GREGORY BENFORD
Physicist, UC Irvine; Author, Deep Time

Evolving the laws of physics

Richard Feynman held that philosophy of science is as useful to scientists as ornithology is to birds. Often this is so. But the unavoidable question about physics is — where do the laws come from?

Einstein hoped that God had no choice in making the universe. But philosophical issues seem unavoidable when we hear of  the "landscape" of possible string theory models. As now conjectured, the theory leads to 10500 solution universes — a horrid violation of Occam's Razor we might term "Einstein's nightmare."

I once thought that the laws of our universe were unquestionable, in that there was no way for science to address the question. Now I'm not so sure. Can we hope to construct a model of how laws themselves arise?

Many scientists dislike even the idea of doing this, perhaps because it's hard to know where to start. Perhaps ideas from the currently chic technology, computers, are a place to start. Suppose we treat the universe as a substrate carrying out computations, a meta-computer.

Suppose that precise laws require computation, which can never be infinitely exact. Such a limitation that might be explained by counting the computational capacity of a sphere around an "experiment" that tries to measure outcomes of those laws. The sphere expands at the speed of light, say, so longer experiment times give greater precision. Thinking mathematically, this sets a limit on how sharp differentials can be in our equations. A partial derivative of time cannot be better than the time to compute it.

In a sense, there may be an ultimate limit on how well known any law can be, especially one that must describe all of space-time, like classical relativity. It can't be better than the total computational capacity of the universe, or the capacity within the light sphere we can see.

I wonder if this idea can somehow define the nature of laws, beyond the issue of their precision? For example, laws with higher derivatives will be less descriptive because their operations cannot be carried out in a given volume over a finite time.

Perhaps the infinite discreteness required for formulating any mathematical system could be the limiting bound on such discussions. There should be energy bounds, too, within a finite volume, and thus limits on processing power set by the laws of thermodynamics. Still, I don't see how these arguments tell us enough to derive, say, general relativity.

Perhaps we need more ideas to derive a Law of Laws. Can we use the ideas of evolution? Perhaps invoke selection among laws that penalize those laws that lead to singularities — and thus taking those regions of space-time out of the game? Lee Smolin tried a limited form of this by supposing universes reproduce through black hole collapses. Ingenious, but that didn't seem to lead very far. He imagined some variation in reproduction of budded-off generations of universes, so their fundamental parameters varied a bit. Then selection could work.

In a novel of a decade ago, Cosmo, I invoked intelligent life, rather than singularities, to determine selection for universes that can foster intelligence, as ours seems to. (I didn't know about Lee's ideas at the time.) The idea is that a universe hosting intelligence evolves creatures that find ways in the laboratory to make more universes, which bud off and can further engender more intelligence, and thus more experiments that make more universes. This avoids the problem of how the first universe started, of course. Maybe the Law of Laws could answer that, too?

LERA BORODITSKY
Cognitive Psychology & Cognitive Neuroscience, Stanford University

Do our languages shape the nuts and bolts of perception, the very way we see the world?

I used to think that languages and cultures shape the ways we think. I suspected they shaped they ways we reason and interpret information.  But I didn't think languages could shape the nuts and bolts of perception, the way we actually see the world.  That part of cognition seemed too low-level, too hard-wired, too constrained by the constants of physics and physiology to be affected by language.

Then studies started coming out claiming to find cross-linguistic differences in color memory.  For example, it was shown that if your language makes a distinction between blue and green (as in English), then you're less likely to confuse a blue color chip for a green one in memory.  In a study like this you would see a color chip, it would then be taken away, and then after a delay you would have to decide whether another color chip was identical to the one you saw or not.

Of course, showing that language plays a role in memory is different than showing that it plays a role in perception.  Things often get confused in memory and it's not surprising that people may rely on information available in language as a second resort.  But it doesn't mean that speakers of different languages actually see the colors differently as they are looking at them.  I thought that if you made a task where people could see all the colors as they were making their decisions, then there wouldn't be any cross-linguistic differences.

I was so sure of the fact that language couldn't shape perception that I went ahead and designed a set of experiments to demonstrate this.  In my lab we jokingly referred to this line of work as "Operation Perceptual Freedom."  Our mission: to free perception from the corrupting influences of language.

We did one experiment after another, and each time to my surprise and annoyance, we found consistent cross-linguistic differences.  They were there even when people could see all the colors at the same time when making their decisions.  They were there even when people had to make objective perceptual judgments.  They were there when no language was involved or necessary in the task at all.  They were there when people had to reply very quickly.  We just kept seeing them over and over again, and the only way to get the cross-linguistic differences to go away was to disrupt the language system.  If we stopped people from being able to fluently access their language, then the cross-linguistic differences in perception went away.

JAMSHED BHARUCHA
Professor of Psychology, Provost, Senior Vice President, Tufts University

Education as Stretching the Mind

I used to believe that a paramount purpose of a liberal education was threefold:

1) Stretch your mind, reach beyond your preconceptions; learn to think of things in ways you have never thought before.

2) Acquire tools with which to critically examine and evaluate new ideas, including your own cherished ones.

3) Settle eventually on a framework or set of frameworks that organize what you know and believe and that guide your life as an individual and a leader.

I still believe #1 and #2. I have changed my mind about #3. I now believe in a new version of #3, which replaces the above with the following:

a) Learn new frameworks, and be guided by them.

b) But never get so comfortable as to believe that your frameworks are the final word, recognizing the strong psychological tendencies that favor sticking to your worldview. Learn to keep stretching your mind, keep stepping outside your comfort zone, keep venturing beyond the familiar, keep trying to put yourself in the shoes of others whose frameworks or cultures are alien to you, and have an open mind to different ways of parsing the world. Before you critique a new idea, or another culture, master it to the point at which its proponents or members recognize that you get it.

Settling into a framework is easy. The brain is built to perceive the world through structured lenses — cognitive scaffolds on which we hang our knowledge and belief systems.

Stretching your mind is hard. Once we've settled on a worldview that suits us, we tend to hold on. New information is bent to fit, information that doesn't fit is discounted, and new views are resisted.

By 'framework' I mean any one of a range of conceptual or belief systems — either explicitly articulated or implicitly followed. These include narratives, paradigms, theories, models, schemas, frames, scripts, stereotypes, and categories; they include philosophies of life, ideologies, moral systems, ethical codes, worldviews, and political, religious or cultural affiliations. These are all systems that organize human cognition and behavior by parsing, integrating, simplifying or packaging knowledge or belief. They tend to be built on loose configurations of seemingly core features, patterns, beliefs, commitments, preferences or attitudes that have a foundational and unifying quality in one's mind or in the collective behavior of a community. When they involve the perception of people (including oneself), they foster a sense of affiliation that may trump essential features or beliefs.

What changed my mind was the overwhelming evidence of biases in favor of perpetuating prior worldviews. The brain maps information onto a small set of organizing structures, which serve as cognitive lenses, skewing how we process or seek new information. These structures drive a range of phenomena, including the perception of coherent patterns (sometimes where none exists), the perception of causality (sometimes where none exists), and the perception of people in stereotyped ways.

Another family of perceptual biases stems from our being social animals (even scientists!), susceptible to the dynamics of in-group versus out-group affiliation. A well known bias of group membership is the over-attribution effect, according to which we tend to explain the behavior of people from other groups in dispositional terms ("that's just the way they are"), but our own behavior in much more complex ways, including a greater consideration of the circumstances. Group attributions are also asymmetrical with respect to good versus bad behavior. For groups that you like, including your own, positive behaviors reflect inherent traits ("we're basically good people") and negative behaviors are either blamed on circumstances ("I was under a lot of pressure") or discounted ("mistakes were made"). In contrast, for groups that you dislike, negative behaviors reflect inherent traits ("they can't be trusted") and positive behaviors reflect exceptions ("he's different from the rest"). Related to attribution biases is the tendency (perhaps based on having more experience with your own group) to believe that individuals within another group are similar to each other ("they're all alike"), whereas your own group contains a spectrum of different individuals (including "a few bad apples"). When two groups accept bedrock commitments that are fundamentally opposed, the result is conflict — or war.

Fortunately, the brain has other systems that allow us to counteract these tendencies to some extent. This requires conscious effort, the application of critical reasoning tools, and practice. The plasticity of the brain permits change - within limits.

To assess genuine understanding of an idea one is inclined to resist, I propose a version of Turing's Test tailored for this purpose: You understand something you are inclined to resist only if you can fool its proponents into thinking you get it. Few critics can pass this test. I would also propose a cross-cultural Turing Test for would-be cultural critics (a Golden Rule of cross-group understanding): before critiquing a culture or aspect thereof, you should be able to navigate seamlessly within that culture as judged by members of that group.

By rejecting #3, you give up certainty. Certainty feels good and is a powerful force in leadership. The challenge, as Bertrand Russell puts it in The History of Western Philosophy, is "To teach how to live without certainty, and yet without being paralyzed by hesitation".

DENIS DUTTON
Professor of the philosophy of art, University of Canterbury, New Zealand, editor of Philosophy and Literature and Arts & Letters Daily

The Self-Made Species

The appeal of Darwin's theory of evolution — and the horror of it, for some theists — is that it expunges from biology the concept of purpose, of teleology, thereby converting biology into a mechanistic, canonical science. In this respect, the author of The Origin of Species may be said to be the combined Copernicus, Galileo, and Kepler of biology. Just as these astronomers gave us a view of the heavens in which no angels were required to propel the planets in their orbs and the earth was no longer the center of the celestial system, so Darwin showed that no God was needed to design the spider's intricate web and that man is in truth but another animal.

That's how the standard story goes, and it is pretty much what I used to believe, until I read Darwin's later book, his treatise on the evolution of the mental life of animals, including the human species: The Descent of Man. This is the work in which Darwin introduces one of the most powerful ideas in the study of human nature, one that can explain why the capacities of the human mind so extravagantly exceed what would have been required for hunter-gatherer survival on the Pleistocene savannahs. The idea is sexual selection, the process by which men and women in the Pleistocene chose mates according to varied physical and mental attributes, and in so doing "built" the human mind and body as we know it.

In Darwin's account, human sexual selection comes out looking like a kind of domestication. Just as human beings domesticated dogs and alpacas, roses and cabbages, through selective breeding, they also domesticated themselves as a species through the long process of mate selection. Describing sexual selection as human self-domestication should not seem strange. Every direct prehistoric ancestor of every person alive today at times faced critical survival choices: whether to run or hold ground against a predator, which road to take toward a green valley, whether to slake an intense thirst by drinking from some brackish pool. These choices were frequently instantaneous and intuitive and, needless to say, our direct ancestors were the ones with the better intuitions.

However, there was another kind of crucial intuitive choice faced by our ancestors: whether to choose this man or that woman as a mate with whom to rear children and share a life of mutual support. It is inconceivable that decisions of such emotional intimacy and magnitude were not made with an eye toward the character of the prospective mate, and that these decisions did not therefore figure in the evolution of the human personality — with its tastes, values, and interests.  Our actual direct ancestors, male and female, were the ones who were chosen by each other.

Darwin's theory of sexual selection has disquieted and irritated many otherwise sympathetic evolutionary theorists because, I suspect, it allows purposes and intentions back into evolution through an unlocked side door. The slogan memorized by generations of students of natural selection is random mutation and selective retention. The "retention" in natural selection is strictly non-teleological, a matter of brute, physical survival. The retention process of sexual selection, however, is with human beings in large measure purposive and intentional. We may puzzle about whether, say, peahens have "purposes" in selecting peacocks with the largest tails. But other animals aside, it is absolutely clear that with the human race, sexual selection describes a revived evolutionary teleology. Though it is directed toward other human beings, it is as purposive as the domestication of those wolf descendents that became familiar household pets.

Every Pleistocene man who chose to bed, protect, and provision a woman because she struck him as, say, witty and healthy, and because her eyes lit up in the presence of children, along with every woman who chose a man because of his hunting skills, fine sense of humor, and generosity, was making a rational, intentional choice that in the end built much of the human personality as we now know it.

Darwinian evolution is therefore structured across a continuum. At one end are purely natural selective processes that give us, for instance, the internal organs and the autonomic processes that regulate our bodies. At the other end are rational decisions — adaptive and species-altering across tens of thousands of generations in prehistoric epochs.  It is at this end of the continuum, where rational choice and innate intuitions can overlap and reinforce one another, that we find important adaptations that are relevant to understanding the human personality, including the innate value systems implicit in morality, sociality, politics, religion, and the arts. Prehistoric choices honed the human virtues as we now know them: the admiration of altruism, skill, strength, intelligence, industriousness, courage, imagination, eloquence, diligence, kindness, and so forth.

The revelations of Darwin's later work — beautifully explicated as well in books by Helena Cronin, Amotz and Avishag Zahavi, and Geoffrey Miller — have completely altered my thinking about the development of culture. It is not just survival in a natural environment that has made human beings what they are. In terms of our personalities we are, strange to say, a self-made species. For me this is a genuine revelation, as it puts in a new genetic light many human values that have hitherto been regarded as purely cultural. 

CLAY SHIRKY
Social & Technology Network Topology Researcher; Adjunct Professor, NYU Graduate School of Interactive Telecommunications Program (ITP)

Religion and Science

I was a science geek with a religious upbringing, an Episcopalian upbringing, to be precise, which is pretty weak tea as far as pious fervor goes. Raised in this tradition I learned, without ever being explicitly taught, that religion and science were compatible. My people had no truck with Young Earth Creationism or anti-evolutionary cant, thank you very much, and if some people's views clashed with scientific discovery, well, that was their fault for being so fundamentalist.

Since we couldn't rely on the literal truth of the Bible, we needed a fallback position to guide our views on religion and science. That position was what I'll call the Doctrine of Joint Belief: "Noted Scientist X has accepted Jesus as Lord and Savior. Therefore, religion and science are compatible." (Substitute deity to taste.) You can still see this argument today, where the beliefs of Francis Collins or Freeman Dyson, both accomplished scientists, are held up as evidence of such compatibility.

Belief in compatibility is different from belief in God. Even after I stopped believing, I thought religious dogma, though incorrect, was not directly incompatible with science (a view sketched out by Stephen Gould as "non-overlapping magisteria".)  I've now changed my mind, for the obvious reason: I was wrong. The idea that religious scientists prove that religion and science are compatible is ridiculous, and I'm embarrassed that I ever believed it. Having believed for so long, however, I understand its attraction, and its fatal weaknesses.

The Doctrine of Joint Belief isn't evidence of harmony between two systems of thought. It simply offers permission to ignore the clash between them. Skeptics aren't convinced by the doctrine, unsurprisingly, because it offers no testable proposition. What is surprising is that its supposed adherents don't believe it either. If joint beliefs were compatible beliefs, there could be no such thing as heresy. Christianity would be compatible not just with science, but with astrology (roughly as many Americans believe in astrology as evolution), with racism (because of the number of churches who use the "Curse of Ham" to justify racial segregation), and on through the list of every pair of beliefs held by practicing Christians.

To get around this, one could declare that, for some arbitrary reason, the co-existence of beliefs is relevant only to questions of religion and science, but not to astrology or anything else. Such a stricture doesn't strengthen the argument, however, because an appeal to the particular religious beliefs of scientists means having to explain why the majority of them are atheists. (See the 1998 Larson and Witham study for the numbers.) Picking out the minority who aren't atheists and holding only them up as exemplars, is simply special pleading (not to mention lousy statistics.)

The works that changed my mind about compatibility were Pascal Boyer's Religion Explained, and Scott Atran's In Gods We Trust, which lay out the ways religious belief is a special kind of thought, incompatible with the kind of skepticism that makes science work. In Boyer and Atran's views, religious thought doesn't simply happen to be false -- being false is the point, the thing that makes belief both memorable and effective. Psychologically, we overcommit to the ascription of agency, even when dealing with random events (confirmation can be had in any casino.) Belief in God rides in on that mental eagerness, in the same way optical illusions ride in on our tendency to overinterpret ambiguous visual cues. Sociologically, the adherence to what Atran diplomatically calls 'counter-factual beliefs' serves both to create and advertise in-group commitment among adherents. Anybody can believe in things that are true, but it takes a lot of coordinated effort to get people to believe in virgin birth or resurrection of the dead.

We are early in one of the periodic paroxysms of conflict between faith and evidence. I suspect this conflict will restructure society, as after Galileo, rather than leading to a quick truce, as after Scopes, not least because the global tribe of atheists now have a medium in which they can discover one another and refine and communicate their message.

One of the key battles is to insist on the incompatibility of beliefs based on evidence and beliefs that ignore evidence. Saying that the mental lives of a Francis Collins or a Freeman Dyson prove that religion and science are compatible is like saying that the sex lives of Bill Clinton or Ted Haggard prove that marriage and adultery are compatible. The people we need to watch out for in this part of the debate aren't the fundamentalists, they're the moderates, the ones who think that if religious belief is made metaphorical enough,  incompatibility with science can be waved away. It can't be, and we need to say so, especially to the people like me, before I changed my mind.

RANDOLPH M. NESSE
Psychiatrist, University of Michigan; Coauthor, Why We Get Sick

Truth does not reside with smart university experts

I used to believe that you could find out what is true by finding the smartest people and finding out what they think. However, the most brilliant people keep turning out to be wrong.  Linus Pauling's ideas about Vitamin C are fresh in mind, but the famous physicist Lord Kelvin did more harm in 1900 with calculations based on the rate of earth's cooling that seemed to show that there had not been enough time for evolution to take place. A lot of the belief that smart people are right is an illusion caused by smart people being very convincing… even when they are wrong.

I also used to believe that you could find out what is true by relying on experts — smart experts — who devote themselves to a topic.  But most of us remember being told to eat margarine because it is safer than butter — then it turned out that trans-fats are worse.  Doctors told women they must use hormone replacement therapy (HRT) to prevent heart attacks — but HRT turned out to increase heart attacks.  Even when they are not wrong, expert reports often don't tell you what is true.  For instance, read reviews by experts about antidepressants; they provide reams of data, but you won't often find the simple conclusion that these drugs are not all that helpful for most patients.  It is not just others; I shudder to think about all the false beliefs I have unknowingly but confidently passed on to my patients, thanks to my trust in experts. Everyone should read the article by Ioannidis, "Why most published research findings are false."

Finally, I used to believe that truth had a special home in universities.  After all, universities are supposed to be devoted to finding out what is true, and teaching students what we know and how to find out for themselves. Universities may be best show in town for truth pursuers, but most stifle innovation and constructive engagement of real controversies, not just sometimes, but most of the time, systematically. 

How can this be? Everyone is trying so hard to encourage innovation!  The Regents take great pains to find a President who supports integrity and creativity, the President chooses exemplary Deans, who mount massive searches for the best Chairs. Those Chairs often hire supporters who work in their own areas, but what if one wants to hire someone doing truly innovative work, someone who might challenge established opinions?  Faculty committees intervene to ensure that most positions go to people just about like themselves, and the Dean asks how much grant overhead funding a new faculty member will bring in.  No one with new ideas, much less work in a new area or critical of established dogmas, can hope to get through this fine sieve.  If they do, review committees are waiting. And so, by a process of unintentional selection, diversity of thought and topic is excluded.  If it still sneaks in, it is purged.  The disciplines become ever more insular. And universities find themselves unwittingly inhibiting progress and genuine intellectual engagement.  University leaders recognize this and hate it, so they are constantly creating new initiatives to foster innovative interdisciplinary work.  These have the same lovely sincerity as new diets for the New Year, and the same blindness to the structural factors responsible for the problems.

Where can we look to find what is true?  Smart experts in universities are a place to start, but if we could acknowledge how hard it is for truth and its pursuers to find safe university lodgings, and how hard it is for even the smartest experts to offer objective conclusions, we could begin to design new social structures that would support real intellectual innovation and engagement.

BART KOSKO
Information Scientist, USC; Author, Noise

THE SAMPLE MEAN

I have changed my mind about using the sample mean as the best way to combine measurements into a single predictive value.  Sometimes it is the best way to combine data but in general you do not know that in advance.  So it is not the one number from or about a data set that I would want to know in the face of total uncertainty if my life depended on the predicted outcome.

Using the sample mean always seemed like the natural thing to do.  Just add up the numerical data and divide by the number of data.  I do not recall ever doubting that procedure until my college years.  Even then I kept running into the mean in science classes and even in philosophy classes where the discussion of ethics sometimes revolved around Aristotle's theory of the "golden mean." There were occasional mentions of medians and modes and other measures of central tendency but they were only occasional.      

The sample mean also kept emerging as the optimal way to combine data in many formal settings.  At least it did given what appeared to be the reasonable criterion of minimizing the squared errors of the observations.  The sample mean falls out from just one quick application of the differential calculus.  So the sample mean had on its side not only mathematical proof and the resulting prominence of appearing in hundreds if not thousands of textbooks and journal articles.  It was and remains the evidentiary workhorse of modern applied science and engineering.  The sample mean summarizes test scores and gets plotted in trend lines and centers confidence intervals among numerous other applications.

Then I ran into the counter-example of Cauchy data.  These data come from bell curves with tails just slightly thicker than the familiar "normal" bell curve.  Cauchy bell curves also describe "normal" events that correspond to the main bell of the curves.  But Cauchy bell curves have thicker tails than normal bell curves have and these thicker tails allow for many more "outliers" or rare events.  And Cauchy bell curves arise in a variety of real and theoretical cases.  The counter-example is that the sample mean of Cauchy data does not improve no matter how many samples you combine.  This result contrasts with the usual result from sampling theory that the variance of the sample mean falls with each new measurement and hence predictive accuracy improves with sample size (assuming that the square-based variance term measures dispersion and that such a mathematical construct always produces a finite value — which it need not produce in general).  The sample mean of ten thousand Cauchy data points has no more predictive power than does the sample mean of ten such data points.  Indeed the sample mean of Cauchy data has no more predictive power than does any one of the data points picked at random.  This counter-example is but one of the anomalous effects that arise from averaging data from many real-world probability curves that deviate from the normal bell curve or from the twenty or so other closed-form probability curves that have found there way into the literature in the last century.

Nor have scientists always used the sample mean.  Historians of mathematics have pointed to the late sixteenth century and the introduction of the decimal system for the start of the modern practice of computing the sample mean of data sets to estimate typical parameters.  Before then the mean apparently meant the arithmetic average of just two numbers as it did with Aristotle.  So Hernan Cortes may well have had a clear idea about the typical height of an adult male Aztec in the early sixteenth century.  But he quite likely did not arrive at his estimate of the typical height by adding measured heights of Aztec males and then dividing by the number added.  We have no reason to believe that Cortes would have resorted to such a computation if the Church or King Charles had pressed him to justify his estimate.  He might just as well have lined up a large number of Aztec adult males from shortest to tallest and then reported the height of the one in the middle.

There was a related and deeper problem with the sample mean:  It is not robust.  Extremely small or large values distort it.  This rotten-apple property stems from working not with measurement errors but with squared errors.  The squaring operation exaggerates extreme data even though it greatly simplifies the calculus when trying to find the estimate that minimizes the observed errors.  That estimate turns out to be the sample mean but not in general if one works with the raw error itself or other measures.  The statistical surprise of sorts is that using the raw or absolute error of the data gives the sample median as the optimal estimate.

The sample median is robust against outliers.  If you throw away the largest and smallest values in a data set then the median does not change but the sample mean does (and gives a more robust "trimmed" mean as used in combining the judging scores in figure skating and elsewhere to remove judging bias).  Realtors have long since stated typical housing prices as sample medians rather than sample means because a few mansions can so easily skew the sample mean.  The sample median would not change even if the price of the most expensive house rose to infinity.  The median would still be the middle-ranked house if the number of houses were odd.  But this robustness is not a free lunch.  It comes at the cost of ignoring some of the information in the numerical magnitudes of the data and has its own complexities for multidimensional data.

Other evidence pointed to using the sample median rather than the sample mean.  Statisticians have computed the so-called breakdown point of these and other statistical measures of central tendency.  The breakdown point measures the largest proportion of data outliers that a statistic can endure before it breaks down in a formal sense of producing very large deviations.  The sample median achieves the theoretical maximum breakdown point.  The sample mean does not come close.  The sample median also turns out to be the optimal estimate for certain types of data (such as Laplacian data) found in many problems of image processing and elsewhere — if the criterion is maximizing the probability or likelihood of the observed data.  And the sample median can also center confidence intervals.  So it too gives rises to hypothesis tests and does so while making fewer assumptions about the data than the sample mean often requires for the same task. 

The clincher was the increasing use of adaptive or neural-type algorithms in engineering and especially in signal processing.  These algorithms cancel echoes and noise on phone lines as well as steer antennas and dampen vibrations in control systems.  The whole point of using an adaptive algorithm is that the engineer cannot reasonably foresee all the statistical patterns of noise and signals that will bombard the system over its lifetime.  No type of lifetime average will give the kind of performance that real-time adaptation will give if the adaptive algorithm is sufficiently sensitive and responsive to its measured environment.  The trouble is that most of the standard adaptive algorithms derive from the same old and non-robust assumptions about minimizing squared errors and thus they result in the use of sample means or related non-robust quantities.  So real-world gusts of data wind tend to destabilize them.  That is a high price to pay just because in effect it makes nineteenth-century calculus computations easy and because such easy computations still hold sway in so much of the engineering curriculum.  It is an unreasonably high price to pay in many cases where a comparable robust median-based system or its kin both avoids such destabilization and performs similarly in good data weather and does so for only a slightly higher computational cost.  There is a growing trend toward using robust algorithms.  But engineers still have launched thousands of these non-robust adaptive systems into the stream of commerce in recent years.  We do not know whether the social costs involved from using these non-robust algorithms are negligible or substantial.

So if under total uncertainty I had to pick a predictive number from a set of measured data and if my life depended on it — I would now pick the median.

DAVID GELERNTER
Computer Scientist, Yale University; Chief Scientist, Mirror Worlds Technologies; Author, Drawing Life

Users Are Not Reactionary After All

What I've changed my mind about is that the public is wedded to obsolete 1970s GUIs & info mgmt forever — PARC's desktop & Bell Labs' Unix file system. I'll give two example from my own experience. Both constitute long term ideas of mine and might seem like self-promotion, but my point is that as a society we don't have the patience to develop fully those big ideas that need time to soak in.

I first described a GUI called "lifestreams" in the Washington Post in 1994. By the early 2000s, I thought this system was dead in the water, destined to be resurrected in a grad student's footnote around the 29th century, The problem was (I thought) that Lifestreams was too unfamiliar, insufficiently "evolutionary" and too "revolutionary" (as the good folks at ARPA like to say [or something like that]); you need to go step-by-step with the public and the industry or you lose.

But today "lifestreams" are all over the net (take a look yourself), and I'm told that "lifestreaming" has turned into a verb at some recent Internet conferences. According to ZDnet.com, "Basically what's important about the OLPC [one laptop per child], has nothing to do with its nominal purposes and everything to do with its interface. Ultimately traceable to David Gelernter's 'Lifestreams' model, this is not just a remake of Apple's evolution of the original work at Palo Alto, but something new."

Moral: the public may be cautions but is not reactionary.

In a 1991 book called Mirror Worlds, I predicted that everyone would be putting his personal stuff in the Cybersphere (AKA "the clouds"); I said the same in a 2000 manifesto on Edge called "The 2nd Coming", & in various other pieces in between. By 2005 or so, I assumed that once again I'd jumped the gun, by too long to learn the results pre-posthumously — but once again this (of all topics) turns out to hot and all over the place nowadays. "Cloud computing" is the next big thing: What does this all prove? If you're patient, good ideas find audiences. But you have to be very patient.

And if you expect to cash in on long-term ideas in the United States, you're certifiable.

This last point is a lesson I teach my students, and on this item I haven't (and don't expect to) change my mind. But what the hell? It's New Year's, and there are worse things than being proved right once in a while, even if it's too late to count.