Apple's Jonathan Ive's wonderfully thoughtful interview segment

...from the movie Objectified. (h/t gizmodo)

"Watch Jonathan Ive's Segment in Objectified."

g

A Creative Whack: Exaggerate

Hi -

Do you like to collect creativity-enhancing techniques?

I've enjoyed books like James Adams's Conceptual Blockbusting, which helps you notice (and thereby eliminate) your own blocks to creativity. You've probably also seen Brian Eno's "Oblique Strategies," a deck of pithy thought-provoking sayings intended you to turn your creative problem on its ear.

Another creativity card-deck is Roger von Oech's "Creative Whack Pack." Many of the techniques are well-documented elsewhere, but I like how a deck of cards invites browsing. (I keep it in my briefcase and flip through it now and then.)

Here's a good example. It's a technique used by many successful entrepreneurs and inventors:

EXAGGERATE (#22)

Imagine a joke so funny that you can't stop laughing for a month. Paper stronger than steel. An apple the size of a hotel. A jet engine quieter than a moth beating its wings. A home-cooked meal for 25,000 people. Try exaggerating your idea. Think big: what if it were a thousand times bigger, louder, stronger, faster, or brighter? What if the number of people who could use it were increased a thousandfold? Now think small; what if it were only one-thousandth as powerful, fast, costly, or complicated as before? How can you exaggerate your idea?

-gregg

Sony's 360-degree display

Sony is demonstrating a 360-degree tabletop display this week. They claim it is a "highly-multiview" autostereoscopic display, rather than a 360-degree 2-D display.

Have any of you seen it? How did it look? Is it a rotating LED paddle, or something different?

Link: http://av.watch.impress.co.jp/docs/news/20091019_322773.html?ref=rss

Google-English translation: http://translate.google.com/translate?sl=auto&tl=en&u=http://av.watch.impress.co.jp/docs/news/20091019_322773.html%3Fref%3Drss

Recent update, with more images:

http://translate.google.com/translate?hl=en&sl=auto&tl=en&u=http://av.watch.impress.co.jp/docs/news/20091022_323667.html

I dimly recall coming up with the term "highly-multiview" for an article a while back... but maybe that was just a coincidence of the translation.

g-fav

ps Thanks to Andrew Woods for bringing this to my attention.

Two sub-$5 toys with curious operational mechanisms

Hi -

Ever wish your 3 year-old had a toy or two which, despite being < $5, triggers a sense of wonder about how on earth it works? Here are two toys that I (finally!) found at Henry Bear's Park. Both are made by Toysmith (Auburn, WA).

Toysmith #9236 "Flashing floating DISCO glide ball"

I first saw this at a friend's house and spent two years tracking it down. This odd little ball is black with colored dots. When you roll it across the floor, it seems to frictionlessly glide along - it's actually floating within a clear sphere. If you smack it just right, the dots glow and blink. $4!

Toysmith "Never fall UFO"

It's a wind-up plastic UFO that scoots along a tabletop, but a little mechanism inside makes it change direction just before it would have plummeted off the edge! The salesperson suggested that it would be a fun way to keep your one-year-old entertained at a restaurant, since it can't fall off the table. $2!

I hope to see more inventive toys with curious mechanisms inside...

g-fav

Optics: 2 papers @ OSA meeting. Me on 3-D, and Accardi / Wornell on QUASI light fields

Hi - 

I finally posted the 2-pager for an invited talk I'm giving at the OSA's 93rd Annual Meeting (Frontiers in Optics 2009 / Laser Science XXV): "Progress in Volumetric Three-Dimensional Displays and Their Applications."

It'll be on Tuesday, Oct. 13 (2009) in San Jose, CA.

Also: Anthony Accardi and his advisor Gregory Wornell will be publicly presenting very creative and multi-disciplinary work, along the lines of their recent JOSA paper, "Quasi light fields: extending the light field to coherent radiation." For free previews, see the PDF of one of Anthony's recent talks mirrored at Prof. Raskar's site, or an earlier post on related topics at ArXiv.

-g

Whew! Now you can swing side-to-side without fear of being sued!

Hi -

If, for fear of patent infringement, you had been avoiding swinging side-to-side on a swing suspended by trees - - breathe a sigh of relief! Read on...

One example of some of the problems with the U.S. patent system is exemplified by U.S. Pat. 6,368,227, "METHOD OF SWINGING ON A SWING," which generated substantial publicity for its young inventor when it issued in 2002.

Or, I should say, the expired '227. According to the USPTO database, it expired in 2006.

What did it cover?

1. A method of swinging on a swing, the method comprising the steps of:
a) suspending a seat for supporting a user between only two chains that are hung from a tree branch;
b) positioning a user on the seat so that the user is facing a direction perpendicular to the tree branch;
c) having the user pull alternately on one chain to induce movement of the user and the swing toward one side, and then on the other chain to induce movement of the user and the swing toward the other side; and
d) repeating step c) to create side-to-side swinging motion, relative to the user, that is parallel to the tree branch.

2. The method of claim 1, wherein the method is practiced independently by the user to create the side-to-side motion from an initial dead stop.

3. The method of claim 1, wherein the method further comprises the step of:

e) inducing a component of forward and back motion into the swinging motion, resulting in a swinging path that is generally shaped as an oval.

4. The method of claim 3, wherein the magnitude of the component of forward and back motion is less than the component of side-to-side motion.

Whew! Google Patents: 6,368,227. Or search for the prosecution history in the file wrapper at the USPTO, here.

-g

Mass High Tech profile

Hi there - 

I'm profiled in this week's Mass High Tech in a piece by Jim Connolly: "Gregg Favalora sets his sights on the optics industry."

Maybe it’s in his genes or maybe it was from his environment, but Gregg Favalora feels he was destined to work with optics and imaging. In fact he has been working with 3-D imaging since he was in junior high...

.

-g

(Edge) Short course on synthetic genomics

Hello -

Brockman's Edge.org continues to showcase interesting work from his cadre of leading-edge thinkers (or, as I think of them, "the smart cool-kids.") George Church and Craig Venter taught a 6-hour course on synthetic genomics, presented at Edge.org, for free.

I didn't make it through all six hours, though what I watched I found really absorbing. I also admit I find the selection of students equally fascinating - they introduce themselves in the first minutes of part one. If you are a frequent visitor to Edge.org, or watch TED videos, or, really, go to any of the "conferences by big thinkers," they'll be familiar (though mostly West Coast) names.

Edge Master Class: A Short Course on Synthetic Genomics

g-fav

Watching simulated chromosomes mutate as they try to reach a "goal"

Hi -

What might a time-lapse of evolution look like, at the genetic level, watching genes win and lose on the basis of how well the organism was able to survive? In this blog post, we'll look at a really dumbed-down version of this that's still, I think, conceptually compelling.

How is a recipe different from an evolving organism? What if real-world problems, like designing bridges or electronic circuits, were solved by a computer that hunted around for a good solution rather than at the explicit, exacting direction of a team of engineers?

Genetic algorithms are computational "recipes" that function in this manner. Broadly, it's been possible to program a computer to program itself, using a technique that has some analogies with real-life evolution. This won't surprise any computer scientists, since the technique has been around since the 1950s and gained popularity in the 1970s. But I thought you might be interested in GAs because we can peer (diagrammatically) into an accelerated world of simulated, simplified "evolution" - and, also, it's a fun programming exercise that lets me get a little more experience with languages like Mathematica and C.

(If this is new to you, see the brief overview that I blogged about in 2007.)

In short, we give our evolution-simulating program a few things:

• A population of organisms (usually called chromosomes), each chromosome being composed of a bunch of genes... and each gene can be as simple as a 0 or 1 -- or as complicated as a mathematical function like Sin(x). Usually the population is random, a bunch of Frankensteins.
• A goal to achieve, like "figure out what math function best approximates this stock-ticker," or "figure out how to place 100 Lego blocks to make a really long bridge that doesn't collapse under its own weight." Usually the goal is presented in terms of a test called a fitness function. We test each organism (or chromosome's) ability to pass the test. The chromosomes that do better survive - pairs of chomosomes (parents) breed, make children, the kids are mutated, and the weaker parents are killed. Repeat.
• And you need to let the simulation know how likely it is that genes will mutate or that breeding parents will swap chromosomal segments ("crossover").

As a prelude to a larger project, I put together a simple Mathematica notebook that would try out a problem in a textbook on the subject, Melanie Mitchell's An Introduction to Genetic Algorithms. The test?

Turning random "binary" chromosomes into a freaky population that loves the number 1

Start with a bunch of random chromosomes, like:

{0,0,1,0,1,1,1,0} (chromosome 1)

{1,0,0,1,1,0,0,0} (chromosome 2)

...and let them breed and mutate. The chromosomes that looked most like {1,1,1,1,1,1,1,1} were not killed off.

Our population had 40 chromosomes, each with 10 genes valued 0 or 1. The universe runs for 100 time-steps.

Here's a chromosome from the randomized primordial ooze, our friend chromosome 5:

And here is its universe, textually:

..and graphically (Mr. Five is in the fifth row from the top):

A few simple experiments: does it work?

Of course, I implemented the basic GA stuff, like a function for "fitness-proportionate selection using roulette wheel sampling," a fancy way of saying "a way to favor the parents that most look like a bunch of ones." At the end I'll link to the notebook so you can use it, laugh at my code, and offer suggestions.

Here I'll show you a few plots that show how the population's fitness improves (hopefully) over 100 sessions of selection, mating with crossover, and "honey, I mutated the kids."

Above, we use the recommended parameters for mutation (0.001 = 0.1% likely) and crossover (70%). As you can see, the population approaches the theoretical maximum fitness (400, which is the number of chromosomes multiplied by the number of genes each) at about step 20.

The curve goes up, meaning that the "fitness" of the chromosomes is improving.  A plot of the chromosomes at ten snapshots in their history are provided right under that.  They're getting more black - which means that they're becoming filled with 1s.  (Under that is what the first ten generations looked like.)

It bounces around, too. That's because the genes mutate now and then.

What if we go outside without sunblock?

What if we turn the mutation probability up to a whopping 10%?

Augh! The population never has a chance to improve its fitness; just as it tries to get healthy, we flip bits and really screw things up.

One other quick experiment: do what extent does the crossover probability matter? Let's turn it down to 0.5%, so that crossover happens about 1 in 200 matings. We'll increase the number of generations to 500:

Neat! Maybe I'm reading too much into this, but it's as if improvements in fitness happen in punctuated jumps.

Oh, what does the life of one chromosome over the whole history look like? Going back to our original parameters for mutation and crossover, Mr. Five indeed tends toward all ones (or "all black," reading downward.)

Well, that's it. If you find this interesting, you might want to code some of this stuff up on your own. I suggest a few video clips of neat stuff (e.g. synthetic organisms learning to crawl, Lego bridges...) and good books on these blog posts:

• g-fav's first attempt to write a GA, in C (May 2007)
• "virtual schadenfraude" - turn on your speakers! (Oct 2008)
• Lego bridges and truck backer-uppers (May 2009)

The Mathematica notebook (do whatever you want with this, was written for fun and not error-tested, not responsible for damages...) max_ones.nb - free Wolfram Mathematica Player

Have a good weekend!

Gregg

Maps on FedEx shipping labels

Hi - 

Among Kevin Kelly's numerous thoughtful blog posts about network effects, the Web, and various design-issues is this brief post about FedEx's use of road maps printed directly on packages.  Check it out.  (One commenter noted that it would be more useful in the cab of the truck on a clipboard, instead of in the back of the truck stuck to a package, but still...)

g