will be dominated.

Of course the bad thing about all this is that it will lead to retailation from your opponents. You will then be in a state of war with each other and everyone will do worse, unless one side wins and dominates the other. This is how I see it all unfolding.


Addendum – I did graduate work and published multiple peer-reviewed articles on game theory. Here is a more detailed summary of game theory and the prisoner’s dilemma.

 Game Theory

For our purposes, a game is an interactive situation in which individuals, called players, choose strategies to deal with each other in attempting to maximize their individual utility. There are several ways of distinguishing games including: 1) in respect to the number of players involved; 2) in respect to the number of repetitions of play; 3) in respect of the order of the various player’s preferences over the same outcomes. On the one extreme are games of pure conflict, so-called zero-sum games, in which players have completely opposing interests over possible outcomes. On the other extreme are games of pure harmony, so-called games of coordination. In the middle are games involving both conflict and harmony in respect of others. It is one particular game that interests us most, since it describes the situation in Hobbes’ state of nature, and is the central problem in contractarian moral theory.

The Prisoner’s Dilemma

The prisoner’s dilemma is one of the most widely debated situations in game theory. The story has implications for a variety of human interactive situations. A prisoner’s dilemma is an interactive situation in which it is better for all to cooperate rather than for no one to do so, yet it is best for each not to cooperate, regardless of what the others do.

In the classic story, two prisoners have committed a serious crime but all of the evidence necessary to convict them is not admissible in court. Both prisoners are held separately and are unable to communicate. The prisoners are called separately by the authorities and each offered the same proposition. Confess and if your partner does not, you will be convicted of a lesser crime and serve one year in jail while the unrepentant prisoner will be convicted of a more serious crime and serve ten years. If you do not confess and your partner does, then it is you who will be convicted of the more serious crime and your partner of the lesser crime. Should neither of you confess the penalty will be two years for each of you, but should both of you confess the penalty will be five years. In the following matrix, you are the row chooser and your partner the column chooser. The first number in each parenthesis represents the “payoff” for you in years in prison, the second number your partner’s years. Let us assume each player prefers the least number of years in prison possible. In matrix form, the situation looks like this:

                                                                                                                                Prisoner 2

    Confess  Don’t Confess
 Prisoner 1 Confess (5, 5) (1, 10)
Don’t Confess (10, 1) (2, 2)

So you reason as follows: If your partner confesses, you had better confess because if you don’t you will get 10 years rather than 5. If your partner doesn’t confess, again you should confess because you will only get 1 year rather than 2 for not confessing. So no matter what your partner does, you ought to confess. The reasoning is the same for your partner. The problem is that when both confess the outcome is worse for both than if neither confessed. You both could have done better, and neither of you worse, if you had not confessed! You might have made an agreement not to confess but this would not solve the problem. The reason is this: although agreeing not to confess is rational, compliance is surely not rational!

The prisoner’s dilemma describes the situation that humans found themselves in in Hobbes’ state of nature. If the prisoners cooperate, they both do better; if they do not cooperate, they both do worse. But both have a good reason not to cooperate; they are not sure the other will! We can only escape this dilemma, Hobbes maintained, by installing a coercive power that makes us comply with our agreements (contracts). Others, like the contemporary philosopher David Gauthier, argue for the rationality of voluntary non-coerced cooperation and compliance with agreements given the costs to each of us of enforcement agencies. Gauthier advocates that we accept “morals by agreement.”

How Democracies Die: Trump and American Democracy

Trump’s long-term effect on American democracy: How worried should we be?

Retired mathematics professor Doug Mudar penned the above-named essay Monday on his blog “The Weekly Sift.” Its careful analysis reflects a well-ordered mind molded by the rigors of earning a Ph.D. in mathematics at the University of Chicago. Mudar begins with a partial list of just a few of the many things that are concerning about the current regime.

To say the above is banana republic stuff and that Trump is authoritarian and undemocratic is an understatement. Still, the government hasn’t totally collapsed and much of it is fighting back. So how worried should we be? Will we get over Trump or are authoritarianism and dictatorship near? As Mudar notes we don’t need to speculate as the research on this question has been done by Harvard political scientists Steven Levitsky and Daniel Ziblatt in their new book How Democracies Die.

One of the key findings of this research is that laws, constitutions, and political institutions aren’t enough to protect against autocrats. Instead, democracies only survive if certain norms are respected—for example not denying supreme court nominations a hearing, misusing the filibuster,  shutting down the government or threatening the world economy with debt ceiling brinkmanship. As Mudar puts it:

The Constitution never says that the President can’t order the FBI to investigate the candidate he just defeated, that he can’t tell big whopping lies on a regular basis, or that he has to give the public enough information to judge whether or not he is corrupt. Those aren’t rules, they’re just good practices.

While a list of the norms that Trump violates would fill volumes, Levitsky and Ziblatt point to two essential norms:

  • mutual toleration, “the understanding that competing parties accept one another as legitimate rivals”
  • forbearance, “the idea that politicians should exercise restraint in deploying their institutional prerogatives”

Without these restraints, mutual cooperation if you will, the tit for tat will continue. Mudar offers a poignant example of where this could lead:

The 12th Amendment specifies that the sealed votes of the Electoral College are sent to the President of the Senate, who counts them “in the presence of the Senate and House of Representatives.” What if the President of the Senate, with the connivance of majorities in both houses, simply miscounted the votes and proclaimed someone else to be president?

There’s no provision for dealing with that scenario — and with innumerable similar situations — because the Founders never anticipated that our political leaders would go that far. And they wouldn’t. Or would they?

Next Mudar considers that the old model of democratic breakdown, the coup, has been replaced by a newer model. You maintain the outward appearance of democracy but attack voting rights, political rivals, journalists, and media that doesn’t spout the party line. To better explain all this Levitsky and Ziblatt use a soccer analogy to show how an elected president becomes an autocrat:

  • Capture the referees. In other words, get your people in charge of the judiciary, law enforcement, and intelligence, tax, and regulatory agencies. Anyone who used to be a neutral arbiter must become your partisan. You can do this in the judiciary, for example, by expanding the size of the Supreme Court and appointing your people to the new positions (as Roosevelt tried to do), or by impeaching judges who rule against you (as the Republican-controlled legislature is trying to do in Pennsylvania). (In North Carolina, the gerrymandered Republican majority in the legislature has done court-packing in reverse: It shrunk the size of the State Court of Appeals to prevent the new Democratic governor from filling the open seats.)
  • Sideline star players on the other side. “Opposition politicians, business leaders who finance the opposition, major media outlets, and … religious or other cultural figures” are “sidelined, hobbled, or bribed into throwing the game.” With the referees already in your pocket, the carrots of government contracts and positions, or the sticks of ruinous regulations, taxes, and prosecutions can hollow out the institutions that otherwise might channel public opinion against you.
  • Rewrite the rules in your favor. We were already seeing a lot of rule-rewriting on the state level prior to Trump: Gerrymandering and voter suppression have locked in large Republican majorities in states (like North Carolina) where the voters are more-or-less evenly split between the parties. In last November’s election in Virginia, Democratic candidates for the House of Delegates won the popular vote 53%-44%, but Republicans maintained a 51-49 majority. Combining a biased legal system with a lifetime ban on felon voting (as in Florida, where the Sentencing Project estimates that 20% of adult blacks can’t vote) can sideline a large chunk of the opposition electorate. In countries like Russia, field-tilting rules make it difficult for new parties to form, for genuine opposition candidates to get on the ballot, or for opposition voices to get their message out.

Once all this has transpired an autocrat can act with impunity without secret police and gulags. Of course, this takes time and while he is corrupting some referees others are standing up to oppose him. We might also get lucky that the aspiring dictator isn’t very talented or adept at taking control—as is the case with Trump. Hopefully, Trump will leave peacefully and we will enact new measures to protect ourselves in the future— forcing candidates to release their tax returns, divest from their companies, etc.

Mudar also gives a warning about a dystopian future:

So far, democracy has been protected by two main forces: The so-called “Deep State” (i.e., career government officials who are more committed to the missions of their organizations than to the orders they receive from the White House) and Trump’s overall unpopularity.

So, for example, career prosecutors—even if they are Republicans—have not been willing to sacrifice their integrity by manufacturing a case against Hillary Clinton, or ignoring evidence against Trump himself, just because he tweets that they should. Career EPA officials are refusing to become pawns of the fossil fuel industry no matter how much Scott Pruitt wants them to. Career economists at the Treasury didn’t concoct a bogus tax-cuts-pay-themselves analysis just because Steve Mnunchin promised they would.

That’s the Deep State in action: It’s not a conspiracy masterminded by some shadowy cabal. It’s the professional integrity of people who believe that their jobs mean more than just a paycheck or their bosses’ approval …

That’s both its strength and its weakness. You can’t kill the Deep State just by finding its leader and bribing, threatening, or imprisoning him or her. But conversely, it has no sense of strategy. It is made up of individuals, and individuals can be worn down. The Deep State has held its own for a little over a year, but can it hold for four years or eight?

But what if Trump got to replace one or two of the liberals on the Supreme Court or his popularity increased? It’s easy to imagine democracy crumbling altogether if:

the economy stays strong, the country avoids any new shooting wars or trade wars, and Trump’s victims immigrants, Muslims, LGBT people, etc.—remain isolated. Much of the country then starts to say, “What was all that alarmism about?” When Jim Comey or Andrew McCabe winds up in jail, it seems like a one-off case rather than an assault on law enforcement.

Mudar thinks this will all come to a head in 2018. The outcome will depend on many factors including whether Democrats stay united against Trump, regain power despite gerrymandering and start to reestablish democratic norms. It also depends on whether Republicans will come to understand what is at stake and join the resistance or at least not oppose it. Mudar concludes with a chilling warning:

Long term, both parties need to figure out how to strengthen the norms of forbearance and tolerance, which were in trouble long before Trump arrived on the scene. Unless we can re-establish them, getting past Trump will not solve our problems. His failure, if it happens, might simply be a training example for new and better demagogues.