Messaging Extraterrestrials: A Waste of Time and Hope

Humanity has an awful track record of attempts to communicate with hypothetical alien civilizations. Leaving aside the publicity stunts, what few deliberate messages we've beamed at the heavens have been halfhearted at best. Like the famous Arecibo message, which assumes a human-like visual cortex must be present in beings that don't even share our evolutionary heritage.

A new paper (PDF) submitted to the journal Space Policy lays out a framework for constructing better messages. For example, the authors suggest sending proposed messages to different cultures, in order to weed out culturally specific information. The approach outlined in the paper may or may not have merit, but the underlying assumption is fundamentally flawed.

We are in a very small technological window where we can send and receive information encoded in electromagnetic waves, but cannot build atomically precise machines. If we could, then sending radio or laser messages would be far less efficient than sending tiny, self-replicating probes at near-lightspeed. Packed with information at atomic density, such probes could carry with them a significant portion of our cultural identity. More importantly, they could carry intelligences to act as ambassadors. The advantage of sending out actual emissaries as opposed to pure information is huge on astronomical scales, where even light takes years to travel between the closest stars.

It's probable that our small technological window exists for other civilizations as well. So, even if we assume a benevolent, communicative alien civilization exists close to Earth (close in astronomical terms, <100 light years); it's still unlikely our messages will do any good. We must further assume that the alien civilization is in the same 100 to 200 year technological window as we. For us, that's hundreds of years out of the millions of years it's taken to evolve modern human beings.

Should we send messages to hypothetical alien civilizations? If we just want to signal our far values, then sure. But if we value the act of communication itself, we should only send messages if they stand a reasonable chance of facilitating understanding between us and another civilization. 

Sending messages to the stars is like buying a lottery ticket. We expend some small quantity of resources to make token attempts at winning an unknown (but presumably large) jackpot at very long odds. The resources we expend are wasted. And like lotteries, our efforts are also a waste of hope.

HT: George Dvorsky

Egypt and Mesh Networks

The Egyptian government's recent move to disconnect the country from the internet brought to mind a similar scenario from science fiction. In Greg Egan's novel Zendegi, the protagonist is a reporter in a near-future Iran besieged by civil unrest. As in Egypt, the government of Iran pulls the internet plug. However, the protestors have futuristic cellphones that can form ad-hoc networks with each other, frustrating the government's information blockade.

This type of capability is called mesh networking. Mesh capable products have been around for years, but as yet there is no significant deployment of the technology on the personal level. Technical and regulatory issues aside, I suspect the delay in rolling out mesh enabled phones is lack of demand.

As yet, there's little demand from big telco. Big telecommunications companies tend to exert heavy-handed control over the ways their networks can be used. AT&T, for example, won't allow me to use the basic internet sharing capabilities of my iPhone without paying an additional monthly fee. What is pricing model for a customer that accesses the internet through a chain of mesh networked phones?

There's little demand from the public, either. A robust mesh network is the sort of feature that's infinitely handy when it's needed, e.g., accidental or deliberate service outages. It's not a feature that sells phones, though. There's also the chicken-and-egg problem of adoption, and the difficulty of simplifying the tech for mass consumption.

Mesh networking will eventually arrive on our phones, sooner or later. A big, innovative company like Google or Apple could smash through these barriers and deliver a mesh-enabled phone in the next few years. Or it could slowly trickle down from high-end phones to mass adoption, like Bluetooth. Either way, in the near future governments will have to resort to more sophisticated measures to disconnect their citizens from the internet than just flipping a few switches.

Tiny Pico Projector

As I mentioned in my post about augmented reality display tech, pico projectors have impressive capabilities. And they keep getting smaller:

Marcel Sieler of the Fraunhofer Institute has developed a prototype pico projector that is ridiculously tiny. It uses an array of 45 microscopic lenses, each driven by their own LCD, to stitch together an image that is both brighter and higher resolution than most pico projectors on the market.

The light source for the projector isn't included in the above image and would likely be the bulkiest component of the fully assembled device. However, the light source needn't be located directly behind this tiny module. A wearable system built into a shirt might place the LED lamps in an out-of-the-way location, like the small of the back. Fiber optics woven into the fabric would then direct the light to tiny modules like the one above, located on the shoulders or chest. The system could be flexible, lightweight, and nearly invisible.

A system like this could also be built into accessories, like belts. Bring on the projector pants!

Distributed Emergency Response

The San Ramon Valley fire department has released an iPhone app that allows users to act as impromptu EMTs. From Tim O'Rielly's report:

It was developed after the San Ramon fire chief had the horrendous experience of sitting at lunch with his team (including a paramedic) only to discover that next door someone had been dying of a heart attack. He didn't learn of it till the fire truck pulled up, siren blaring. He realized the need for an alert system that would reach out to the mobile phone of anyone trained in CPR who is nearby. Hence this app. Now, when an ambulance or fire truck is dispatched to help a heart attack victim, the same dispatch is sent to the app, which checks to see if the user is nearby to the person in crisis. In addition, the app shows the location of the nearest AED (defibrillator).

According to the American Heart Association, the first 3-5 minutes after ventricular fibrillation are critical for long-term survival. Also consider that heart disease is the leading cause of death in the United States.

One can easily imagine a more evolved version of this app becoming commonplace. If the app knew about the user's capabilities, it could preferentially dispatch more qualified personnel close to the emergency. Furthermore, it could provide video and audio instruction tailored to the user's skill level. Dispatchers, paramedics, or even physicians could be video conferenced through the app to provide diagnosis or help until the ambulance arrives.

Why not use such an app for other emergencies? Stroke is another top killer in the US. Clot-busting drugs can dramatically improve outcomes for stroke victims, but only if administered quickly. Is it possible that autoinjectors loaded with this drug could be included in emergency AED boxes?

One problem blocking such a wide deployment of powerful drugs is potential for misuse, but an app could alleviate this problem. If the autoinjector were electronically locked, it could use short-range wireless to establish a link to the smartphone running the app. From there, the video conferencing capabilities could allow remote diagnosis of stroke and the physician would unlock the auto-injector with an encrypted signal.

As smart devices diffuse into the general population, there will be opportunities for defense-in-depth. Emergency response is only the tip of the iceberg.

The Dangers of Overspecialization

This op-ed in the Washington Post lends (anecdotal) support to my position that today's university students are too specialized:

An outstanding biochemistry major wants to be a doctor and supports the president's health-care bill but doesn't really know why. A student who started a chapter of Global Zero at his university hasn't really thought about whether a world in which great powers have divested themselves of nuclear weapons would be more stable or less so, or whether nuclear deterrence can ever be moral. A young service academy cadet who is likely to be serving in a war zone within the year believes there are things worth dying for but doesn't seem to have thought much about what is worth killing for. A student who wants to study comparative government doesn't seem to know much about the important features and limitations of America's Constitution.

When asked what are the important things for a leader to be able to do, one young applicant described some techniques and personal characteristics to manage a group and get a job done. Nowhere in her answer did she give any hint of understanding that leaders decide what job should be done. Leaders set agendas.

I wish I could say that this is a single, anomalous group of students, but the trend is unmistakable. Our great universities seem to have redefined what it means to be an exceptional student. They are producing top students who have given very little thought to matters beyond their impressive grasp of an intense area of study.

Robin raises an important question:

Just how many people should be thinking about what problems are important and why?

This is a good question, and worthy of debate and study. But considering the dangers we're facing as a species, I think it's better to err on the side of excessive generalization than excessive specialization.

Too much generalized education might slow down the pace of innovation and thus slow down the economy. Increased interdisciplinary studies would lengthen the already considerable ramp-up time from college freshman to full fledged innovator. Overspecialization, on the other hand, may have much worse effects.

There are already precious few people and resources committed to the study of existential risks. The trend of increased specialization threatens to weaken this already thin safety net. I've no doubt that specialization will be required to mitigate existential risks. But to get involved in such a minimally profitable field in the first place requires a certain amount of "big-picture" thinking. I'd feel much safer in a world with too many big thinkers than too few.

Why Do Worms Suck?

As computer worms go, Conficker was a beast. First detected in 2008, at it's peak it infected at least seven million computers worldwide. Millions of systems remain infected, today. Given the worm's success, this passage from the Conficker Working Group's new report really stands out:

[...] it does come up in questions of why Conficker was never used for anything more devious than scareware. It is likely that the Conficker Working Group effort to counter the spread did make it more difficult for the author to act with impunity, but the author did not seem to have tried his or her hardest.

Given that the most widespread computer worm in history was the result of this level of effort, why are we not seeing more advanced and capable worms? For example, why not leverage the spare compute cycles of infected machines to help improve the capabilities of the malicious software? Fuzzing in particular seems scaleable in this way.

Fixing the Innovation Shortfall

As scientific and technical knowledge continues to accumulate it takes longer and longer for innovators to get up to speed. Ben Jones finds that overall, innovators are losing ground: (HT: Tyler Cowen)

I estimate shifts in life-cycle productivity and show that innovators have become especially unproductive at younger ages. Meanwhile, the later start to the career is not compensated for by increasing productivity beyond early middle age. I further show that the early life-cycle dynamics are closely related to variation in the age at Ph.D. and discuss a theory where accumulations of knowledge across generations lead innovators to seek more education over time. More generally, the results show that individual innnovators[sic] are productive over a narrowing span of their life cycle, a trend that reduces, other things equal, the aggregate output of innovators.

I see three points where this trend could be slowed or reversed:

• Reduce the ramp-up time for innovators, i.e., improved educational systems. Peter Thiel gets it.
• Increase productivity during peak years. 
• Extend the productive years into later life. Anti-aging research may help here.

Upcoming posts will address these points in more detail.

Weep Not for Earth

Heaven lasts long, and Earth abides.

-Lao Tzu

I occasionally hear statements of this sort: "We're trashing the planet", "We're ruining the environment", and "The Earth would be better off without us". Others have pointed out flaws with these positions, but I find it useful to remind myself just how much catastrophe the Earth has borne in it's 4.5 billion year lifetime. Here are just a few highlights:

• The Great Dying occurred 251.4 million years ago. New research suggests an unprecedented series of volcanic eruptions massively altered the environment. The climate changed severely, the oceans acidified, ash blocked out the sun. Magma contacted huge coal deposits in Siberia. What happened next may be similar to the disaster that befell Centralia, Pennsylvania; but on an astronomical scale. Coal, burning underground, may have put as much as 3 trillion tons of carbon into the atmosphere. To put that into perspective, in 2008 all the fuel burned worldwide released under 30 billion tons of carbon (source). 96% of all marine species went extinct, plus 70% of all land species.
• 65.5 million years ago the Cretaceous-Tertiary extinction event happened. The scientific consensus is that an asteroid at least 6 miles across impacted in what is now the Gulf of Mexico. Large parts of the planet were roasted by the 100,000,000 megaton equivalent energy release. Later, the Earth was shrouded in ash and dust, cooling and then starving what few organisms survived the impact. 75% of all species went extinct.
• More recently and on a smaller scale, a swarm of cometary debris may have impacted present day North America about 13,000 years ago. If there were people living in North America when this occurred, they would have witnessed the closest thing to nuclear holocaust ever seen by human eyes. Thousands of explosions, each comparable to a hydrogen bomb, would have occurred as the ice and rock fragments either impacted the ground or vaporized in the atmosphere. Ejecta from this disaster absorbed or reflected the sun's radiation and prolonged the last ice age.

It takes a lot to actually hurt our planet. Short of astronomical phenomena or advanced technology, Earth itself is quite safe. We, on the other hand, are not. It's our job to remember that distinction, so we know what it is we are fighting for.

Placebos and Nocibos

Ben Goldacre speaking at Nerdstock, via George Dvorsky:

Near the end of the video, Dr. Goldacre mentions a study of the muscle relaxant carisoprodol. Subjects were divided into six groups, three were given the drug and three received a placebo. Some groups were told the drug was a relaxant, others a stimulant, and the remainder were given no information about the drug. After administering the drug, blood concentration was measured for carisoprodol and meprobamate (its metabolite):

Consistent with the placebo effect, those that were told the drug was a relaxant had higher concentrations of the drug compared with subjects who received no information. Interestingly, the converse is also true. Subjects who were told they had taken a stimulant had lowered drug concentrations compared to the controls. In the paper this is called a nocibo effect; meaning the same placebo mechanism is at work but is causing undesired effects.

The placebo effect is still poorly understood. It's a fascinating subject for me. Is it possible to make a drug more effective simply by saying the right things to the patient? If so, perhaps the dosage could be lowered with a commensurate reduction of side effects. Or, could the side effects themselves be mitigated by priming the patient with counter-factual information? Would you prefer to be told lies at the hospital if it meant less severe side effects?

Avoiding the Focusing Illusion

Like correspondence bias, the focusing illusion is one of those quietly horrifying cognitive biases that influences us both deeply and broadly. It describes the human tendency to overemphasize how possible future circumstances will affect our day-to-day happiness. As Daniel Kahneman puts it:

Paraplegics are often unhappy, but they are not unhappy all the time because they spend most of the time experiencing and thinking about other things than their disability. When we think of what it is like to be a paraplegic, or blind, or a lottery winner, or a resident of California we focus on the distinctive aspects of each of these conditions. The mismatch in the allocation of attention between thinking about a life condition and actually living it is the cause of the focusing illusion.

For example, Kahneman, et al. find that:

Most people believe that they would be happier if they were richer, but survey evidence on subjective well-being is largely inconsistent with that belief.

Kahneman and Schkade (gated paper) lend more supporting evidence:

Large samples of students in the Midwest and in Southern California rated satisfaction with life overall as well as with various aspects of life, for either themselves or someone similar to themselves in one of the two regions. Self-reported overall life satisfaction was the same in both regions, but participants who rated a similar other expected Californians to be more satisfied than Midwesterners.

If you are at home, look around. Chances are, you've surrounded yourself with products that you thought would make you happier than they actually did. Not only that, but our relationships, jobs, and habits are all vulnerable to the focusing illusion. Avoiding or compensating for this bias is therefore crucial to achieving our goals, in other words, this is a part of instrumental rationality.

One possible way to nullify the focusing illusion is by taking the outside view. For example, when considering a purchase or change of lifestyle, do some research among others in similar circumstances. If a broad consensus emerges opposing your current opinion, its may be a good opportunity to practice changing your mind. The biggest problem with this approach may be an excess of low-quality information. Surveying your friends for their opinions invites selection bias. The internet is full of product reviews, but many are initial reactions and don't reflect the long-term effects on the purchaser's well-being.

The focusing illusion is strongly linked with the concept of a hedonic treadmill. Although events in our life may make us more or less happier for a time, we tend to to return to a "hedonic set point". For example, Brickman, et al. (gated paper) famously discovered that lottery winners and paraplegic accident victims tend to have the same day-to-day happiness levels, i.e., hedonic set points. This undercuts our expectations of how such life-changing events might influence our happiness, suggesting the focusing illusion is at work.

It's possible that folks with lower hedonic set points might be more vulnerable to the focusing illusion. If lower hedonic set points lead to an increase in happiness seeking behavior, the focusing illusion might be triggered more frequently. Perhaps increasing one's hedonic set point could lower their susceptibility to the bias? I've been unable to locate any articles on this approach (consider this an active solicitation).

I'll address one more way to avoid the focusing illusion. Kahneman points out that marketers and politicians excel at exploiting the focusing illusion to influence our behavior. Ignoring advertisements and political rhetoric is good general advice, of course. But being aware of the focusing bias may even the playing field.

Cheap, Clean Power From a Very Big Gun

TekLaunch (formerly QuickLaunch) plans to build a kilometer long light gas gun to cheaply launch payloads into orbit, but they don't have a customer yet. Space-based solar power could provide inexpensive, renewable energy anywhere in the world, but high launch costs make it impractical.

 Let's see if we can get these two attractive singles together.

Space-based solar power (SBSP) is a proposed system of orbiting solar panels to send power to ground stations. Earth's atmosphere blocks a significant amount of solar radiation from reaching the ground. In space, photovoltaic panels can receive over 140% more energy. Furthermore, there's almost always a clear line of sight to the sun, so SBSP satellites operate at peak efficiency nearly all the time. Satellites would beam power back to Earth using lasers or more likely microwaves. The terrestrial receiver stations would be cheap and simple compared to other power plants. They can be built in poor or remote areas of the world. And, to correct a common misconception(*cough* SimCity *cough*), the system is quite safe. Since the microwave beam would have to be over a kilometer wide, its average power intensity would be low. Even if the beam went off-course humans wouldn't be in danger.

As previously mentioned, the biggest problem with implementing SBSP is launch costs. SpaceX's Falcon 9 is the most efficient launcher in the world at $2430/lb to low earth orbit (LEO). A 1 gigawatt SBSP satellite might (optimistically) weight 1,000 metric tons, meaning it would cost over $5 billion to launch into orbit. The satellites themselves and the ground stations would add additional cost to the system. In comparison, the equivalent earthbound station would cost $4.8 billion (source) and have no additional costs. Launch costs must come down for SBSP to be workable.

The QuickLaunch gun. Hydrogen is heated in the lower right segment of the gun, the projectile exits from the upper left.

A light gas gun is quite a bit simpler than a rocket. Hydrogen or helium is pressurized in a sealed chamber attached to a long barrel (typically the barrel is evacuated). When a critical pressure is reached, a valve permits the hydrogen to flow into the barrel and out of the muzzle, pushing a projectile to high speeds in the process. Firing a projectile into orbit requires some added complexity, like a heat shield to protect the cargo while it exits the atmosphere and a small rocket engine to provide a final boost into orbit. QuickLaunch's clever twist on this design is placing nearly the entire barrel underwater. Since it's neutrally buoyant, the water acts to support the barrel equally along the entire length. Sea launch also permits the gun to be adjusted to different azimuths.

QuickLaunch's business plan is to launch rocket fuel and set up an orbital gas station. Their literature claims they'll sell fuel for $500/lb at LEO, but there's no reason why payloads must be limited to fuel. Anything that can withstand the very high G-forces from a gun launch can be put into orbit. This rules out astronauts or moving parts, but not solar panels. In fact, solar panels have been fired out of light gas guns before. The Super High Altitude Research Project tested photovoltaic panels at up to 3200 g's. And in this video, Dr. John Hunter of QuickLaunch says, "Photovoltaics are the easiest thing to launch because they're so slim."

At $500/lb, that optimistic 1 gigawatt SBSP satellite I mentioned would cost $1.1 billion to launch into orbit. Even if costs double or quadruple SBSP could compete with terrestrial solar power stations, thanks to it's unique advantages. QuickLaunch estimates 4 million lbs per year could be launched from one gun, meaning 1.8 gigawatts of power could be onlined every year.

Of course, some components of the SBSP satellites can't be launched from a gun. The structural elements, pointing motors, and thrusters would have to be launched conventionally, with rockets. On-orbit assembly of the solar arrays could be conducted with remotely operated robots, which would also have to be launched from rockets. 

The solar arrays themselves would have to be designed differently than for most applications because of the high G-forces at launch. Conventional satellite solar panels are rigid and fold together to fit into the rocket's payload fairing. This type of system would likely fail in a gun launch. Instead, flexible photovoltaic panels could be rolled up and stacked inside the launch vehicle, like this:

Not to scale. Orange section holds fuel and oxidizer for orbital insertion. White cylinders represent rolled flexible solar panels.

Breathless enthusiasm aside, there's huge engineering challenges awaiting anyone who tries mating two relatively untested technologies. I've made many assumptions here, for example I've assumed the majority of a SBSP satellite's mass is solar panels. Also, I should mention that I'm not the first to think of combining these two clearly complementary systems.

An Economics Post!

Click to enlarge

This chart is from the Stock Trader's Almanac. The Y axis is logarithmic. The gray boxes are supposed to represent periods of increased inflation. Note that all of these periods include a major war, brackets indicate the war's duration.

One interpretation of this chart is that we're due for a economic boom, to the tune of at least 500% DJIA. Does that fit with the data in the chart? The War on Terror period is enclosed in a gray box, but the Consumer Price Index doesn't show increased inflation relative to the boom-times of the 80's and 90's. If there is a pattern here, then either we are due for an inflationary period or we are in a new type of situation entirely.

First Impressions and Correspondance Bias

New research shows that it's possible to change people's first impressions, but only if they are presented with conflicting evidence in multiple contexts. (Gated paper here)

Although these results support the common observation that first impressions are notoriously persistent, Gawronski notes they can sometimes be changed. "What is necessary is for the first impression to be challenged in multiple different contexts. In that case, new experiences become decontextualized and the first impression will slowly lose its power. But, as long as a first impression is challenged only within the same context, you can do whatever you want. The first impression will dominate regardless of how often it is contradicted by new experiences."

This is good news, because it gives us another weapon in the fight against correspondance bias. Also known as the fundamental attribution error, correspondance bias is one of the most pernicious of the cognitive biases. To understand correspondance bias, try to remember an occasion when someone was being particularly offensive, rude, or stupid. Attempt to recall your thoughts at the time; how did you explain that person's behavior? Did you explain their behavior in terms of reasons, e.g., "Perhaps they're having a bad day"? Or did you simply write them off as an offensive, rude or stupid person.

Now try to remember an occasion where you did something particularly offensive, rude or stupid. Chances are good that you remember reasons (perhaps not particularly good reasons) for your bad behavior. Correspondance bias describes the excessive association between actions and personality traits when we explain others' behavior, and also our excessive disassociation when we explain our own behavior. It's actually a bit disturbing to realize: this bias colors nearly every interaction we have with each other!

So, how to counter this insidious force? Since first impressions and correspondance bias are tightly connected, reducing the former may simultaneously reduce the latter. According to the research, the key is to present behavior that contradicts the first impression in as many contexts as possible. This means, for example, if your boss has the first impression of you as a lazy person, working harder by itself may not eliminate the impression. Instead, consider showing off your productivity outside the workplace.

When fighting correspondance bias, the standard approach has been to recognize the bias when it occurs in our own minds. This new research give us an opportunity to fight correspondance bias in other minds, not just our own. Extremely handy, considering most folks aren't even aware that such a bias exists.

Further reading on correspondance bias can be found here.

Set a Password on Your Smartphone!

There's not much I can add to this Ars Technica article:

Last week, California's Supreme Court reached a controversial 5-2 decision in People v. Diaz (PDF), holding that police officers may lawfully search mobile phones found on arrested individuals' persons without first obtaining a search warrant. The court reasoned that mobile phones, like cigarette packs and wallets, fall under the search incident to arrest exception to the Fourth Amendment to the Constitution.

California's opinion in Diaz is the latest of several recent court rulings upholding warrantless searches of mobile phones incident to arrest. While this precedent is troubling for civil liberties, it's not a death knell for mobile phone privacy. If you follow a few basic guidelines, you can protect your mobile device from unreasonable search and seizure, even in the event of arrest.

If you have already pre-committed to not volunteering information to the police, then by extension you should at least set a password on your phone. If you haven't pre-committed, then watch this video:

Space is Far

This video has been making the rounds lately. It's a video montage with audio from Carl Sagan's Pale Blue Dot. I do love me some Sagan, but I've got a bone to pick with his message in the video. (And why the NASA plug? Has NASA done anything to significantly advance human space exploration since Apollo?)

Carl Sagan speaks beautifully about humanity's future in space. We'll have changed by the time we're ready to colonize other worlds, he says. The passage of generations and necessity will make us more far-seeing and prudent, among other things. He claims that our descendants will be very much like us, but with more of our strengths and fewer of our weaknesses. It's a hopeful and life-affirming message, but is it based on much more than wishes?

When we humans think about the future, we go into far mode. As Robin Hanson wrote:

In far mode we tend to focus more on our simple abstract ideals and values, relative to messy desires and practical constraints. We also tend to neglect our messy internal contradictions and conflicts, and therefore assume our values and actions are coherent and consistent. So in far mode we tend more to explain good acts as virtue, and bad acts as vice or evil. We assume future folk are less driven by base desires, more strongly committed to their ideals, less tolerant of domination, more morally enlightened, and more morally judgmental about others’ failings.

Far mode is a great way for us to convey our values without having to spend any actual resources. This extra signaling capability is accompanied by a commensurate loss in accuracy. We're less likely to be correct if we massage the data to come across as high-minded.

So, back to Sagan's message. It's possible that we'll become more far-seeing and prudent in the future. But does it follow from the data? "Necessity" could cause us to change in any number of ways. If our technological prowess becomes great enough, for example, we could very well become less far-seeing and prudent as those capabilities are offloaded to machine intelligences.

Is it probable that our star-faring descendants will be similar to us? This requires one of two things to happen: technological change must slow down or we must all agree (or be forced) to remain baseline human-like despite the capabilities to improve ourselves. Neither seems particularly likely. Even if most of humanity decided to forgo self-improvement technologies (uploading or augmentation), the minority that embrace new tech will become more capable and likely more numerous as a result.

I think Carl Sagan's view is limited by his desire to see contemporary humans expand into space. Our descendants may tread the stars, but they won't look or act much like us.

An Overview of Displays for Augmented Reality Applications, Pt. 2

Yesterday I described four existing display technologies and how well each measures up to criteria relevant for AR. Today I'll put on my more speculative hat and explore where these might be going.

One way to compensate for the weaknesses of any individual type of display is to use multiple displays, either one at a time, or in tandem. This type of hybrid system will likely dominate in the near future: as less obtrusive and more constantly available displays go to market consumers will be carrying around a flat panel display (their smartphone), as well as a newer type of display.

A head-up display or retinal image display complements a flat panel well. HUDs provide a private, always-on display while the user's smartphone is in her pocket. If glare washes out the HUD or if the user wishes to share imagery he can utilize the smartphone display. This is less than ideal because it requires the user to physically reposition her flat panel display when it's needed. Offsetting this disadvantage is a huge existing install base of smartphone displays, and well developed idioms for interacting with them.

Another pair of complementary display techs is the retinal imaging display combined with a pico projector. A forward projection system's main failing is lack of environmental robustness. Projected images are easily drowned out in bright lights, and dark colors reflect poorly. This is neatly complemented by the retinal display's relative insensitivity to local lighting. Relative to the flat panel/HUD hybrid described previously, this combination is more available. Although a projection display might require the user to reposition himself, local surfaces such as walls will often be available with no effort from the user.

There's yet another possibility aside from multiple display solutions. With continuing research and development, one type of display may manage to hit all four criteria (availability, unobtrusiveness, share-ability, and robustness). Projection systems in particular continue to make the most astounding advances. Real-time calibration and compensation may advance to the point that any odd-shaped surface can make a perfect projection screen. Problems with direct sunlight or glare washing out the picture can be mitigated with laser projectors. Of course, if such a system became too popular we might end up with some serious light pollution. Just imagine a train full of commuters, all using projected AR.

An Overview of Displays for Augmented Reality Applications, Pt. 1

Note: please forgive the somewhat esoteric nature of this post. This is a pet interest of mine and hopefully by the end you'll have an idea of how augmented reality displays might be of importance to you.

Augmented reality (AR) is likely the next step in our interactions with computers. AR refers to computer generated information that is overlaid or composited in with our real surroundings. As computers continue to shrink while becoming more powerful, we'll integrate them more with our lives. A large part of that integration will be new and better display technologies.

Ideally, we'd like our displays to have the following four properties:

• Available. For AR to work we need a display that is almost never inaccessible to the user. The user shouldn't have to move in order to see the display.
• Unobtrusive. If users are going to wear the display for hours at a time, it needs to not interfere with the user's senses or movement. It should also be comfortable to wear.
• Shareable. The should be able to be shared with other people in close physical proximity without compromising the owner's experience. This will become less important as display tech diffuses into the population. If everyone has a display then users can invoke consensual imagery to share.
• Robust. The display has to work in a wide range of conditions. It should be fully visible in any ambient light, from direct sunlight to total darkness.

So, how do existing and developing displays measure up to these ideals? The most common display is the emissive or reflective flat panel. From LCD's to AMOLED's to e-ink, they are all shareable and they are mostly robust. However, if your display is in your pocket or bag it is not available. And if it is available, whether in your hand or strapped to your wrist (or worse, your head), it is not very unobtrusive. They can be made less obtrusive, as in the EyeTap, which is a semi-transparent display incorporating a camera mounted at eye-level. It uses a beam splitter, which means that the unaided view is diminished somewhat.

The sort of head-up display most of us are familiar with is the type common in many commercial aircraft cockpits. A transparent panel sits in front of the pilot and a projector displays on to the panel, giving the appearance of a display hovering in front of the eye. Miniaturized versions of this type are currently in the prototype stage. Since they sit between the eye and the subject, HUDs are always available. If built into lightweight glasses, they are also unobtrusive. However, they can get washed out in bright light, and you won't be sharing your YouTube kitty videos with that cute girl/guy on the bus without taking off the display and handing it over.

Retinal imaging displays are a specialized type of head-up display that scans a laser beam across the user's retina, similar to the way the electron gun scans the phosphor screen of a CRT television. There are only two companies (that I know of) showing off prototype retinal displays: NEC and Brother. The user experience of a retinal display is said to be similar to a projection-type HUD, but with better contrast. This increases the robustness, plus the lack of a transparent panel makes retinal displays less obtrusive. They are just as available as projection HUDs and just as un-shareable.

The final type of display tech I'll evaluate is the pico-projector. This is probably the newest and most experimental display for AR applications, even though it is quite simple in theory. A tiny projector is worn somewhere on the body with a field of view co-incidental with the user. Information is displayed on a wall, the user's hands, or any other close surface. Operating out of the MIT Media Lab, SixthSense is the pioneer this field. As long as the pico-projector has a clear line of sight to a nearby surface this display is available. So, less available than the two HUDs but more available than a flat panel.

Providing pico-projectors continue to shrink, I'd consider this the most unobtrusive display. Already we're seeing projectors that can fit in a cell phone. It is likely that in the near future projector modules will be small enough to build into clothes or accessories without much added bulk. This display is, of course, shareable. Projected images wash out in bright light, however. This lack of robustness is exacerbated by the projected surface's unpredictable geometry. A fixed projection display might look fine when facing a white wall, but it would be an unreadable jumble when cast upon your hand or the portly gentleman in front of you in the Starbucks line.

So, I've examined four display technologies and they each have at least one flaw that keeps them from AR perfection. All is not lost, however! My next post will address a way forward for pico-projectors and the possibility of combined display systems.

The Quick And Easy Path

Elevating public discourse about new technology is hard. The idea of the Singularity has been popularized for over a decade and the popular press still gets stuck on the same links of the inferential chain. Robin Hanson suggests that it's better to avoid public discourse for disruptive new technology:

Typically, when a public debate begins decades in advance of a potential new tech, it becomes a far-minded symbolic battle ground, where folks express grand positions on family values, materialism, inequality, nationalism, etc. The net effect is usually to inhibit the useful application of such techs. In contrast, when a tech appears mostly out of the blue, people tend to focus on whether they’d actually like to use it now.

For example, the pill and the web were both largely unheralded, and were thus quickly adopted and integrated into our lives. But if folks had seen thirty years in advance how the pill would change sexual practices, or how easily folks would give up privacy for web access, such techs might have been blocked or more heavily regulated, to our detriment.

He's right on the money about the two examples given here. But there is a certain class of technology that is incredibly dangerous to avoid publicly discussing before it's developed. Hint: one example of it has come within a few momentary decisions of destroying most human life on Earth.

Nuclear weapons went from theoretical possibility to reality in 12 years. Most of the technological progression during that time was top secret, shielded from the public until the bombs were dropped on Japan. Would a robust public debate prior to the development of the atomic bomb have stopped or slowed the insanity of the Cold War? I doubt it. But we would at least have had a chance, albeit a small one, of avoiding a world that has come so close to accidental devastation.

Nuclear weapons are the first in a class of technology which has the capability to destroy us. There are more, some of which exist now and some which will exist in the near future. Among them are biotech, molecular nanotech and recursively self-improving intelligences. Avoiding public discourse about the risks of these technologies does not now seem to be a hard problem. Indeed, it seems to have been the default position of our recent history for several existential risks.

The Cold War infested our planet with over 65,000 nuclear weapons at its peak, and very nearly destroyed modern civilization. The next bout of global insanity will be faster and more dangerous as the technologies involved become more capable. Public debate, alone, prior to the development of new tech won't be sufficient to avert catastrophe. And I agree with Robin that the level of discourse has been miserable, historically. Raising public debate from empty rhetoric to rationality would be a huge project to be sure. To which I say:

Better start now.

For Women, More Controversial Looks = More Male Attention?

OkTrends has posted a statistical analysis which relates men's attractiveness rating of women to the number of messages women receive on their dating site:

The less-messaged woman was usually considered consistently attractive, while the more-messaged woman often created variation in male opinion.

I think this is fascinating info. There's a serious problem with the article, however. At various points, the author assumes that the number of messages a female receives on OKCupid correlates with:

• Male attention
• Liking the recipient of the message
• A better response from men

Furthermore, the conclusion of the article makes suggestions with an eye towards maximizing the number of messages a female will get. These assumptions indicate a linear correlation in the author's mind: more messages = more success at dating websites. Isn't it likely that the quality of the messages (not to mention the qualities of the sender) have a significant impact on success, even though these metrics are outside the scope of the article?

Since only two datasets were used in this analysis, attractiveness ratings and quantity of messages, we can only relate one to the other. If OkTrends had included data on the content of the messages, or satisfaction ratings from relationships that were initiated by the messages, then we could draw conclusions about dating website success.

It is possible that increased messages does correlate with increased success. But its also possible that certain women receive more, lower quality messages (perhaps from lower quality males), which might negatively correlate with success. This data doesn't point towards either conclusion.

Hat tip: Alex Tabarrok

Bored With Molecular Nanotechnology?

Me too! Evolution did the best it could when it built you and I out of molecular machines. But really, can't intelligence do better than blind evolution? Just as molecular nanotechnology builds one atom or molecule at a time, femtotechnology would build one nucleon or quark at a time. Two articles in H+ Magazine explore this highly speculative proto-technology:

If ever a femtotech comes into being, it will be a trillion trillion times more “performant” than nanotech, for the following obvious reason. In terms of component density, a femtoteched block of nucleons or quarks would be a million cubed times denser than a nanoteched block. Since the femtoteched components are a million times closer to each other than the nanoteched components, signals between them, traveling at the speed of light, would arrive a million times faster. The total performance per second of a unit volume of femtoteched matter would thus be a million^3 times a million = a million^4 = a trillion trillion = 10^24.

Heady stuff! Femtotech has been a pet interest of mine for some time. Greg Egan fueled my fascination with his novel Schild's Ladder. In it, the protagonist "goes femto", that is to say, her mind is computed in a femtotech substrate. However, femtotech envisioned in Schild's Ladder has significant drawbacks. The femtocomputer in the novel is as unstable as it is fast, and must essentially blow itself apart in the course of normal operation.

Instability seems to be the Achille's Heel of femtotech speculation. Machinery made out of nucleons would tend to collapse into a sphere, and very quickly. Ben Goertzel mentions dynamic stabilization as a possible solution. If we could apply correctly timed high-frequency vibrations to an unstable nucleon assembly it could maintain its shape, just like an inverted pendulum can be stabilized by oscillating it at the right frequency. As mentioned in the article, there are many problems with dynamically stabilizing femto-scale structures. I believe the vibrations would have to be similar in frequency to gamma radiation, and gamma rays are notoriously difficult to manage with any sort of precision.

My take-away from the pair of articles is this: tremendous potential lies at the femto-scale. Humans may only reach down as far as nanotech, but our descendants won't stop there. A universe awaits us inside every nucleus.

Waste Aversion vs. Loss Aversion

This, from the journal of the Society for Judgment and Decision Making:

Spending too much hurts the pocket. Appearing to have spent too much hurts the ego. People may willingly overpay in economic currencies to reduce psychological costs. However, losses that are not one’s “own fault” (due, e.g., to theft, accident, or market behavior) are not accompanied by this second-order pain. Thus, people may also prefer to tolerate additional expenditures (of time, effort, pain, discomfort, etc.), if only to avoid the psychological costs of feeling that they wasted their money.

The authors of the paper conducted surveys and found that a majority of people choose the more wasteful of two options as long as it made prior decisions appear less wasteful. I immediately thought of the sunk cost fallacy. However, the authors differentiate:

Our phenomenon is related to the sunk-cost fallacy. The prototypical sunk-cost effect involves a preference for A over B (e.g., driving to a ballgame versus watching it at home on TV) when one had made a non-refundable payment for A, even though circumstances have changed since payment was made (e.g., the weather is foul and traffic is impossible), so that B would have been preferred to A in the absence of the sunk cost (e.g., Thaler, 1980; Arkes & Blumer, 1985). Sunk-cost effects are weaker in one sense, and stronger in another, than the present effect.

Sunk-cost effects are weaker inasmuch as A is not dominated by B. In fact, at the time payment was made, A had not only been preferred to B, but even worth paying for. It is the change in circumstance that changed the preference. In our scenarios, on the other hand, one option dominates another unequivocally, and only the prepayment shifts preference to the dominated outcome.

Sunk-cost effects are stronger than the present effect inasmuch as they relate directly to behavior, and therefore can be observed in the real world, such as when public projects whose costs exceed expectation are pursued beyond a point at which they would not have been undertaken had a prior investment not already been made.

Loss aversion seems to explain the sunk cost fallacy. Similarly, waste aversion seems to explain the paper's results. I wonder how loss aversion and waste aversion are related.

When Stalker Behavior Becomes Mainstream

From a recent NYT article:

Google could have put face recognition into the Goggles application; indeed, many users have asked for it. But Google decided against it because smartphones can be used to take pictures of individuals without their knowledge, and a face match could retrieve all kinds of personal information — name, occupation, address, workplace.

“It was just too sensitive, and we didn’t want to go there,” said Eric E. Schmidt, the chief executive of Google. “You want to avoid enabling stalker behavior.”

This strikes me as disingenuous. Anyone as well-informed as Eric Schmidt must realize that in the very near future our wearable computers will be doing this sort of "stalker behavior" all the time, and at our behest. Whether the computer is in our pocket or built into our clothing or glasses, continuing advances in augmented reality and machine vision will combine to deliver a background hum of salient facts about our environment.

Having the right information about a particular person can make all the difference in a social situation. Our wearable computers will sift through masses of social networks, (micro)blogs, images, and video to deliver what info we might want about a person. It's likely that much of that info will be what we today consider private. If Google won't provide this service, someone else will.

But Google has to stay competitive. So it seems that what Eric Schmidt really means is that he's waiting for the perception of "stalker behavior" to change. As we adapt to shrinking privacy, the line between stalker and well-informed stranger will shift. I wonder if manners will dictate that we pretend we know less about a new acquaintance than we really do?

The Incessant Obsolescence Postulate and Space Telescopes

The Incessant Obsolescence Postulate (IOP): "Interstellar probes become faster and cheaper over time as technology improves. Therefore a pressure exists to delay launching such probes." A recent paper examines how the incessant obsolescence postulate, among other factors, influence timeframes for humanity's first forays to other stars. The always excellent George Dvorsky presents three flaws with the paper, all of which I agree with:

• Millis's extrapolations assume a linear progression of available energy density; technological development is showing a strong tendency to progress non-linearly

• He assumes that there won't be a "wild card" type breakthrough in propulsion technology and energy extraction; it's not unreasonable to assume that there will be a sudden breakthrough that could serve as a significant game changer 

• His 500 passenger colony ship is ludicrous; biological humans won't be making such a journey, and most certainly not 200-500 years from now

But what of the incessant obsolescence postulate itself? Its a fascinating notion, and the paper concludes that:

Due to the combination of the nonlinear nature of both advancement trends and relativistic spaceflight, there will be a point where an optimum launch opportunity occurs.

How might the existence of advanced space telescopes influence this optimum launch opportunity? For this purpose, space telescopes can be modeled as "probes" that reach their destination immediately after being commissioned. They have the further advantage of being able to shift observation targets much, much faster than interstellar probes. They are also more upgradeable than distant probes.

Obviously there are huge limitations, as well. Some types of science may never be done remotely. Still, I would expect continual improvements in space telescopes to push interstellar probe launches into the future. If telescopes continue to improve at the pace we've been experiencing, they may become the single biggest factor driving the incessant obsolescence postulate.

Raising the Sanity Waterline: College Mostly Fails

Folks believe lots of crazy things. Is our educational system mitigating this problem, worsening it, or neither? Ars Technica posts a review of a paper (not online as of this post), which may shed some light on this issue:

The new survey tested for informal reasoning in the biological sciences, using over 500 students at a variety of colleges, enrolled in classes ranging from introductory biology to advanced ecology. The results show that, even as the students are immersed in things like trophic pyramids and the Calvin cycle, they don't always come to grips with basic things like conservation of matter and energy.

For example, most students could describe how the process of photosynthesis involves removing carbon dioxide from the atmosphere and combining it with water to form carbohydrates, which then get used to build the cellulose that forms most of the plant's bulk. But, when asked to actually trace what's going on on the organismal scale, many students have problems recognizing that a substantial proportion of a tree's solid bulk originated from a gas, instead suggesting that most of it was brought up from the soil.

It gets worse:

Nearly 70 percent "chose 'sunlight' as a possible source of atoms in chlorophyll molecules," according to the results. They had problems with other aspects of energy, too, not recognizing that moving up a trophic level in an ecosystem generally entailed the loss of energy to the environment—"only 44 percent of students thought the top of a food web would have 'less available energy than the trophic levels below it.'"

This is the tip of a huge iceberg, most of which is outside the scope of this post. The paper's authors suggest a possible fix: more precise language on the part of the faculty. In the spirit of holding off on proposing solutions, I'll only address the survey's findings here. What appears to be happening is this: students are mostly getting a "technician's eye" towards the subjects they are studying. What I mean by this is that they know the jargon, and they are very knowledgeable about specific parts of their chosen field. The authors note that the surveyed students had "a tendency to substitute a wall of scientific terminology for actual understanding".

Now, obviously one has to narrow one's education in order to be effective in nearly any field. However, the normative college curriculum: "build a broad base of understanding early on and proceed to more specialized understanding", doesn't seem to be working properly. Speaking metaphorically, the "glue" that is supposed to bind all these specialized bits of knowledge is too thin. I see this problem only getting worse as the corpus of scientific knowledge grows. More and more specialized pieces of information will have to be taught at our colleges, crowding out more and more generalized classes. If I'm correct, then we should see similar surveys yield even poorer results as time goes on.

Mind Emulation and Stoicism

Robin Hanson on mind emulations; which he calls "ems":

Ems would need some resources, and their bodies and minds would collect defects and errors, even if none of these resources or defects are biological. So it would be functional for ems to redirect their hunger systems to track the collection of needed resources, and redirect their pain systems to monitor possible defects and errors.

Is this likely? Imagine an entirely virtual whole brain emulation, running in a datacenter environment. The only resource it would consume is electricity. "Defects and errors" would be handled seamlessly on the hardware side. Today, for example, many servers use a kind of memory which automatically corrects random errors.

It seems probable that ems will not have to deal with anything remotely analogous to hunger or pain. Assuming they can pay for their own electricity (and an occasional migration to new hardware), an em can happily hum along until the heat death of the universe. So why keep the ancestral pain or hunger pathways? For that matter, why sleep? Will ems feel and think so similarly to modern humans, once the barriers to self-modification come down?