Traveling Wave Reactors

The WSJ has a new article about TerraPower, a start-up that plans to expand nuclear energy with a new reactor design. They claim their traveling wave reactor (TWR) could run for up to 100 years without needing to be refueled, making it safer and more efficient than conventional reactors.

The secret to the TWR's endurance is in the core. Conventional reactors sustain nuclear fission more or less uniformly throughout the entire fuel supply. In a TWR's core, only a small region of the fuel is undergoing fission at any given time. Because this limits the core's rate of fission, much more fuel can be built into the core without increasing the power output of the reactor. If conventional reactors are sprinters, then TWRs are marathon runners.

TerraPower hopes that the relatively maintenance-free TWR will gain a foothold in developing nations, where technical knowledge and specialized equipment are in short supply. The article points out that prospects for TWRs in the United States are bleak. Despite their advantages, regulatory restrictions make licensing any new reactor design in the US a very lengthy task.

Unfortunately, no amount of clever engineering can help to improve the public perception of nuclear power. In my experience, there are many people who dismiss nuclear power out of hand, without even trying to balance risks and benefits. Even a perfectly safe, perfectly efficient reactor will do nothing to change the minds of some folk.

Maybe a name change is in order. Magnetic resonance imaging (MRI) used to be called nuclear magnetic resonance (NMR). The name was changed for clinical applications to soothe patient fears over harmful radiation. Current generation nuclear power plants are so much safer than older style plants (like Chernobyl) that it seems a shame for them to share the same name.

How about "Neutron Cascade Power Plants"?

Observing Rotating Black Holes

In my post on ergospheres, I mentioned how spinning massive objects drag spacetime around themselves like a whirlpool. An article (abstract) published in Nature Physics suggests it's possible to directly observe the effects of rotating spacetime around spinning black holes:

In the latest work, Fabrizio Tamburini of the University of Padova in Italy and colleagues instead show how to detect the rotation by measuring changes to the light from a distant star or from the disk of accreted material surrounding a black hole. They point out that a wavefront travelling in a plane perpendicular to the black hole’s axis of spin will get twisted as it passes close to the black hole, since half of the wave front will be moving in the direction of advancing space-time and the other half in the direction of receding space-time. In other words, the phase of the radiation emanating from close to a rotating black hole should have a distinctive distribution in space.

The authors claim that the altered light from distant spinning black holes should be observable with the best modern telescopes. This is good news because rotating spacetime and spinning black holes have yet to be observed. Indirectly detecting a spinning black hole would be strong evidence for both phenomena.

Source: Physorg.com

Did the Ancient Romans Have Hot Air Balloons?

What a coincidence! Over dinner last week I was struck by the absurdity of the ancient Egyptians *not* inventing hot air balloons. I promptly forgot all about the subject until, quite at random, I came across a blog post by Mike Darwin touching on nearly the same subject:

The ‘mundane’ observation that caused Joseph-Michel Montgolfier (1740 –1810 CE) to invent the hot air balloon was just as accessible to the Romans of ~400 BCE, as were the materials and technologies required to construct human carrying hot air balloons. Certainly, the same motivations were present in both cultures at both times: Joseph Montgolfier was contemplating how to successfully assault the British fortress of Gibraltar, which had proved impregnable to the French by both sea and land, when he noticed how floating embers from a fire he was laying next to were carried aloft and over great distances; giving him the idea of lighter than air flight. The Romans, a military people with similar problems, as well as a love of spectacle and a penchant for technological innovation in war, could just as easily have developed lighter than air manned flight – and yet they did not. There are no Roman frescoes of hot air balloons, whether for war or celebration, drifting over the Empire’s capital.

He uses this to springboard into a much larger discussion about technological progress. The post is well worth reading in its entirety. But I disagree with the certainty Mike expresses here. We don't know that the Romans lacked hot air balloons, only that there's no surviving evidence they did. And evidence from that time is sparse at best.

The Antikythera Mechanism which Mike mentions in the post is a good example of just how little we know. As he says:

The Antikythera Mechanism has forced a complete re-evaluation of the technology of the ancient world. The device contained 32 gears, assembled into a mechanism that accurately reproduced the motion of the sun and the moon against the background of fixed stars, with a differential drive giving their relative position, and thus the phases of the moon.4 More recently, it has been discovered that device also integrates eclipse prediction with cycles of human institutions, most notably the Olympics!

The technology used to produce the Antikythera Mechanism rivals that used in the best 16th century clocks, and the understanding of planetary motions embodied in the workings of the device suggest that some form of the calculus may have been in use by its makers. It is also clear from the complexity and precision of the device that it was not a prototype, but rather represents a well developed, and arguably a mature technology, which must have had other applications.

The Antikythera Mechanism was only discovered very recently. Furthermore, it's entirely singular in age and complexity. An entire industry of mathematicians, astronomers, and craftsmen must have existed in order to build such a machine. Of that industry, only one device survives to the present day. If we never found the Antikythera Mechanism then we would be entirely ignorant of the advanced industry that must have existed around 80 BC.

The Antikythera Mechanism doesn't imply that the Romans had hot air balloons, but it does underscore our inadequate knowledge of ancient technology. And I still think it would be damn cool if the Egyptians had hot air balloons.

Too Many Lasers

The historical trend in large scale fiber optic networks has been towards fewer lasers. A typical network architecture will place a laser on both ends of a fiber optic line to allow two-way communication, plus additional lasers in between for switching or amplification. Passive optical networks (PONs) such as Verizon's FiOS replace the lasers between the two network endpoints with optical splitters. Splitters work similarly to prisms, breaking up a single optical signal from the central office into independant beams that serve a dozen or more customers.

Now, a team at A*STAR Institute for Microelectronics have demonstrated (abstract) a fiber optic network with even fewer lasers. The architecture is similar to standard PON, but there's additional fiber that carries a continuous laser beam from the central office out to all the subscribers on the network. Like a self-addressed stamped envelope (I'm showing my age here), the continuous beam does most of the work for the terminals on the customer's end. A small silicon chip at the customer's terminal is all that's required to modulate the incoming beam and send it back out again, carrying data to the central office.

What does this mean for consumers? The network is cheaper overall to build, meaning future fiber-to-the-home service will be less expensive. And without a power-hungry laser at the customer's end, services will be more resistant to blackouts. Battery packs that power today's fiber terminals for 1-2 hours during power failures should be good for at least double that time using the new architecture.

Unless a way can be found to remove *all* lasers from a PON, I think this team has reached the logical endpoint of the laser reduction trend in fiber optic networks.

War: Shut Up and Multiply

A new study (abstract) finds that people stop balancing consequences when considering whether or not to go to war:

The prospect of crippling economic burdens and huge numbers of deaths doesn't necessarily sway people from their positions on whether going to war is the right or wrong choice. One possible explanation is that people are not weighing the pros and cons at all, but rather using a moral logic of "sacred values"—convictions that trump all other considerations—that cannot be quantified.

To try to capture people in the act of making such decisions, psychologist Jeremy Ginges of the New School for Social Research in New York City and anthropologist Scott Atran of École Normale Superieure in Paris challenged people around the world with a series of difficult questions.

They started by surveying 656 Israeli settlers in the West Bank. The researchers asked the settlers about the hypothetical dismantlement of their settlement as part of a peace agreement with Palestinians. Some subjects were asked about their willingness to engage in nonviolent protests, whereas others were asked about violence. Besides their willingness to violently resist eviction, the subjects rated how effective they thought the action would be and how morally right the decision was. If the settlers are making the decision rationally, in line with mainstream models, their willingness to engage in a particular form of protest should depend mostly on their estimation of its effectiveness. But if sacred values come into play, that calculus should be clouded.

When it came to nonviolent options such as picketing and blocking streets, the rational behavior model predicted settlers' decisions. But in deciding whether to engage in violence, the settlers defied the rational behavior models, the researchers report today in the Proceedings of the Royal Society B. Rather than how effective they thought violence would be in saving their homes, the settlers' willingness to engage in violent protest depended only on how morally correct they considered that option to be. (more)

As an avowed consequentialist, I find this behavior abhorrent. No matter how "right" something feels, we owe it to the real human beings that will be affected by our decision to make a rational decision. That is, we must do the math and then follow through on the results. If going to war results in tens of thousands of deaths, and avoiding war allows a tyrant to roam free killing hundreds every year; then war is obviously not the way to fix the problem.

A common argument against such analysis is that real world situations are messy and cannot be stated in mathematical terms. However, uncertainty is no excuse. Just because there are probabilities involved doesn't mean we should give up and start thinking in terms of "sacred morals".

Slightly Less Invasive Brain-Computer Interfaces

The EEG approach to brain-computer interfacing (BCI) consists of detecting the electrical emissions of groups of brain cells and interpreting the signals to control a machine. Traditional scalp-mounted EEG can do very basic controls because the electrodes are separated from the brain by about an inch of bone and flesh. This distance limits resolution, because the electrodes can only pick up on the largest groups of active brain cells. Scalp-mounted EEG devices have come to market, but they're not replacing keyboard and mouse any time soon.

Electrocorticography (ECoG) surmounts the resolution problem by placing the electrodes directly on the brain surface. However, this requires invasive surgery involving drilling through the skull and cutting open the brain's protective outer covering. Over time, the body reacts to the presence of foreign matter by building an organic sheath over the ECoG electrodes, degrading their performance.

Epidural electrocorticography (EECoG) is a new development in BCI, and it promises to deliver the advantages of ECoG with fewer drawbacks. In EECoG, the electrode array is placed beneath the skull but above the dura mater. This is a simpler procedure with less risk of infection or degradation over time.

A new primate study (abstract) has shown that EECoG can be used to control an on-screen cursor using electrodes placed only a millimeter apart. With electrodes of this density, the array itself can be smaller and capture more information than previously thought possible.

Daniel Moran, one of the authors of the study, has high hopes for EECoG:

"Eventually," he says, "we'll have a little piece of Saran Wrap with telemetry. We'll drill a small hole in the skull, pop the bone out, drop the device in, replace the bone, sew up the scalp and you'll have what amounts to Bluetooth in your head that translates your thoughts into actions."

Would you have a hole drilled in your skull if it meant never having to touch another keyboard or mouse ever again? Surgery has risks and complications can be costly and inconvenient. On the other hand, repetitive strain injuries cost the US $20 billion per year...

Source: physorg.com

Ergospheres!

Massive objects bend the fabric of space and time. Spinning massive objects distort space and time in another way, they drag spacetime around themselves. Almost like liquid in a blender, spacetime itself rotates around spinning masses, faster towards the mass and slower further away. 

This frame-dragging is now being quantified. In a 2004 experiment which I can only describe as rad, glass spheres, perfectly round to within 40 atoms, were cooled to near absolute zero, launched into orbit, and spun up to create an exquisitely sensitive set of gyroscopes. The intention was to monitor the direction of spin of the gyros, looking for the tiny distortions caused by frame-dragging. However, a small manufacturing error introduced noise into the system, all but drowning out the telltale signature. Data analysis continues in an attempt to salvage the mission.

You wouldn't need super-sensitive gyroscopes to detect frame dragging near a rotating black hole. A spinning black hole is so massive and so dense that it drags spacetime around itself faster than the (local) speed of light.

(This next bit is written with a somewhat shaky understanding of relativity, corrections are welcome.)

As an observer directly approaches a rotating black hole, frame-dragging becomes more and more intense. The spinning current of spacetime threatens to sweep the observer into an orbit around the black hole, requiring more and more thrust to counter the effect as the observer approaches. At a certain point, all the thrust in the universe is sufficient only to keep the observer from being swept along, she is no longer able to fly against the spinning vortex of spacetime no matter how hard she tries. At this point the observer has entered the ergosphere.

At the outer edge of the ergosphere, spacetime rotates past at the speed of light. A photon precisely counter-orbiting the black hole at this distance would remain motionless. Further in, even light cannot fight the current. Past the boundary of the ergosphere, it is impossible to remain stationary relative to the black hole. Doing so would require the observer to move faster than the local speed of light.

The ergosphere is particularly cool because our observer can experience all this physics weirdness and later tell her friends about it. Parts of the ergosphere lie outside the black hole's event horizon, meaning that one can swoop in and out again without getting caught in any embarrassing and awkward singularities.

ARgh

Route 66 and TomTom are collaborating on an Android app that gives turn-by-turn directions, with a twist. Their "Follow Me" mode generates a virtual AR car for you to follow, complete with blinkers so you know, in advance, when to turn:

Cool idea, no? Except their execution fails. Notice that the AR car is opaquely overlaid on the real-time video. This obstructs a significant portion of the road ahead. It's not a problem if you're driving with your eyes on the actual road, as you should. But fatigued or distracted drivers might focus on the app's live video instead, leading to dangerous situations. Most nav systems don't feature a live video feed, so there's little danger of a driver using only the system to perceive the road ahead. 

Developers of AR systems have a responsibility to their users: if an AR system gives you enough information (in this case a live video feed of the road), and is meant to be used in a possibly hazardous task (like when driving); then the system must not hinder your execution of the task.

Selective Perceptive Boosts Via Magnets

An article (abstract) published in Current Biology shows how visual perception can be selectively boosted in two different ways:

In the study, letters (H,S, and D) made up of smaller letters were displayed on computer monitors. Researchers asked participants to detect the presence of the letter "H" on either the global or local level. Participants who received transcranial magnetic stimulation (TMS) during the task showed significant improvements in detection times, compared to sham stimulation.

What's particularly interesting is this: the TMS frequency determined the type of perceptive boost. At 5 Hz, a frequency associated with theta brainwaves, participants were quicker to identify the global presence of the letter "H" when it was made of the letters "S" or "D". 20 Hz TMS bursts, similar in frequency to beta waves, improved performance in the more detail-oriented task of picking out the individual letters forming the whole.

Beyond the basic science aspect, this is a fascinating example of using technology to "change mental gears", as 'twere. I've often found that, after fiddling around with one or another detail of a project, it's difficult to return to a more integrative, global viewpoint. Programmers, artists, and engineers would all benefit from being able to switch modes back and forth at will.

I've already blogged about brain enhancement via transcranial direct current stimulation (tDCS). TMS is even less invasive than tDCS, although the equipment is bulkier and more expensive. I suspect tDCS will achieve mass commercialization faster than TMS for these reasons, although in the long term both may be used in tandem.

A (Hypothetical) Road To Autonomous Vehicles

Autonomous cars are a frequent sci-fi trope. Yet, the technology has matured to the point where Google's self-driving cars have safely logged over 140,000 miles with only occasional human intervention. In a sense, the capabilities of autonomous vehicles have outpaced our perception. What's standing in the way of widespread adoption?

We can safely assume the costs for the required processors, sensors, and servos will plummet in the coming years. Legal and regulatory hurdles will be cleared so long as large automakers have reason to throw lawyers and lobbyists at the problem. As I see it, the main issue is public opinion. Most folks seem unwilling to hand over the wheel to a computer, despite the potential safety and efficiency gains of self-driving cars.

But what about handing over the brake and throttle? Adaptive cruise control (ACC), a system where the car handles braking and acceleration based on inputs from radar or laser rangefinders, is offered in many cars today. Consumer confidence in ACC isn't terribly surprising. Braking inappropriately isn't usually dangerous. And decades of older, dumb cruise control have prepared drivers to hand over control of the throttle.

The next logical step after ACC is limited steering autonomy. For example, the Lexus LS offers a Lane-keeping Assist (LKA) option, wherein stereo cameras monitor the road, detect lane markers, and use warnings and slight steering torque to prevent accidental lane departure. One can imagine more advanced systems that detects cars and other obstacles, applying just enough steering to avoid accidents.

Integrated with ACC, I envision a mature system that makes it very, very difficult to get into collisions; even deliberately. By way of analogy, imagine trying to force together the positive ends of two magnets. The repulsive force increases as the magnets get closer. Similarly, the autonomous driving systems would become more assertive as necessary to avoid a collision.

This type of limited autonomy safety feature will likely arrive in advance of fully self-driving cars, and may ease the transition. It's even possible that with limited autonomy would come less attentive driving. The autonomous features would activate more frequently to avoid accidents, creating a kind of feedback loop of increased consumer confidence and more autonomous vehicles.

We may see a driverless future sooner than we think.

Placebo Power, Cont.

I've posted about placebos and nocibos before. Now, a study (abstract) published in Science Translational Medicine reports large placebo and nocibo effects, backed up with MRI scans:

The volunteers were placed in an MRI scanner and heat applied to the leg at a level where it begins to hurt – set so that each individual rated the pain at 70 on a scale of 1 to 100. An intravenous line for administration of a potent opioid drug for pain relief was also introduced.

After an initial control run, unknown to the participants, the team started giving the drug to see what effects there would be in the absence of any knowledge or expectation of treatment. The average initial pain rating of 66 went down to 55.

The volunteers were then told that the drug would start being administered, although no change was actually made and they continued receiving the opioid at the same dose. The average pain ratings dropped further to 39.

Finally, the volunteers were led to believe the drug had been stopped and cautioned that there may be a possible increase in pain. Again, the drug was still being administered in the same way with no change. Their pain intensity increased to 64. That is, the pain was as great as in the absence of any pain relief at the beginning of the experiment.

The researchers used brain imaging to confirm the participants’ reports of pain relief. MRI scans showed that the brain’s pain networks responded to different extents according to the volunteers’ expectations at each stage, and matching their reports of pain.

This showed the volunteers really did experience different levels of pain when their expectations were changed, although the administration of pain relief remained constant.

Notice that the placebo effect reduced the participant's pain ratings more than the actual drug! This also caught my attention:

On the basis of subjective and objective evidence, we contend that an individual’s expectation of a drug’s effect critically influences its therapeutic efficacy and that regulatory brain mechanisms differ as a function of expectancy. We propose that it may be necessary to integrate patients’ beliefs and expectations into drug treatment regimes alongside traditional considerations in order to optimize treatment outcomes.

I speculate that clinicians may be conflicted between a desire to maximize treatment outcomes for their patients and a compulsion towards honesty. A clinician trained in managing beliefs and expectations might exaggerate the effects of a pain-relieving drug, knowing full well that the exaggeration itself will increase the drug's efficacy. In a sense, the prophecy could become self-fulfilling.

Source: Physorg.com

Don't Reject When Deciding Between Two Alternatives

When deciding between two alternatives, one might choose by focusing on the more suitable aspects of the better alternative. Or one might reject by focusing on the less suitable aspects of the poorer alternative. An article (PDF) published in the Journal of Consumer Research finds that rejecting actually increases the chances of making the wrong decision:

We propose that consumers’ decisions are determined by how much they elaborate on information associated with their preferences. The extent to which consumers elaborate on information that is aligned with their preferences depends on whether they choose or reject alternatives. Choosing encourages elaboration on information that is consistent with one’s preference, which leads to preference-consistent decisions. Rejecting encourages elaboration on information that is inconsistent with one’s preference, which leads to preference-inconsistent decisions. For example, if a consumer is looking for a phone for business, a choice task would lead her to focus mostly on business features and purchase, for instance, a Blackberry. A rejection task, however, would lead her to elaborate on entertainment features and purchase, for instance, an iPhone. This is especially true in purchase contexts in which consumers have a baseline preference (business) but also want to serve other interests (entertainment) with the purchase.

Generalization is difficult because the studies all involved choosing between two and only two alternatives. But even in larger sets of alternatives, decisions often come down to two options. In these situations, rejecting an alternative leaves you open to having your preferences hijacked.

Building A Better Internal Combustion Engine

Your average car engine uses the four-stroke cycle:

1. Intake: Starting near the top of the cylinder, the piston travels downward, drawing in a mixture of air and fuel.
2. Compression: The piston reaches the bottom of the cycle and starts moving upwards, compressing the air/fuel mixture.
3. Power: As the piston nears the top of the cylinder, the spark plug ignites the air/fuel mixture. This causes an increase in pressure that presses the piston downward again.
4. Exhaust: By the time the piston reaches the bottom of it's stroke, all the fuel has combusted. When the piston reverses direction it forces the exhaust products out of the engine and the cycle starts again.

This means that each cylinder is pulling double duty. A cylinder will use one reciprocation to intake and compress the air and fuel, and the next reciprocation to ignite and exhaust. A split-cycle engine shares this load between two cylinders: one for intake and compression, one for power and exhaust. 

I first heard of Scuderi's split-cycle engine way back in 2006. At that time, they just had some ideas and computer models. Now they've released some more information, including extended testing with a prototype engine. Based on this prototype, they project fuel efficiency improvements of up to 36 percent over standard gasoline engines.

There's a significant improvement in power density as well. The prototype is a 1 liter, 2 cylinder engine producing 135 horsepower at 6,000 RPM. To put that in perspective, my car has a 2 liter, 4 cylinder engine and, despite the doubling in cylinders and volume, only generates 5 more horsepower.

I won't delve too deep here, but Scuderi also have plans to build a regenerative braking system into their engines. The power/exhaust piston is shut off while the vehicle brakes, and this basically makes the engine into a fancy air compressor. The compressed air is stored in a tank, and when the car takes off again the intake/compression piston is shut off while the engine runs off its reserve of compressed air.

I know, I know. Internal combustion is so last century! But there's too much technological inertia, and too much oil in the ground, to make a clean break. Current hybrid automobiles try to bridge this divide, but their power-trains are heavy and complex. They have to carry a complete internal combustion engine, plus batteries and an electric motor. The Scuderi design is a better compromise because it is simpler, lighter, and ultimately cheaper.

"Illusionary" Benefits

The size-weight illusion refers to our tendency to incorrectly perceive the weight of two objects of equal mass but different sizes. Even if we are aware of the equal weight of both objects, we tend to judge the larger object as lighter. An article (abstract) published in the journal Evolution and Human Behavior suggests that the illusion is functional:

Here we show that perception of heaviness (including the illusion) and perception of optimal objects for throwing are equivalent. Thus, the illusion is functional, not a misperception: optimal objects for throwing are picked as having a particular heaviness. The best heaviness is learned while acquiring throwing skill.

This immediately reminded me of another functional illusion: Robin Hanson's Homo Hypocritus hypothesis. Robin proposes that our big brains evolved in response to social pressures, specifically to bend rules in our favor. Self-deception is central to being perceived as honest, even when breaking or bending the rules:

In a messy real world, social norms expressed in language typically have many iffy boundary cases and ambiguities. How much of what sort of food of what quality offered how conveniently counts as food sharing? How big a frown is a grimace? Sex with how close a relative counts as incest? And so on. This wouldn’t matter if boundary cases were decided randomly, but that seems unlikely. Instead big brain gains come five ways:

Unnormed – coalition politics on acts uncovered by norms.
Skirt – keep actions near but not over edge of violating norms.
Cover – politics of observers on if to report an act to others.
Frame – lawyer-like arguing on if acts violate social norms.
Conspire – form coalitions on how to publicly interpet iffy acts.

Most norms have meta-norms against consciously trying to evade them. Self-deception should help here; foragers might sincerely believe they usually just do their job and “tell it like it is”, and then unconsciously try to act, selectively report and frame acts, and support interpretation coalitions, to their advantage. Instead of “man the tool user”, we might be better understood as “man the sly rule bender.”

The authors of the size-weight illusion article speculate that the advantage granted by the illusion may have helped us out-compete the Neanderthal:

The human ancestor with the larger brain was always the one to succeed and survive with the sole exception of Neanderthal, whose brain was actually larger than that of homo sapiens. Why did homo sapiens win out despite the smaller brain size? Evidence now reveals structural differences, namely, homo sapiens had relatively enlarged cerebellum and posterior parietal cortex as compared to Neanderthal. Perhaps our throwing abilities were the key.

I wonder, what other illusions have benefited us in our rise to biological dominance? And I wonder at the ultimate cost we may pay for allowing some of our illusions to run amok.

Hiring Bias

If a firm were to take exceptional measures to guard against biases in their hiring procedure, what might they do? Data from several studies suggests they could start by removing names and images on all incoming resumes prior to review.

A 1999 study (abstract) of 238 academic psychologists found that otherwise identical CVs with male names were more likely to be considered hirable than those with female names. Four CVs were generated based on an actual scientist's career. Two were those of an applicant looking for a job after graduate school, and two were more experienced CVs aimed at getting tenure. In both pairs, the only difference was the name on the CV: either "Karen Miller" or "Brian Miller".

A 2003 study (PDF) of Chicago and Boston employers found a significant bias against resumes from applicants with African-American sounding names:

To measure whether there is racial discrimination in the labor market, we send fictitious resumes in response to help-wanted ads in Chicago and Boston newspapers. We manipulate perceived race by randomly assigning to the resumes either a very African American sounding name or a very White sounding name. Employers are 50 percent more likely to call back resumes with White names for interviews. Moreover, we find that the returns to better credentials differ significantly by race. For White names, higher quality resumes elicit 30 percent more callbacks. For African Americans, however, higher quality resumes elicit a far smaller increase in callbacks. Applicants living in better neighborhoods are also more likely to receive callbacks but, interestingly, this effect does not differ by race. The amount of discrimination is uniform across occupations and, to a lesser extent, industries. Federal contractors and employers who list “Equal Opportunity Employer” in their ad discriminate as much as other employers.

And finally, a 2010 study (abstract) finds that men are more likely to be called back if they include a photo of themselves in their CV, while women with photos are less likely to get a callback:

To put our results in perspective, a plain male needs to send over twice as many CVs as an attractive male for an equal chance at a callback. This result is robust across industries and job and employer characteristics and ought to encourage attractive males to attach a photograph to their resumes in cultures like Israel in which the inclusion of a picture is left up to the applicant. On the other hand, attractive and plain women alike are better off omitting their photographs from their CVs since their inclusion decreases their chances of a callback by 20% to 30%.

With all this evidence of bias in hiring, why don't more HR departments remove the names and images from applicants' resumes? I'd take this a further step and suggest that interviews should be conducted via instant messaging when possible.

An OKCupid Datum

OKCupid's data mining blog OkTrends continues to generate fun and interesting correlations. Of the religious denominations in the data set, the least serious believers were more likely to have higher writing proficiency. Although the differences in atheists and agnostics are probably not significant, it's still amusing that the most serious nonbelievers score higher than the less serious.

HT: Boing Boing

Sinking the Sunk Cost Fallacy

Sunk costs can be thought of as stuff you've already paid for and cannot be sold or traded. Many investments, whether of time or money, are sunk costs. Rationally, we shouldn't let sunk costs affect our decision making. If A is strictly better than B, we should choose A regardless of how much unrecoverable cost we have incurred on B. The sunk cost fallacy refers to the human tendency to continue poor investments, that is, to choose B instead. A study (press release) by Daniel C. Molden and Chin Ming Hu shows that the sunk cost fallacy can be reduced by priming:

Volunteers were first asked to write generally about either their personal duties and obligations or their personal hopes and aspirations. Then they were asked to imagine that they were the president of an aviation company who had committed $10 million to building a special kind of airplane. But with $9 million already spent and the project nearly done, another company had announced they had made a better, cheaper plane. They were asked whether they would spend the last $1 million or cancel the project.

People who had thought about their hopes and aspirations were more likely to say they would abandon the project. On the other hand, people who had written about their duties and obligations felt it was their responsibility to finish the project—thus throwing good money after bad. Molden suggests that this is because people who focus on their own hopes are thinking positively about opportunities for growth, and are therefore more likely to appreciate what else might be accomplished with the remaining $1 million, whereas people who focus on duties and obligations are feeling anxious and thinking about how abandoning the project would be accepting complete failure.

My interpretation: putting volunteers into far mode (hopes and aspirations) mitigated the sunk cost fallacy while near mode (duties and obligations) exacerbated it. I doubt these results can be generalized, however. Imagine Alice has a gym membership that she never uses. I don't see how priming Alice with hopes and aspirations would make her more likely to drop the unproductive membership.

Which is not to say that more creative solutions aren't possible. (HT: Tyler Cowen)

On Procrastination And Perfectionism

People may procrastinate for any number of reasons. For me, the explanation that resonates most deeply has deep ties to perfectionism. From Katja Grace's blog:

The link between procrastination and perfectionism has to do with construal level theory. When you picture getting started straight away the close temporal distance puts you in near mode, where you see all the detailed impediments to doing a perfect job. When you think of doing the task in the future some time, trade-offs and barriers vanish and the glorious final goal becomes more vivid. So it always seems like you will do a great job in the future, whereas right now progress is depressingly slow and complicated. This makes doing it in the future seem all the more of a good option if you are obsessed with perfection.

That "depressingly slow and complicated" progress is made worse if, like me, you're hypercritical. John Siracusa coined the term in a 2009 Ars Technica piece to describe an extreme form of perfectionism:

Knowing what's wrong with something (or thinking that you do, which, for the purposes of this discussion, should be considered the same thing) does a fat lot of good if you lack the skills to correct it. And thinking that you know what's wrong with everything requires significant impulse control if you want to avoid pissing off everyone you meet.

But much worse than that, it means that everything you ever create appears to you as an accumulation of defeats. "Here's where I gave up trying to get that part right and moved on to the next part." Because at every turn, it's apparent to you exactly how poorly executed your work-in-progress is, and how far short it will inevitably fall when completed. But surrender you must, at each step of the process, because the alternative is to never complete anything—or to never start at all.

Creative work is painful for me. In straightforward tasks, the space of all possible outcomes is tightly constrained. For example, cleaning the kitchen counter is completed when all non-permanent blemishes are gone. But creative work has a much larger possibility space. For any piece of creative output I can find fault, and for many of these faults I cannot provide an immediate fix. Some I cannot even concretely define.

Managing this sort of debilitating perfectionism is what separates the successful from the chronically unfulfilled. John Siracusa cites Steve Jobs as an example of a successful hypercritical thinker. There is an ocean of capability between someone like Jobs and myself, and crossing that ocean will be a continuous process.

For now, there are two complementary goals: minimize procrastination and manage perfectionism. Breaking these up into actionable chunks will be a task for future blog posts.

Surprising No One, Mixed-Sex Interactions Make Us Dumber

A 2009 article published in the Journal of Experimental Social Psychology (abstract) found significant impairment in men's cognitive function after they had a neutral, five minute long conversation with a woman:

Y-axis is response times for a cognitively demanding task. Lower is better.

A caveat: the participants were in late adolescence. The authors of the paper acknowledge that the effect probably decreases with age.

Women aren't immune to mixed-sex cognitive impairment, either. A new study published in Psychology of Women Quarterly shows how an "objectifying gaze" by men can lower women's math performance:

As hypothesized, the objectifying gaze caused decrements in women’s math performance but not men’s. Interestingly, the objectifying gaze also increased women’s, but not men’s, motivation to engage in subsequent interactions with their partner. Finally, the objectifying gaze did not influence body surveillance, body shame, or body dissatisfaction for women or men.

Since the circumstance tested in this paper (an objectifying gaze) is less common than that of the first paper (a five minute long conversation), we can tentatively conclude that women are less susceptible than men to mixed-sex cognitive impairment.

When To Guard Against Optimism Bias?

Research subjects enrolled in cancer treatment trials are overconfident, according to an article (abstract) published in IRB: Ethics & Human Research:

Participants in our study were asked to compare their own chances of experiencing a range of risks and benefits related to the trial they were enrolled in with the chances of the other trial participants. We found a significant optimistic bias in their responses. Respondents tended to overestimate the benefits of the trial they were enrolled in and underestimate its risks. In addition, we found no significant relationship between respondents’ understanding of the trial’s purpose and how susceptible they were to unrealistic optimism. Our findings suggest that improving the consent process for oncology studies requires more than addressing deficits in understanding.

In other words, even well-informed respondents were overly optimistic about the benefits and risks of experimental treatments. But optimism bias has much larger effects than this. When we buy insurance, we estimate how likely we are to have an accident or contract a disease. When we choose a romantic partner, we estimate how happy and fruitful the relationship will be. Positive outcomes are imagined as more likely and negative outcomes are discounted. (see also: the affect heuristic)

The authors suggest that optimism bias should be taken into account when research subjects give informed consent. Why apply this reasoning in so narrow a context? If compensating for optimism bias is good in a limited scope like medical research, why is it not also good for, say, choosing which insurance plan to buy?

Many firms sell products that are based on strongly affective future events. Should they be required to follow an anti-optimism protocol?

Are College Students Learning Less?

Followup to: The Dangers of Overspecialization, Raising the Sanity Waterline: College Mostly Fails

For their book Academically Adrift: Limited Learning on College Campuses, Richard Arum and Josipa Roksa conducted analysis for 2,300 undergraduates at 24 colleges. They found that 45% of them demonstrated no significant improvement in critical thinking, complex reasoning, or writing in their first 2 years of college. I haven't read the book, but based on the blurb, their conclusion is that less learning is occurring now than in the past.

This jives with my personal opinion. A working paper (PDF) published by the National Bureau of Economic Research found that:

Full-time students allocated 40 hours per week toward class and studying in 1961, whereas by 2004 they were investing about 26 to 28 hours per week.

But, I've blogged before about the increasingly specialized education our students seem to experience. Can the poor results Arum and Roska find in critical thinking and complex reasoning be attributed to less learning, increased specialization, both, or neither?

My intuition says both, but I've no idea as to the proportions.

The Planetary Cobweb of Collaboration

Click to enlarge

Olivier H. Beauchesne mined data from scientific journal aggregators to make this beautiful map. The lines represent collaboration between researchers. The lines' brightness is determined by the logarithm of the number of collaborators and the logarithm of the distance between cities.

Better Creativity Through Electricity

This, from a new paper published in PLoS ONE:

Our experiences can blind us. Once we have learned to solve problems by one method, we often have difficulties in generating solutions involving a different kind of insight. Yet there is evidence that people with brain lesions are sometimes more resistant to this so-called mental set effect. This inspired us to investigate whether the mental set effect can be reduced by non-invasive brain stimulation. 60 healthy right-handed participants were asked to take an insight problem solving task while receiving transcranial direct current stimulation (tDCS) to the anterior temporal lobes (ATL). Only 20% of participants solved an insight problem with sham stimulation (control), whereas 3 times as many participants did so (p = 0.011) with cathodal stimulation (decreased excitability) of the left ATL together with anodal stimulation (increased excitability) of the right ATL.

This is exciting research. The team applied a very low current of 1.6 milliamps to the participant's scalps (your normal AAA battery can put out 1.6 milliamps for around 700 hours, though at a lower voltage). The results are impressive, showing a threefold improvement on a test of creativity and insight.

If this can be replicated, I think it's only a matter of time before some enterprising company starts selling tDCS for healthy users. College students are already using powerful drugs like Modafinil and Adderall to optimize performance, and these are legally restricted substances with significant side-effects. tDCS by comparison has almost no side effects and isn't as regulated. A 2006 paper (gated link) found headache and nausea to be the most common side effects, both were reported infrequently. As yet there's no research on the long-term effects of recurring tDCS use, though I expect this will be remedied shortly after products go on the market.

The products I'm envisioning are something like Emotiv's EPOC. The EPOC is a wireless EEG that can be used as a computer input device. Unlike hospital EEGs, the EPOC is designed to be worn for long periods. I expect the same design constraints will inform the first generation of at-home tDCS equipment.

Imagine: you sit down at your computer, put on your tDCS headset, and fire up Photoshop. The software that came with the headset has pre-set triggers that you've adjusted to your desired usage, so when you start up a graphics editing program it knows to watch for certain behavior patterns. You're chugging along, some time passes, but eventually you get stuck. You keep going down the same avenues, trying and retrying similar approaches that just aren't making this image look right. Your frequent undo and redo commands trigger the tDCS control software and it sends a signal to your headset. You feel a slight tingle on your scalp.

This is not to say that tDCS is a cure-all for writer's block and its many cousins. Instead, I think tDCS will join caffeine as a mild, socially acceptable nootropic.

Worlds Enough and Time

The results from NASA's planet-hunting Kepler spacecraft have been all over the blogosphere today. For me, the most impressive numbers were in the first paragraph of the paper (PDF):

Our current best estimates of the intrinsic frequencies of planetary candidates, after correcting for geometric and sensitivity biases, are 6% for Earth-size candidates, 7% for super-Earth size candidates, 17% for Neptune-size candidates, and 4% for Jupiter-size candidates.

These are early findings and I expect they'll change quite a lot as more observations are made. Still,

whoa.

Metacognition in Academia

Metacognition, or thinking about one's own thoughts or knowledge, seems to me an obvious foundational element of education. A study published last year (gated paper) found that students with higher metacognitive skills performed better at tests than their peers. So, it seems odd that there is so little emphasis on metacognition at the university level. Scott Jaschik writes in Inside Higher Ed:

Karl Wirth, professor of geology at Macalester College, said that metacognition requires students both to understand how they are learning and to develop the ability to make plans, to monitor progress and to make adjustments.

So one technique used is to distribute "knowledge surveys" to classes, in which students are asked whether they know the answers to questions (with the ability to say that they know half answers in some cases). He demonstrated by giving the audience a survey on metacognition with various prompts such as "What is metacognition?" and "Describe several metacognitive activities that can be used to improve student learning."

Wirth said that by using these surveys (in which students estimate their knowledge, separately from taking the actual test), students learn to accurately predict how much knowledge they have for an upcoming test or requirement. Over time, students become accurate with their predictions, and those who are predicting low performance are thus challenged to tackle whatever learning or studying issues are holding them back.

These are good, elementary first steps. Then I was blown away:

John Ottenhoff, vice president of the Associated Colleges of the Midwest, said it was too early to use the results to suggest any broad plan for reform of teaching. But he said that the program in just two years has generated "lots of practical ideas and some signs of improvement" in student learning.

Further, he noted that many professors worry that any emphasis on study skills will come at the expense of content, which many faculty members feel they lack enough time for as it is.

Many of these exercises amount to very short periods of time at various times in the semester, not so much time as to "sacrifice" any content, "and not burdening a professor."

So, not only is there virtually no existing recognition of metacognitive skills; there is actual debate as to whether basic metacognitive training is worthwhile? I'm flabbergasted.

Android App for Basic Mesh Networking

Followup to: Egypt and Mesh Networks

Ars Technica reports on an Android app currently under development that uses the built-in WiFi on Android smartphones to do basic mesh networking:

The system, named Serval, can relay VoIP calls between phones using their WiFi networking. Individual phones can also act as relay points, and theoretically should be able to bridge together a phone in a remote area with no service to one with access to the cellular network, where the call can finally be relayed to its intended recipient.

If the system can do VoIP, then it can host other IP services, like a VPN. Good news indeed for anyone living under a regime that controls access to the internet.