Any real skunkworks team could develop a viable battery-kill scheme in less than a week… upon impact (same sensors that deploy airbags) flood the battery electrolyte with a "poison" to kill the electrochemical reactions. Duh! The slowpoke arrogance of GM is holding true. Expect a few deaths (collateral losses, so sorry) attributable to battery self-burnout before GM middle-management suckups gets off their collective backsides and allow the engineers to do their jobs.

A single large battery has the advantage of on 2 connections.  Many smaller batteries could be distributed through out the car and certainly reduce the fire and with increased surface area per watt, eliminate the need for a cooling jacket or water-cooling.  The problem is not to get too small or too many batteries to increase the connections and harness weight excessively.  I would like to suggest an overall change to distribute batteries and even uses 8 individual motors as opposed to one and eliminate the transmission but I haven’t seen a website for GM that allows this.

Design the battery system into a tray that slides out under the back of the car. Design air channels into the heatsunk carrier for passive cooling. Eject the pack in detected emergencies.
The packs could also be swapped at battery service stations in 5 minutes (like people swap propane bottles) instead of the utterly idiotic thought that people will sit at a charging station for 45 minutes.
Obsoleted packs would work their way out of rotation over time. More efficient / lighter packs enter rotation in new vehicles and are swapped for similar packs at the stations. Pack cycles used (condition data) is maintained in the pack controller and swap charges are adjusted accordingly. Or people can never swap and keep their new snazzy pack! Charge at home.
Heavy users are charged more for swaps (ala degrading their pack's life cycle if the pack were not modular) to cover costs of depleted pack replacement in the system. Pack service station locations / inventory identified on GPS.
This really isn't rocket science. Create a specification for the battery tray and stick to it.

Some of the suggestions really deserve some abuse, starting with the one to inject a chemical to "Poison" the coolant. Knowing about the chemistry of the coolant, that just can not happen. But if it could, how would the poison circulate? The concept of a quickly removable battery would assure that you had NO battery, at least here in Michigan, where they steal the welded in catalytic converters. How long could you keep an easy to remove $5000 battery?  I did like the concept of a battery made up of a bunch of small sections, it does have many advantages, except for being practical or reliable. The actual solution to the battery fires in crashed vehicles is to provide a few battery drain plugs accessible from under the car. While the car is still raised by the tow truck, crawl under and remove as many of the plugs as cam be reached. The slow leak will drain the coolant in a few hours, removing the hazard.  Just be sure to keep pets and kids from drinking the spilled coolant, since it is bad for your insides.

I am amazed at the naive responses and personal agendas posted.  Only a couple really addressed the issue.  Simple understanding says that there is a very lot of stored energy in the battery structure somewhere.  Upon being damaged the objective must be to prevent the release of this energy at a high enough rate to injure occupants or bystanders.  The problem has existed and still exists in liquid fueled vehicles and the total vehicle design has evolved to do the best possible and still carry enough energy (fuel) to make the vehicle useful.  Today they make the tank tough and locate it where it is as protected as it can be in the event of a crash.  There are still some fires associated with accidents and in most localities they roll a fire truck first to cover automobile wrecks.  Why any higher expectation should be placed on an electric vehicle fails logic and reason.  Suggest a three fold design objective: 1- locate the battery in as protected a place as possible (the volt is long and thin and appears kind of fragile). 2- being as the energy in the now common battery is located in small separate packages (cells) make them separate electrically in the event of damage so if any release their energy it will be only a few. Something like a fuse at each cell may not work but it is a starting concept. 3- Add similar current interrupting devices at progressively higher levels to actuate much like an air bag to head off the possible fire from the energy output within the rating of the battery.  The legitimate output of a 15 kW battery will surely start a fire under many otherwise insignificant conditions.
Anyone who has looked at the many times published concept of engineering investigation will see it does not involve off the wall unrelated inputs, personal or political agendas, or violations of the well known physical and chemical laws.  The term "off the wall may apply to good and useful approaches but only if they are directed at a solution.
I suspect that the whole issue is a little out of context in that there is a higher standard being placed on a battery than is in current place for a fuel tank but good effort in advancing the science of safety is never wrong.  It must however, be approached and led by "cooler heads" to assure that the result is really progress and is not obviated by unintended consequences.

BMW engineers on more than one occasion disclosed that "AIR COOLED" battery pack in MINI was a major mistake !
According to them NISSAN is about to find out !
In BMW tests no battery lasted more than 18 months till it lost lot of the "range", LEAF will kill NISSAN in California where they have to replace the batteries for FREE for up to 8 years !
 
At about $16,000 per battery pack, NISSAN is likely to have loss of $30,000 to $50,000 for each LEAF sold, so no wonder they really promote the 2 years LEASE, but to theri surprise most people are actually BUYING the LEAF !!!
The losses to NISSAN will be in millions.
 
VOLT at least does not use the full battery capacity and the engine "masks" the range deterioration with time and temperature, but in BMW and LEAF you just can not hide the fact that the vehicle will loose lot of range when 2 years old or when heater or A/C is used.
 
 

Gas powered vehicles have fuel pump cutoffs activated by crash sensors. In a similar manner, the suggestions advanced by some posters that the same crash-sensors could be used to safely release the battery's energy in case of a severe accident. Whether it is in the form of a switcheable load, a chemical stabilizer or a combination of both, has to be determined with the aid of  the time-proven engineering tools of  failure analysis and testing.
Hindsight is always 20/20, so I could be wrong in my thinking, but I think that probably the battery team did not perform a thorough DFMEA.
Anyways, the engineering community has always learned its lessons from from accidents like this, and applied those lessons towards corrective actions that improve the product….. If this is the case, and I'm confident it is, this accident will prevent future life and material losses.

I'd make the battery management protocol an open system, so EVs can easily have their batteries replaced, even by independent mechanics, as the technology improves.

Run the car with a gas engine and forget the electric.

1) A sensor will detect a serious crash. 2) A pump will force a harmless liquid into the battery. 3) A flush-valve opens allowing the battery electrolyte to discharge into a resevoir. 4) If the battery is deemed reuseable after crash restoration, the process is reversed. 

I would replace the battery with an internal combustion engine.  They are a proven technology and can be very energy efficient.