Attention Span by Google

The rise of Twitter and tl;dr may seem to presage an ever shrinking attention span. However, Anders Sandberg suggests that our attention spans are simply becoming over-saturated with an excess of interesting content:

So my theory is that we can still pay attention for a long time - but we want a lot more to happen per unit of time too. We want faster rewards, more action.

Why? Perhaps because there is so much stuff out there, so the alternative cost of spending a lot of time on something that does not turn out to be worthwhile is higher. In the time you have spent reading this post (and I writing it) we could have read several RSS entries and short blog posts, watched a YouTube clip, browsed Wikipedia or run a calculation in our favourite math program.

[...]

If this model is true, then we should expect the trend to continue. In the future, we are going to have far more good books, films, comics, papers and other documents instantly available.

It is rational to demand quick and reliable evidence that whatever we have in front of us is relevant or interesting. Spending a lot of time finding out if it actually is by just consuming it would mean we would often waste precious time and attention on things that are not as good.

There is of course a trade-off here, since some important things do not look inviting (since they were made before the current attention economy) and some unimportant things masquerade as important. Smart agents balance the exploration with exploitation.

This is why reliable filtering and reviewing actually are key transhuman technologies. And why training to recognize the real cost and value of what you are doing is such a key transhuman virtue.

This is (to me) an intuitively correct notion. I'm also struck by how extraordinarily bad we are at filtering our various incoming data streams. We're accessing ever larger portion's of the world's information output, without a commensurate boost in discrimination.

In fact, we're so incapable of filtering that we often rely on our peers to do the job for us. Social networks are a useful but nonetheless inefficient way to sift for valuable information. Word of mouth can be hit or miss, possibly because of cognitive biases that make us believe our friends are more similar to us than they may be.

Google continues to work in the field of "attention management". Unfortunately the killer app has yet to be invented. Increasing privacy concerns may hinder Google's efforts to gather the kind of highly personalized information required for the task.

I, for one, would gladly trade away most of my privacy for an effective service to guide my attention.

Dreamers Mostly Fail

A 2002 study finds that positive expectations predict success, while positive fantasies correlate with failure. From PsyBlog:

Across four studies the researchers examined how people thought about each of these challenges. They measured how much they fantasised about a positive outcome and how much they expected a positive outcome.

[...]

Take those looking for a job. Those who spent more time dreaming about getting a job, performed worse. Two years after leaving college the dreamers:

• had applied for fewer job,
• unsurprisingly had been offered fewer jobs,
• and, if they were in work, had lower salaries.

On the other hand those who entertained more negative future fantasies were more likely to achieve their goals. Similar results were seen for the other goals.

This seems to be yet another warning about the dangers of excessive far thinking.

HT: Eliezer Yudkowsky

Journal of Personality and Social Psychology, 2002. DOI: 10.1037/0022-3514.83.5.1198

A Gentle Touch For Space Junk Removal

Radiation pressure is the minuscule force exerted by a photon when it strikes an object. We don't experience radiation pressure in our everyday lives because it is drowned out by much, much larger forces. But in the vacuum of space, the feeble push of radiation pressure isn't opposed by significant aerodynamic drag. Over time, it can do real work.

A NASA study aims to do just that. The authors propose using a commercial, off-the-shelf telescope and laser to point a 5-10kW beam at potentially dangerous space debris. The radiation pressure won't be enough to de-orbit the debris, but according to simulations it should alter the debris' orbits enough to avoid a collision. And at orbital speeds, a collision is catastrophic.

The authors also suggest the telescope-and-laser system could be employed to boost the orbits of small satellites. This is an intriguing notion to me. Nano- and pico-satellites are a relatively recent development, but they're limited by a lack of thrusters. Without thrust, the orbits of tiny satellites inevitably decay and the vehicle burns up as it reenters the atmosphere.

But imagine if the satellite operators could purchase time on a ground-based laser facility. The laser would push their satellite to a higher orbit, provided it has the correct mass-to-area ratio. This sounds promising because it provides thrust "on demand", without much additional capital cost on the part of the satellite operator.

Tasty Fruit, Minus the Seeds

New research published in PNAS has identified a gene for seedlessness in the sugar-apple. This is the first such gene ever isolated in a crop plant, and it looks like they couldn't have chosen a more delightful fruit to make seedless. The flesh of the sugar-apple is very sweet and soft, and is described as custard-like. The major roadblock to wider commercial cultivation of the fruit is an abundance of gigantic seeds:

Image from Muhammad Mahdi Karim

Future research may lead to seedless cultivars of the related custard-apple and cherimoya. I'm looking forward to more choices at the supermarket.

PNAS, 2011. DOI: 10.1073/pnas.1014514108

Interesting Fact of the Day

The world's first jet vs. jet aerial combat was against robots.

In 1944, a desperate Nazi Germany began launching unmanned, jet-engined V-1 flying bombs against targets in England. The Royal Air Force countered by scrambling fighters to shoot down the V-1's before they reached their targets. From the Wikipedia page:

Even though it was not fully operational, the jet-powered Gloster Meteor was rushed into service with No. 616 Squadron RAF to fight the V-1s. It had ample speed but its cannons were prone to jamming, and it shot down only 13 V-1s. These were the first air combats between two jet-powered aircraft in history, although that achievement is usually ascribed to the battles between manned jet aircraft during the Korea War in the fall of 1950.

The Onion Is A Vegetable After My Own Heart

As usual, The Onion is the only mainstream news outlet I can relate to:

Calling the situation "none of our business" and "not worth a second of our time, quite frankly," a responsible and thoughtful U.S. populace uniformly decided this week to ignore Charlie Sheen's recent outbursts, saying they had far more important things to focus on than a sitcom actor's personal troubles.

A Guilty Conscience Makes Us Punish Ourselves and Others

For today, I present a pair of articles on the effects of guilt. First up is Robin Hanson:

A psych study asked people to think of someone they felt guilty toward, or made them imagine feeling guilty toward someone (e.g., slacking off on a joint project, or being careless with something borrowed). Researchers then had these guilty folks divide up money between themselves, the victim, and a third party (e.g., a deserving charity or random person). Compared to controlled conditions, such people give more money to the victim, but at the expense of the third party, not themselves.

Next, a study (gated paper, press release) published in Psychological Science:

When reminded of an immoral deed, people are motivated to experience physical pain. Participants who wrote about an unethical behavior not only held their hands in ice water longer but also rated the experience as more painful than did participants who wrote about an everyday interaction. Critically, experiencing pain reduced people’s feelings of guilt, and the effect of the painful task on ratings of guilt was greater than the effect of a similar but non painful task.

I wonder at these two expressions of guilty feelings. Under what conditions do we choose to punish ourselves versus punishing innocent others?

Prescription Placebos

According to the British Medical Journal, most clinical interventions have unknown effectiveness:

Of course, this doesn't mean that most treatments are ineffective. But the finding does suggest that clinicians may be leaning on unproven treatments for situations where no proven remedies exist. And for these situations, placebos may be a good substitute for unproven treatment.

Doctors are already warming up to the potential of placebos. A study for the German Medical Association (BÄK) finds that fully half of all German doctors prescribe placebos, and for good reason:

The report recommends that students and doctors should be taught about placebos and their usage.

"Placebos can maximise the effect of medication," says Robert Jütte, author of the study and a BÄK board member.

"They can reduce undesirable side-effects and are a more efficient usage of our healthcare budget."

Recent research, he said, showed that placebos had helped 59% of patients who had been suffering from an upset stomach. Used to treat depression, placebos have the same effect as antidepressants in about a third of cases. (more)

The report also warns about ethical issues, and recommends that placebos only be prescribed for minor illnesses. But what about seriously ill patients for which no proven treatments exist? This is a thornier issue, and the BÄK has called for internationally recognized guidelines for the use of placebos.

HT: Tyler Cowen

Ceci n'est pas une personne.

It's a robot, the product of the Geminoid Lab in Japan. They plan to use it to study human-robot interaction, as well as a particularly expensive variant of telepresence they're calling "blended presence".

Warning: Serious Time Sink Ahead

Fractal Lab is a WebGL fractal explorer. You'll need a cutting-edge browser to use it, like Firefox 4 or Google Chrome's dev channel.

I could spend ages in the mandelbox fractal, alone. Here I found baroque mechanical complexity:

Click to enlarge

And also strangely organic landscapes:

Click to enlarge 

Click to enlarge

Venturing deeper, an alien core:

Click to enlarge

If we live long enough, we may exhaust all novelty in the physical universe. It's comforting to know that the mathematical universe will always be there, waiting for us.

Google's Self-Driving Car, Unleashed

Coincident with Sebastian Thrun's TED2011 talk, Google is allowing a few journalists a rare look at the capabilities of their autonomous cars. Usually the vehicles are programmed to avoid high g-forces and to obey the speed limit. But on a closed course, the system can show off:

Autonomous cars will be handy to have in case of an emergency. Imagine a "Max Emergency" mode that causes the car to aggressively drive to the nearest hospital while putting a call through to 911.

Sounds much better than waiting around for an ambulance.

3D Printers Not Good Enough for the Battlefield?

Over at Accelerating Future, Michael Anissimov has a dim opinion of a US Navy proposal solicitation for "semi-autonomous breeding robots with built-in 3D printers":

In terms of strength, the products of 3D printers are weak and can easily be pulled apart with your bare hands. If you want a strong product you still have to go to the machine shop or foundry.
Interesting proposal solicitation, however it is worth remembering that military commanders have been making breathless requests for futuristic technologies since time immemorial. There will be no “semi-autonomous breeding robots with built-in 3D printers” of practical battlefield value until at least 2025. However, this is the sort of thing a superintelligence could build millions of to do its bidding.

Regarding the strength of 3D printed components, I must wholeheartedly disagree. 3D printing systems like selective laser melting can build parts out of steel, titanium, or aluminum that approach the density of conventionally manufactured objects:

I don't know about Anissimov, but these parts can't easily be pulled apart with my bare hands.

Going a bit further, I don't think the Navy's proposal is that "breathless". I can imagine building a selective laser melting system into a medium to large sized autonomous platform, something like TARDEC's Autonomous Platform Demonstrator, for example:

OK, so you've got a robot tank that can build metallic parts, so what? Well, let's also add a supply of modular servos, processors, sensors, and batteries. A robotic system to assemble all this together into functioning drones might be stretching the limits of available technology. But what you'd get is an autonomous vehicle that can build smaller autonomous vehicles.

Now, whether this is a particularly good idea is a whole other can of worms. For now, it's probably cheaper and more efficient to just pack prebuilt drones into larger vehicles. Anissimov claims this situation won't change before 2025. I'm much less certain.

Americans Overestimate Government Influence

A new working paper co-authored by Bryan Caplan attempts to quantify voter bias:

Systematically biased attributional beliefs turn out to be common and large. Fully 14 out of 16 survey questions exhibit statistically significant biases. Compared to experts in American politics, the public greatly overestimates the influence of state and local governments on the economy, the president and Congress on the quality of public education, the Federal Reserve on the budget, Congress on the Iraq War, and the Supreme Court on crime rates. The public also moderately underestimates the influence of the Federal Reserve on the economy, state and local governments on public education, and the president and Congress on the budget. While we are open to the possibility that non-cognitive factors explain observed belief gaps, controlling for demographics and various measures of self-serving and ideological bias does little to alter our results. A full set of controls reduces the absolute magnitude of the raw belief gaps by less than 13% - and leaves the number of statistically significant lay-expert differences unchanged.

The biases seem to lean in favor of overestimating the influence of government. These results reminded me of a John Stossel video:

We should keep in mind that politicians may overstate their capabilities because voters often expect more than is possible. Unreasonable demands lead to over-inflated claims by candidates, leading to yet more unreasonable demands and so on.