My impression from various sources (partly news, partly being in aerospace, though not aircraft) is that for aircraft crashes people often survive the crash and die from the fire (fueled by lots of jet fuel spilled around), so there's a high priority placed on putting out the fire. The news reports even said that the modern airport fire trucks start dousing the fire with foam from a pretty good distance as they rush to the scene.
The foam certainly complicated things, and even if she was standing, a foam jet could have easily knocked her over. It was probably also complicated by the number of people who survived and were managing to get out on their own-- probably not that common in a crash where fire is breaking out. And firefighters don't experience nearly the number of aircraft crash+fire as they would structure fires, so anticipating what's going to happen when a large number of people survive a crash like that is going to be less predictable.
All that said, they should have some kind of standard practice to minimize the chance of something like this happening again. It's apparently harder than it seems-- an example where conditions are nearly perfect and nobody is any hurry is the vehicles patrolling the beachs in SoCal. Every few years someone gets run over by a lifeguard/police/fire vehicle who either wasn't paying close attention while driving or didn't do a walk around before they started up and drove off.
I don't think I've seen a car made in the past decade or so that had a 0-60 time slower than 8 seconds, and FWIW, one could argue that if everyone's car had 0-60 times in the teens, the roads would be even safer.
Prius. 12.9 seconds. And there are plenty of other cars slower than 8 seconds, mostly in the sub $30K price range.
Try merging from a dead stop into dense 60-70 mph traffic with a short merge lane in a car with a 13 s 0-60 time and three passengers-- you might do it but you won't enjoy it. I've been shopping for a replacement for my old saturn wagon and it's hard to find a small car with as much hauling capability, similar gas mileage and as good aacceleration (0-60 time 9.2 s) at a reasonable price (e.g. comparable inflated cost).
I'm not sure putting everyones acceleration in the teens would make things safer, unless cars had an enforced minimum separation distance-- finding and getting into a hole when merging into dense traffic is *way* easier if you have some acceleration.
$2.80 of the $27.95. Even if you completely remove it, you're still talking about a $22.32 ebook. And if you're going through a distributor, you aren't completely removing it - they still take a cut.
But for an ebook you're also eliminating all the costs of the physical space at the retailer. The $12.58 that the retailer gets isn't pure profit-- they have to pay for space, utilities, employees, etc. out of that
And pre-production is generally easier with an ebook than a paper book. You can script both, as GP mentions, but to make a nice looking paper book generally takes more tweaking than a nice looking ebook.
The cassette player died a while ago, necessitating the FM transmitter. It was also an annoying GM cassette player, which required the slightly more expensive cassette adapter that had a short loop of tape to make the player happy so it wouldn't just eject.
You obviously don't live in Los Angeles. There's no such a thing as a 4 car length gap much of the time. At on and off ramps it's common to have multiple cars merging on and off simultaneously with less than a car length between any of them.
My 15 year old car has a cassette player. A CD changer was an option that the dealer could install. My iPhone communicates with the car via FM transmitter in a Jabra Freeway bluetooth speakerphone. The setup is pretty reasonable, though I find it easier to put the speakerphone in the middle of the dash instead of the visor. The FM transmitter on the Freeway isn't as good as the one that was on the Cruzr2, but it handles multiple phones better. Wheel mounted controls would be nice though.
I use a dash-mounted iphone with a touchscreen interface for nav and music in the car, but I have it set up so that I can use tactile feedback for hitting the spots I need while driving. The traffic map and music apps are set up in corners of the screen, so I just need to hit a corner to get to them.
I also just started shopping for cars and was having thoughts similar to the submitter-- why would I want to pay $$ for a built in, dedicated system from the car maker that's already behind what I can do on a phone or a tablet? They should just provide standardized tactile inputs on the wheel, plus more buttons and an optional screen in the dash. There should also be some standardized interface for sending vehicle specific information back to the user's device (e.g. tire pressure, fuel economy, etc.).
The problem we are trying to optimize is: How can we move the most cars (maximize distance) in the least amount of time. i.e. dX/dT. Which looks like a differential equation.
One small but important nit-- you're not trying to move the most cars, but the most people.
It is a shame that the Dept. of Motor Vehicles doesn't know shit about standing waves nor teaches people how to help optimize keeping vehicles moving in the traffic flow. One of these days every car will be able to pass it's current speed both forward and backwards to its neighbor's car so that people 5, 10, 20 mins down the road can know about future traffic conditions. i.e. Peer-to-Peer Car Knowledge.
In LA they most certainly do. They can't make drivers drive better, but they instrument the hell out of the roads and there are a number of ways for drivers to get detailed real time information. The whole region's freeway system is instrumented and you can get near real time updates of freeway speeds at sigalert.com as well as most of the popular mapping apps. It even gives the dispatch reports from incidents so you know which lanes are blocked. I glance at sigalert and replan my route in an instant due to a collision 15 miles away. The city of LA also just completed integration of city street sensors and lights system wide, and it does work. You can get street speed data on google maps, and there are also crowdsourced apps like waze (which I don't use since it put me on a dead stopped 405) or trapster, that integrate realtime data from drivers on the road.
I temporarily commute from the SGV to the south bay (I'd move if it was permanent), and my experience is consistent with your claim. Getting into DTLA is a mess-- I have a bunch of bailout points that I use (and know in part from biking around the same areas) to at least keep rolling. I basically get off the freeway and take surface streets through DTLA then get back on the 110 outbound, both north and south, and in the morning there's generally less traffic outbound than inbound. I also get to cheat a little because I have a transponder and get reimbursed for tolls-- it saves a lot of time and drive stress, and I'd probably stay overnight a couple nights a week rather than do the round trip every day without it.
It works well in LA, too, where the weather is nearly always ideal. Better station planning and car design could speed the boarding times. Or having more secure bike parking at the ends (shared use lockers or valet services-- there are a few stations in LA that have bike valets).
Building a train parallel to the freeway, especially in Los Angeles, doesn't do anything to relieve congestion. They've already tried that with MetroLink, as well as the Bus Rapid Transit along the 10 freeway, and they are just as clogged as they were before. Oh, and on national average of all US cities with population of 500k or more, a transit commute takes more time than a car commute by a significant amount. Source
That doesn't even get into the financial aspect, where cost per passenger-mile traveled on rail projects is 4x what a car costs when you factor in construction costs for both rail and road, as well as maintenance costs for both rolling stock and car, and personal ownership costs of the car (insurance, title fees, etc.) and fuel.
Outside of incredibly dense population centers, rail just doesn't make sense.
Nothing actually relieves congestion-- you just move more people using less space or resources per person but the same congestion. There's a certain amount of congestion/delay/commute time that people are willing to tolerate, and people will keep saturating transportation modes back to that level no matter what you do to relieve it. New freeways or lanes rapidly return to the previous levels of congestion-- you can't pave your way out of it, but you can provide modes that are more pleasant or more efficient.
Transit being slower than driving in US cities is because most transit is buses that travel in the same lanes as auto traffic, but make more stops. Of course it's slower. Light rail can be quite fast, and in LA light rail plus bicycle can be much faster than driving at peak times. You also can do something like read while you're on the train, which you can't in the car. I live in the Pasadena area, and it's much faster to take the train to DTLA at peak hours than to drive. I temporarily have a commute several times a week to the south bay-- that's faster by car because I'd have to take 4 different trains or bus lines to do it. But traffic is so bad that I can get to Sunnyvale in about the same time it takes me to get to the south bay. Leaving early and coming home after peak make it tolerable for a little while (and it's usually only 3 days/week). I'd move if it was permanent.
My partner commutes (bike+train) sometimes to DTLA, and when she's late getting home it's because someone gave her a ride partway-- it's consistently slower than bike+train. Now that Metro lets you take a bike any time, there are a lot of people who do bike+train commutes. They should really consider a different style car, with bike hooks and flip up seats along one side, and regular seats along the other. It would make it easier to pack more people+bikes in. Shared use bike lockers (which are now available in the bay area) instead of long term rental lockers would also make bike commuting/shopping easier-- they work like parking meters, but keep people from stealing parts off your bike.
SpaceX is located smack in the center of the LA aerospace industry- there's a huge pool of experienced engineers to draw from who have been working around that area for ages.
I've seen an electric crotch rockets that will do at least 55 mph. They're not allowed to go faster than that on the indoor velodrome where we motorpace behind it (generally not nearly that fast). It uses some huge AGM batteries.
I just mapped it, and it comes up about 17-18, but you're still in the right ballpark.
His best bet is to move- as soon as the 405 is built out, it will return to the same congestion as before (that happens to freeways everywhere). The next best thing (and better for LA) would be to fund a rail line that essentially parallels the 405. And maybe throw in a bikeway-- LA has 330 days/year that are good biking weather, but having to do a long commute on city streets can be a pain. There are a few bikeways along the various rivers and/or freeways (SGRT, LARIO) that can make a bike commute competitive with driving, even for very long distances. Shorter than about 10 miles it's faster to bike, and even at 15-20 miles, the combination of bike and train is faster than driving at rush hour.
As a mathematician (now in industry) who knows many other mathematicians, I can assure you many of them are very poorly prepared to handle data or real-world problems.
As a physicist who's known a few mathematicianns, I can assure you, y'all are an odd bunch.
University of Chicago used to have at least a few people doing it-- granular media experiments can be done pretty cheaply, and there was someone there doing theory and experiments on migration of suspensions in droplets as they evaporate that started out with a bunch of experiments using coffee droplets. Poke around 4 year colleges that have good physics departments and there's probably someone doing good physics on the cheap.
On the other hand, the biological sciences are especially tough because experiments are hard, expensive and unreliable, and those doing them typically not so sophisticated with data analyses.
Try low temperature physics...
When I was in grad school I used to ride bikes with a guy who was a biology PhD-- I can't remember if he was a post-doc or staff somewhere. One time we were out and he asked "How many experiments do you do a week?" I almost fell off my bike laughing. I ran my experiment 3 times in 6 years (all in the last 1.5 years), and each time it ran for no less than 4 weeks (I think the longest run was 12 or 13 weeks). But up to the first successful run (as in all the engineering worked and it was possible to get data): design, build, test, fix (hardware, software, and electronics).
"A PhD student has the right to expect a project that generates a decent body of work within those four years."
For a Masters degree, this is acceptable. For a PhD, they had better be coming up with their own idea, a plan, funding, and then have their advisor and committee evaluate during the prospectus defense. Having their topic/project dropped in their lap so they can turn the crank is not what a PhD is all about.
Funding?
There are areas of physics where the cycle time for proposals is 2 years (from announcement to release of funds) with a success rate of less than 10% for even senior people (NIH has an even lower funding rate, and an expectation that most things get proposed a couple times before being funded). Many, if not most, graduate students in science can easily get funding to cover their salary through fellowships/RA positions/TA positions, etc, but the chances of a grad student writing their own grant proposal in most subfields is pretty small. Sure, there are areas where you can do good science with dimestore materials (and a few places that specialize in that), but that's a pretty narrow slice of science in almost any field. Some of the most successful faculty I've known at one of the top science/engineering universities in the world are successful because they let their post-docs be PI on proposals (which is relatively uncommon). Then if the project is awarded the post-doc starts the work as a post-doc and manages to spin it into a faculty job.
Almost everybody I know who got both a masters and PhD in physics did it for one of two reasons-- 1) they started grad work at a smaller school that didn't have a PhD program (or not much of one) and switched to a larger/better program 2) they expected to work at large companies (e.g. 3M) where the pay scale gave you slightly more money if you had a masters+PhD than PhD alone, even if all you did for the masters was fill out a few extra forms and bind up some intermediate result (that you had anyway) into a thesis.
The core of the HR problem is not pay. The core of the problem is lack of women. When you work for a company where the overwhelming majority of the engineers are male (which is pretty much every tech company), assuming you're straight, the opportunities for forming relationships are limited by the nearly nonexistent dating pool.
I work in a place and industry (aerospace R&D) that's overwhelmingly male and know *a lot* of dateable women (I already have a very long term gf, though), virtually all of them outside my workplace and industry. It's not difficult-- just do something for recreation that's even slightly group oriented and not male dominated and they're everywhere.
I have *two* rice cookers in the kitchen. And use them both. One's a little tiny everyday one that does about 2 cups pretty fast. The other is a giant steamer/pressure cooker that we picked up cheap at a garage sale for when we're entertaining and need more than 2 cups of rice, or when we cooked some particularly crusty rice in the other one and it's still soaking to get teh crust off...
It would cost you more to build something to catch it at L2 and refill the dewar than to just build a new one-- it wasn't designed to have any sort of docking and refill capability. It will eventually fall out of L2 into a heliocentric orbit anyway, so they're probably just going to do it in a controlled way.
Slashdot Mirror
My impression from various sources (partly news, partly being in aerospace, though not aircraft) is that for aircraft crashes people often survive the crash and die from the fire (fueled by lots of jet fuel spilled around), so there's a high priority placed on putting out the fire. The news reports even said that the modern airport fire trucks start dousing the fire with foam from a pretty good distance as they rush to the scene.
The foam certainly complicated things, and even if she was standing, a foam jet could have easily knocked her over. It was probably also complicated by the number of people who survived and were managing to get out on their own-- probably not that common in a crash where fire is breaking out. And firefighters don't experience nearly the number of aircraft crash+fire as they would structure fires, so anticipating what's going to happen when a large number of people survive a crash like that is going to be less predictable.
All that said, they should have some kind of standard practice to minimize the chance of something like this happening again. It's apparently harder than it seems-- an example where conditions are nearly perfect and nobody is any hurry is the vehicles patrolling the beachs in SoCal. Every few years someone gets run over by a lifeguard/police/fire vehicle who either wasn't paying close attention while driving or didn't do a walk around before they started up and drove off.
I don't think I've seen a car made in the past decade or so that had a 0-60 time slower than 8 seconds, and FWIW, one could argue that if everyone's car had 0-60 times in the teens, the roads would be even safer.
Prius. 12.9 seconds. And there are plenty of other cars slower than 8 seconds, mostly in the sub $30K price range.
Try merging from a dead stop into dense 60-70 mph traffic with a short merge lane in a car with a 13 s 0-60 time and three passengers-- you might do it but you won't enjoy it. I've been shopping for a replacement for my old saturn wagon and it's hard to find a small car with as much hauling capability, similar gas mileage and as good aacceleration (0-60 time 9.2 s) at a reasonable price (e.g. comparable inflated cost).
I'm not sure putting everyones acceleration in the teens would make things safer, unless cars had an enforced minimum separation distance-- finding and getting into a hole when merging into dense traffic is *way* easier if you have some acceleration.
$2.80 of the $27.95. Even if you completely remove it, you're still talking about a $22.32 ebook. And if you're going through a distributor, you aren't completely removing it - they still take a cut.
But for an ebook you're also eliminating all the costs of the physical space at the retailer. The $12.58 that the retailer gets isn't pure profit-- they have to pay for space, utilities, employees, etc. out of that
And pre-production is generally easier with an ebook than a paper book. You can script both, as GP mentions, but to make a nice looking paper book generally takes more tweaking than a nice looking ebook.
The cassette player died a while ago, necessitating the FM transmitter. It was also an annoying GM cassette player, which required the slightly more expensive cassette adapter that had a short loop of tape to make the player happy so it wouldn't just eject.
You obviously don't live in Los Angeles. There's no such a thing as a 4 car length gap much of the time. At on and off ramps it's common to have multiple cars merging on and off simultaneously with less than a car length between any of them.
My 15 year old car has a cassette player. A CD changer was an option that the dealer could install. My iPhone communicates with the car via FM transmitter in a Jabra Freeway bluetooth speakerphone. The setup is pretty reasonable, though I find it easier to put the speakerphone in the middle of the dash instead of the visor. The FM transmitter on the Freeway isn't as good as the one that was on the Cruzr2, but it handles multiple phones better. Wheel mounted controls would be nice though.
I use a dash-mounted iphone with a touchscreen interface for nav and music in the car, but I have it set up so that I can use tactile feedback for hitting the spots I need while driving. The traffic map and music apps are set up in corners of the screen, so I just need to hit a corner to get to them.
I also just started shopping for cars and was having thoughts similar to the submitter-- why would I want to pay $$ for a built in, dedicated system from the car maker that's already behind what I can do on a phone or a tablet? They should just provide standardized tactile inputs on the wheel, plus more buttons and an optional screen in the dash. There should also be some standardized interface for sending vehicle specific information back to the user's device (e.g. tire pressure, fuel economy, etc.).
The problem we are trying to optimize is: How can we move the most cars (maximize distance) in the least amount of time. i.e. dX/dT. Which looks like a differential equation.
One small but important nit-- you're not trying to move the most cars, but the most people.
It is a shame that the Dept. of Motor Vehicles doesn't know shit about standing waves nor teaches people how to help optimize keeping vehicles moving in the traffic flow. One of these days every car will be able to pass it's current speed both forward and backwards to its neighbor's car so that people 5, 10, 20 mins down the road can know about future traffic conditions. i.e. Peer-to-Peer Car Knowledge.
In LA they most certainly do. They can't make drivers drive better, but they instrument the hell out of the roads and there are a number of ways for drivers to get detailed real time information. The whole region's freeway system is instrumented and you can get near real time updates of freeway speeds at sigalert.com as well as most of the popular mapping apps. It even gives the dispatch reports from incidents so you know which lanes are blocked. I glance at sigalert and replan my route in an instant due to a collision 15 miles away. The city of LA also just completed integration of city street sensors and lights system wide, and it does work. You can get street speed data on google maps, and there are also crowdsourced apps like waze (which I don't use since it put me on a dead stopped 405) or trapster, that integrate realtime data from drivers on the road.
but with a funny accent, nonetheless
I temporarily commute from the SGV to the south bay (I'd move if it was permanent), and my experience is consistent with your claim. Getting into DTLA is a mess-- I have a bunch of bailout points that I use (and know in part from biking around the same areas) to at least keep rolling. I basically get off the freeway and take surface streets through DTLA then get back on the 110 outbound, both north and south, and in the morning there's generally less traffic outbound than inbound. I also get to cheat a little because I have a transponder and get reimbursed for tolls-- it saves a lot of time and drive stress, and I'd probably stay overnight a couple nights a week rather than do the round trip every day without it.
It works well in LA, too, where the weather is nearly always ideal. Better station planning and car design could speed the boarding times. Or having more secure bike parking at the ends (shared use lockers or valet services-- there are a few stations in LA that have bike valets).
And treat the wide line between the carpool lane and the number 1 lane as their own lane when both of those are stopped.
Building a train parallel to the freeway, especially in Los Angeles, doesn't do anything to relieve congestion. They've already tried that with MetroLink, as well as the Bus Rapid Transit along the 10 freeway, and they are just as clogged as they were before. Oh, and on national average of all US cities with population of 500k or more, a transit commute takes more time than a car commute by a significant amount. Source
That doesn't even get into the financial aspect, where cost per passenger-mile traveled on rail projects is 4x what a car costs when you factor in construction costs for both rail and road, as well as maintenance costs for both rolling stock and car, and personal ownership costs of the car (insurance, title fees, etc.) and fuel.
Outside of incredibly dense population centers, rail just doesn't make sense.
Nothing actually relieves congestion-- you just move more people using less space or resources per person but the same congestion. There's a certain amount of congestion/delay/commute time that people are willing to tolerate, and people will keep saturating transportation modes back to that level no matter what you do to relieve it. New freeways or lanes rapidly return to the previous levels of congestion-- you can't pave your way out of it, but you can provide modes that are more pleasant or more efficient.
Transit being slower than driving in US cities is because most transit is buses that travel in the same lanes as auto traffic, but make more stops. Of course it's slower. Light rail can be quite fast, and in LA light rail plus bicycle can be much faster than driving at peak times. You also can do something like read while you're on the train, which you can't in the car. I live in the Pasadena area, and it's much faster to take the train to DTLA at peak hours than to drive. I temporarily have a commute several times a week to the south bay-- that's faster by car because I'd have to take 4 different trains or bus lines to do it. But traffic is so bad that I can get to Sunnyvale in about the same time it takes me to get to the south bay. Leaving early and coming home after peak make it tolerable for a little while (and it's usually only 3 days/week). I'd move if it was permanent.
My partner commutes (bike+train) sometimes to DTLA, and when she's late getting home it's because someone gave her a ride partway-- it's consistently slower than bike+train. Now that Metro lets you take a bike any time, there are a lot of people who do bike+train commutes. They should really consider a different style car, with bike hooks and flip up seats along one side, and regular seats along the other. It would make it easier to pack more people+bikes in. Shared use bike lockers (which are now available in the bay area) instead of long term rental lockers would also make bike commuting/shopping easier-- they work like parking meters, but keep people from stealing parts off your bike.
SpaceX is located smack in the center of the LA aerospace industry- there's a huge pool of experienced engineers to draw from who have been working around that area for ages.
His neighbors in Bel Air will complain about the noise.
I've seen an electric crotch rockets that will do at least 55 mph. They're not allowed to go faster than that on the indoor velodrome where we motorpace behind it (generally not nearly that fast). It uses some huge AGM batteries.
I just mapped it, and it comes up about 17-18, but you're still in the right ballpark.
His best bet is to move- as soon as the 405 is built out, it will return to the same congestion as before (that happens to freeways everywhere). The next best thing (and better for LA) would be to fund a rail line that essentially parallels the 405. And maybe throw in a bikeway-- LA has 330 days/year that are good biking weather, but having to do a long commute on city streets can be a pain. There are a few bikeways along the various rivers and/or freeways (SGRT, LARIO) that can make a bike commute competitive with driving, even for very long distances. Shorter than about 10 miles it's faster to bike, and even at 15-20 miles, the combination of bike and train is faster than driving at rush hour.
As a mathematician (now in industry) who knows many other mathematicians, I can assure you many of them are very poorly prepared to handle data or real-world problems.
As a physicist who's known a few mathematicianns, I can assure you, y'all are an odd bunch.
University of Chicago used to have at least a few people doing it-- granular media experiments can be done pretty cheaply, and there was someone there doing theory and experiments on migration of suspensions in droplets as they evaporate that started out with a bunch of experiments using coffee droplets. Poke around 4 year colleges that have good physics departments and there's probably someone doing good physics on the cheap.
On the other hand, the biological sciences are especially tough because experiments are hard, expensive and unreliable, and those doing them typically not so sophisticated with data analyses.
Try low temperature physics...
When I was in grad school I used to ride bikes with a guy who was a biology PhD-- I can't remember if he was a post-doc or staff somewhere. One time we were out and he asked "How many experiments do you do a week?" I almost fell off my bike laughing. I ran my experiment 3 times in 6 years (all in the last 1.5 years), and each time it ran for no less than 4 weeks (I think the longest run was 12 or 13 weeks). But up to the first successful run (as in all the engineering worked and it was possible to get data): design, build, test, fix (hardware, software, and electronics).
"A PhD student has the right to expect a project that generates a decent body of work within those four years."
For a Masters degree, this is acceptable. For a PhD, they had better be coming up with their own idea, a plan, funding, and then have their advisor and committee evaluate during the prospectus defense. Having their topic/project dropped in their lap so they can turn the crank is not what a PhD is all about.
Funding?
There are areas of physics where the cycle time for proposals is 2 years (from announcement to release of funds) with a success rate of less than 10% for even senior people (NIH has an even lower funding rate, and an expectation that most things get proposed a couple times before being funded). Many, if not most, graduate students in science can easily get funding to cover their salary through fellowships/RA positions/TA positions, etc, but the chances of a grad student writing their own grant proposal in most subfields is pretty small. Sure, there are areas where you can do good science with dimestore materials (and a few places that specialize in that), but that's a pretty narrow slice of science in almost any field. Some of the most successful faculty I've known at one of the top science/engineering universities in the world are successful because they let their post-docs be PI on proposals (which is relatively uncommon). Then if the project is awarded the post-doc starts the work as a post-doc and manages to spin it into a faculty job.
Almost everybody I know who got both a masters and PhD in physics did it for one of two reasons-- 1) they started grad work at a smaller school that didn't have a PhD program (or not much of one) and switched to a larger/better program 2) they expected to work at large companies (e.g. 3M) where the pay scale gave you slightly more money if you had a masters+PhD than PhD alone, even if all you did for the masters was fill out a few extra forms and bind up some intermediate result (that you had anyway) into a thesis.
The core of the HR problem is not pay. The core of the problem is lack of women. When you work for a company where the overwhelming majority of the engineers are male (which is pretty much every tech company), assuming you're straight, the opportunities for forming relationships are limited by the nearly nonexistent dating pool.
I work in a place and industry (aerospace R&D) that's overwhelmingly male and know *a lot* of dateable women (I already have a very long term gf, though), virtually all of them outside my workplace and industry. It's not difficult-- just do something for recreation that's even slightly group oriented and not male dominated and they're everywhere.
I have *two* rice cookers in the kitchen. And use them both. One's a little tiny everyday one that does about 2 cups pretty fast. The other is a giant steamer/pressure cooker that we picked up cheap at a garage sale for when we're entertaining and need more than 2 cups of rice, or when we cooked some particularly crusty rice in the other one and it's still soaking to get teh crust off...
It would cost you more to build something to catch it at L2 and refill the dewar than to just build a new one-- it wasn't designed to have any sort of docking and refill capability. It will eventually fall out of L2 into a heliocentric orbit anyway, so they're probably just going to do it in a controlled way.