(more) TED on games as a way to make change

I posted earlier (here and here) about sustainability and using games to try to change peoples' behaviors toward the more sustainable.

Following is a TED video about that same topic (driving behavior changes with games) from someone whose work it is to drive business by getting people to play games that require interaction with the real world: in this case the companies that want you to visit them and buy their products.

Key take aways for me:

• Last decade was the decade of social. This is the decade of games.
• With 7 dynamics, you can get people to do anything.
• Four dynamics presented in the video (the other 3 are "secret" so that the speaker still has a competitive advantage at the end of the talk :P).
• Appointment dynamic = Do something at a certain time and place.
• Influence and status = Offer things that convey social status
• Progression dynamic = Offer a definition of "100% complete" and track progress against it. Level-up.  Unlock rewards over time.
• Communal discovery = Everyone works together to achieve a common goal.

Solar chargers for electric cars?

OK. This picture isn't what I'm actually talking about, but it does highlight part of the problem that would need to be solved:

Could you reasonably make a passenger car that charged itself using solar panels?

A few assumptions:

• The roof of a Honda Civic is ~56 ft^2 [assuming ~70% of the length is roof and ~95% of the width is usable]
• An area that big could hold solar panels that produce ~565 W [assuming 10 W/ft^2]
• An electric car holds a battery that gives it a range of 100 miles.
• A battery that big holds ~150 MJ of energy [assuming 1.3x10^8 J/gal of gas, 29 mpg and electric motors are 3x more efficient than gasoline motors]
• Some optimistic assumptions: 
• you generate the same energy all day
• you work 8 hours/day
• battery charging is 100% efficient.

That would result in  ~16MJ of energy per day.
So you could recharge 11% of the battery a day, which would yield a range of ~10.8 miles.

If you assume the average commute is 33 miles each day (16.5 miles each way), that would mean:

• To allow for a full trip home would take ~12 hrs of recharge time or
• A panel that could generate ~0.9kW and be ~86 ft^2. That's more than 1.5x the area available on the car roof.

To give a sense of the size of the problem of solar charging exclusively:

• To fully charge the battery would take ~73 hours or 
• A panel that could generate 5.2 kW and be ~519 ft^2. That's more than 9x the area available on the car roof.

Most houses that get solar PV only go for 2 - 3 kW.

So self powered cars ALMOST make sense as long as you can recharge them from the grid after you get home. Fully self powered solar cars don't make sense.
It also strongly suggests that solar powered EV recharging stations (e.g. with the panels on a nearby roof or on the parking lot roof) actually make quite a bit sense.

A better way for electric cars? A Better Place.

What are the two main things you worry about with electric cars?

1. Price
2. Range

Gasoline is great because it is a well established, subsidized fuel with very high energy density (1.3x10^8 J//kg). That makes it easy to carry around and get pretty far with. Plus, when you run out, there's always a gas station somewhere nearby. Price and range... not a problem.

Electric cars are limited by their batteries. Li-Ion battery's energy density is nowhere near as high as gasoline (5.2x10^5 - 7.2x10^5 J/kg - gasoline has >180 times more energy per kg of "fuel"). Even accepting that a gasoline engine is 30% efficient vs a 90% efficient electric motor (i.e. total energy that would be converted to work per kg of fuel is 3.9x10^7 J/kg for gasoline vs 6.5x10^5 J/kg for Li-ion), that yields ~60 times more energy per kg of "fuel" from gasoline than from batteries. So to get the same range as a gasoline car (from one tank of fuel), requires either:

1. A very large battery -or-
• If a typical car holds 15 gal of gasoline (@ ~2.7kg/gal = 40.5 kg) , a battery that contains the same effective energy would need to weigh ~2430kg ... where a Honda Civic weighs about 1270 kg... not practical to carry or to pay for.
2. The ability to "refuel" the battery in less time than refueling a car 
•  Even with the best recharger possible, it takes hours (~4 hrs for a Tesla roadster) vs maybe 5 minutes to refill a gas tank.
• With a 100 mile range per battery charge, that would require about 4 battery charges per tank of gas (assuming CAFE standard of 27.5 mpg and 15 gal tank = 412 mile range). Sixteen hours (battery) vs 5 minutes (gasoline) (assuming both cars start empty).

 Those are pretty bad starting places to catch up from.

So Shai Agassi has an idea: Treat the electric car like a cell phone.
The customer buys miles and gets a subsidized car. The batteries are the medium for the "minutes." You never own them, you just use them.
Which means you can now swap batteries because you don't care about the specific battery that you have, only that you're getting the mileage you're paying for.
Suddenly the range problem goes away (~1min to change the battery).
Suddenly the price has a means of being subsidized (ala cell phone contracts that keep the price of the phone low for the end user).

Kind of like Blue Rhino propane tanks ... but way cooler.

And he created a company Better Place to do it:

• Get the cars built
• Get the battery swap system figured out
• Get the infrastructure in place
• Prove to the world that it can work

Check out this video (or the video below) of the system working in Tokyo. It's real.


So, admittedly, it only works where you have the infrastructure, but it certainly gives a new direction for addressing the two big problems of EVs.

Biking to work and energy efficiency

Bike to work and stop global warming.
Bike to work and get healthy.
Ok. Since it's so good for me and my planet and I'm only about 6 miles from work, I finally decided that I should give this a try.

I used a cool iPhone app (Runkeeper) to log how far I went and how fast I went there. As part of the app I got an estimated calorie usage... and that got me thinking about energy efficiency:

• I burn about 240 calories one way on this trip (~6 miles).
• That is about 1x10^6 Joules (~1 MJ) of energy.
• One gallon of gasoline has about 1.3x10^8 J of energy in it (130 MJ).
• My Prius, on roughly this same route, gets around 42 mpg (I'm a terrible hypermiler). So it takes around 0.14 gal of gas to make the trip, or1.8x10^7 J (~18.6 MJ).

So my bicycle commute is more than 18 times more energy efficient than driving.
...And I drive a Prius.
If you use the CAFE Standard fuel efficiency for a passenger car in 2010 (27.5 mpg), then that bicycle commute is more than 28 times more energy efficient than driving.

To be as energy efficient as cycling, the car would need to use 0.01gal of gas for the 6 mile trip, which is 600 mpg! Whereas the current automotive X-Prize is only looking for 100 mpg... 6x short of what is needed.

So we REALLY need to be doing more of what Copenhagen is doing with making cycling mainstream:

"If you make the bicycle the fastest way to get around the city ... you're going to get everyone and their dog to do it."

• Pervasive use of dedicated bicycle lanes.
• 37% of all people commuting to work or school use bicycles.
• 50% of all trips in the city are made by bicycle.
• Double bike lanes to accommodate bike traffic volume in some places.
• "Green wave" timed to give cyclists a no-stoplight flow into the center of the city for 6km (3.7mi).
• Red LED lights on the bicycle lane demarcation that sense coming bicycles and flash to warn cars to avoid right turn conflicts.
• Dedicated parking spots for cargo bikes (taking away 1 car parking spot for 4 cargo bike spots).

Picture credit: Adam Stein (terapass blog)

The Value of Uncertainty - Subjectivity and High Dynamic Range

It is generally agreed that 24 bit color provides more shades (~16.7 million) than the human eye can perceive (~10 million).
For grayscale, only about 100 shades can be distinguished while typical monitors can display 256 shades (8 bit).

Twenty four bit color is generated on a computer monitor by having 3 colored pixels (Red, Green and Blue) displaying up to 256 shades of brightness each (8 bits). Three colors x 8 bits per color = 24 bit color.

A typical digital camera uses 12 - 14 bit sensors, one for each color. This yields 36 - 42 bit color images. Why is this helpful? Even though the eye cannot see all the colors, a computer can still process them. By saving the additional information (i.e. saving the image in RAW format, instead of a compressed format like JPEG), the programs can rescale and process the images with much lower levels of artifacts. e.g. stretching the contrast can introduce bands in the colors called posturization.

 Another image processing option that the extra colors and range make more effective is High Dynamic Range (HDR) photography. In this computational photographic method multiple images of the same scene are combined to maximize the visibility of all areas of the photo. This can result in some very striking images but often only after a significant amount of manual tweaking to get just the right effect. Here-in lies the value.

"Existing tools are therefore likely to improve significantly; there is not currently, and may never be, an automated single-step process which converts all HDR images into those which look pleasing on screen, or in a print.  Good HDR conversions therefore require significant work and experimentation in order to achieve realistic and pleasing final images."

There is no simple answer to getting the "best" HDR rendering. Therefore hobbyists will pay for tools, new tools and more new tools to let them conquer the uncertainty - that there might be a better picture somewhere in the data.

Inspection Equipment

How can KLA-Tencor get away with a 60% gross margin selling hardware with substantial COGS?
That comes largely from the defect inspection groups vs the metrology groups. Why?
Value, in this case, comes from the lack of a definitive base-line. In metrology the standard is defined: a micron is a micron. Differentiation is hard because there is a clear definition of success. In inspection, there is always the fear that there might be a better result somewhere in the data... You don't know if it's good enough.

What does HDR have to do with this?
Humans can't see the difference between 10,000 gray levels but machines can. Setting up the algorithms to properly deal with this difference is hard. For example, one aspect of tuning comes from the bit depth of the sensor. Combinations of possible parameters increase from 8 bit (multiply combinations by 256) to, say, 14 bit (multiply combinations by 16384). Increasing the bit depth increases the amount of information and the possibility that there may be good results in the data, but it also makes comparisons even more difficult, i.e. higher risk.  Proving that you are the best at reducing this risk is hard. Customers pay a premium for what is hard to do.

Medical Imaging
Why are medical imaging devices constructed to such demanding specifications, particularly native bit depth (12 bit grayscale) processing and display, when the human eye can't see that many shades?

Table of native resolution and bit depth from common medical imaging equipment (slide 9).

My guess (I have not seen a definitive answer): Liability - if a diagnosis was missed or rendered incorrect because of a conversion artifact or loss of data that would be bad.

The risk caused by subjective interpretation of images, particularly if those images have been modified in some way by the system, drives mitigation through rigorous technical specs (extended bit depth, color temperature control, white level balance, color matching, multiple monitor matching). Those specs increase material, installation and maintenance costs directly. They also reduce the number of vendors who can meet those specs further driving up price.
Are those specs and the associated costs really necessary? Fear (and / of lawyers) say "yes" so medical grade display systems sell at a significant premium vs commercial displays.