Boeing doesn't do development work without a contract. So, when they got a contract to start development of their capsule, they started.
And then they stopped working on it as soon as the contract ran out. They're waiting on a new contract to resume work.
And this is a part of why aerospace/defense contractors are so expensive for the government to employ, and I say that having worked at some of them. Stringing together multiple contract awards to try to get to a larger goal, and continually going through the proposal writing process. A commercial company with a goal can strive continuously towards it in a more efficient manner, and the goal is success, not hours billed on a cost plus contract.
The nanoparticles are magnetic, not magnets, which is an important different. It means that the nanoparticles will be attracted to an external magnetic field when it is applied, but they will not be attracted to each other.
Buckyballs were banned because if you swallow permanent magnets they can attract each other and could potentially pinch two parts of your intestine together, or other such unpleasant things which would be bad for you.
Swallowing permanent magnets: Bad idea. Swallowing magnetic nanoparticles: Good idea assuming it passes the relevant medical trials for safety and effectiness.
It's my understanding that the electricity for the trolley buses is free.
If San Fran owns the power plant it means that they have paid for the cost of construction and ongoing operating and staffing costs, and by using the power internally rather than selling it they are paying the opportunity cost of not selling it and making money.
Free in this case just means a cost that appears in a different ledger.
Well in addition to falling apart in your hands there are some concerns that free graphene can be carcinogenic in a similar fashion to asbestos... so I'll let you try the hammock first;)
If it's a fiction it's a brilliant one which helps this story stand out from all of the other "Fantastic New Innovation in Graphene Which Will Lead to Large Scale Production" press releases that get put out every year.
The summary was not clear so here is my version based on my understanding of the work:
Graphite is composed of many randomly oriented and sized layers of graphene. Intercalation is a process where compounds or ions can be inserted between the layers of a layered material such as graphite. These Penn State reseachers have discovered a new way to perform this intercalation by leaving out a strong oxidizing agent which was thought to be necessary but would damage the graphene.
The research advisor Thomas E. Mallouk suggested trying it without the oxidizer. The researcher Nina Kovtyukhona was reluctant to perform this experiment as she thought it would be a failure. Her advisor persuaded her to try it by making a bet that he would pay her $100 if it succeeded, and she would pay him $10 if she failed. The experiment was a success, and researchers now have a new avenue to explore for synthesizing graphene.
My personal thought is that while this is scientifically interesting and could lead to some engineering benefits down the road this will not lead to large scale production of graphene since it is just splitting apart graphene sheets from graphite, and these sheets are generally quite small. Large scale production to me would be getting fairly good uniform growth or deposition of graphene over large areas of a substrate which is of the order of 1cm or larger so that it can eventually be scaled up to the 30cm and 45cm wafers in the silicon industry. So unless other researchers come up with a way to make a graphene boule composed of decent sized graphene sheets this technique does not seem useful for commercial electronics.
One things which makes graphene research expensive is that most growth methods end up with just little flecks of graphene material in random locations on a substrate, so a researcher or technician has to manually search for these and place contacts and gates on them using a manual lithography tool. It could even be automated but this would still be orders or magnitude slower than competing technologies.
A new route to making graphene has been discovered that could make the 21st century's wonder material easier to ramp up to industrial scale.
Whenever a press release uses language like this I am forced to point out that graphene so far has had zero compelling results for electronics applications. It is soundly beat by silicon and III-V semiconductors in terms of speed and dynamic range. Graphene transistors can be made reasonably fast (for certain but not all definitions of fast) but even so the signals that they can handle are only very tiny because of the lack of a bandgap. It has some wonderful properties but also some terrible ones which make its applicability suspect.
Why are guys who run factories employing tons of US citizens in US based factories (like Toyota) who produce super reliable product with great mileage get slapped by the media when a bogus story about a gas pedal getting stuck?
Are you kidding? Just wait until Tesla slips up. They will eventually, and the media will jump all over them. The only thing that the media loves more than an underdog is the story of a fallen angel.
Media are jumping all over Tesla already.
It's happened with the Tesla car fires, which are still less common per mile driven than gasoline car fires. New York Times published a likely faked review where Tesla's data logging disagrees with the reporters account, and shows that the reporter only charged the battery to a small fraction of the capacity and ignored the low battery warnings while driving past charging stations. Top Gear also faked battery trouble on their show (it was apparently in the script before they even received the car from Tesla).
When I was an undergrad, most exams in advanced science and engineering classes allowed you to bring ANYTHING as long as it didn't involve communication with people outside the room. Forget about just calculators (ANY calculator), some people would be STACKS of textbooks, and I even remember some laptops (though those were less common back then -- largescale wireless also didn't quite exist yet).
When I first had a test like this, I packed a pile of books too, along with whatever calculator I had (I think a TI-85), etc. But I quickly realized that most of this was useless. In the limited time we had, if I didn't already know the stuff, I'm not going to have time to learn it from a book.
Same here, many engineering classes I had were open book, but a few profs had closed books and some didn't even allow calculators.
I had one excellent engineering prof who allowed one page of notes to exams, but no calculator. Most of the questions were asked in a way such that there was little to no numeric computation required, and when he did request a numeric answer it was simple stuff that he expected us to be able to simply and quickly do in our head or on paper. His philosophy was that you need to understand the problem well enough and break it down to analyzable pieces sometimes with approximations so that you can get within ~5% of the correct numeric answer by hand analysis, and if higher accuracy is required it will generally be optimized on the computer with software you bought or wrote.
The homework problems would cover more rigorous computations and computer simulations, the tests were designed to see if you truly understood the problems at hand.
Are there e-paper screens that are fast enough? On my Kindle there is a noticeable lag for any screen update, and periodic total screen refreshes are required. That lag would be very annoying when entering formulas, numbers, or programming.
Indeed. Using a CPU as a heater is just silly, it would be like frying an egg on the engine block of your Ferrari: technically possible but there are better and cheaper ways to get the same thing done. It wouldn't be that hard to hack together a decent computer controlled heater and sample holder/heat exchanger from a few hundred to a thousand bucks in parts (depending on how much labor you want to put into it).
Instead they have this fiddly system where they have to load samples onto the heatsink of a running computer with the motherboard out in the open, and they must be careful not to drop anything onto the motherboard which could damage it, there are ESD damage concerns, and liquid damage concerns. The temperature control cannot be that good, given the vertical temperature gradient along the heatsink base to top.
It's also disingenuous (but common) for them to merely compare the raw part cost of a DIY system to a pro system. The Pro system is much more capable and robust, support is available, and it includes the labor cost of designing and assembling it. Too many DIY projects claim a "cost" which treats labor as a free commodity.
Ever since I can remember, we have been bombarded by th e concept that there is a scale of healthy eating, and that elimination of as much fat as possible is desirable, and the ne plus ultra of healthy living is veganism, followed by vegatarianism, then low fat/high carb, then the unwashed masses of high protein, and the soulles spawn of Satan - the Atkinists.
You seem older than I and so our perspectives are a bit different. I started cooking my own food around 10 years ago. I always got the impression that vegan/vegetarian lifestyles *could* be quite healthy, but that this had more to do with the fact that these folks are more likely to put more thought and planning into their diet than the average america, eating more fruit and veg, and not eating as many empty calories in general.
Once the genie's out of the bottle it's over. Apparently Lawrence's lawyers are threatening to sue or refer for criminal prosecution anyone that shares the photos of her. I very much doubt they'll have a lot of luck though, even if the original source of the leak is found. If anything they're just going to make it worse via Streisand Effect.
Seriously, the Streisand Effect? This is already all over the internet and news, going after those responsible for it won't change that in the slightest.
I'm quite sure they're not idiots who really think this is a freakin' 2D plane.
Not to be pedantic, but from the same paper abstract but two sentences ahead of what you quoted:
Creating analogous heterojunctions between different 2D semiconductors would enable band engineering within the 2D plane
Which shows that they describe the full device including multiple atomic layers as a 2D plane.
Now the researchers obviously know that this is not a 2D device, in the same way that graphene researchers know that graphene is not 2D (put a gate dielectric and a gate on it and you have a very much 3D transistor). Except at least in graphene the transport is 2D, for the device is this paper there is vertical transport as well. These researchers are simply jumping on the "2D" buzzword bandwagon because it's a hot research topic.
I happen to dislike a lot of what the 2D folks are claiming in the media because they tend to make very hyped claims about their device performance. And most people understand that 2D materials are not really enablers for flexible electronics or transparent electronics because you can take humble silicon or other semiconductors down to similar thicknesses and achieve similar flexibility and transparency. Or you can make a really tiny silicon device that doesn't need to be flexible even when mounted on a flexible board. Flexible electronics is really all about packaging and interconnect (not to mention a killer app).
This is something where engineers know things that climate scientists apparently don't.
If the positive feedback was so strong, that the system was unstable (right half plane as it were) the earth would already be Venus. Doesn't stop climatologists talking out of the butts and proposing just such strong positive feed-backs.
TL;DR; Don't ask a climatologist a control systems question and expect a reasonable answer.
Real engineers know that reality is rarely a linear system represented only by poles and zeros. And they also know that complex systems can have multiple locally stable points, and that even stable systems may "ring" when perturbed with potentially catastrophic consequences.
Consider the humble electronic oscillator which most certainly has a right half plane zero, does its output rise over time until it blows up the universe (aka turns into Venus)? No it doesn't. And to describe the magnitude of the oscillation generally requires the nonlinear transistor model which includes things like the maximum output current of the transistor, it is not a simple linear model like the one you propose.
Earth is quite complex and while an engineer could model it as a control system it would be much more complex than the one you have proposed.
TL;DR Engineers are just as capable of being incorrect as climate scientists are.
Actually Mikey gives four reasons for the healthcare.gov problems, but the summary just focuses on the last one, probably because it sounds funny:
The original points (as summed up by me in a few words) were (1) Fragmentation of implementation, (2) Lack of monitoring of system, (3) Lack of experience by the companies building it, and (4) workplace culture clash.
Yes and if you read their paper an RF switch is used to adjust the terminating impedance of this antenna which perturbs the reflectivity. That RF switch will consume a finite energy and power. It is an active component and not a passive component. A passive component would not accomplish the modulation.
I think it sounds reasonable that a reflector like this might be able to use less energy per bit than a conventional high-efficiency transmitter, since the RF power is being provided elsewhere, this is effectively just a modulator inserted between (or to the side of) the transmitter and receiver. However the study authors do not perform any theoretical analysis or measurement in their paper to support this notion.
Although this device takes no internal power to cause the back scatter effect
That's incorrect, a switch always requires energy to change state, as well as a usually very small amount of power to maintain at least one of the two states. Quite a bit of a CPU's power dissipation comes from the energy consumed by CMOS switches switching from a one state to another.
It may well be that this is more power efficient than other methods of transmitting information but that has yet to be backed up by theory or demonstration.
It only works because it has a very low bitrate of 1kbps:
The UW’s Wi-Fi backscatter tag has communicated with a Wi-Fi device at rates of 1 kilobit per second with about 2 meters between the devices.
Although the authors claim that "The Wi-Fi Backscatter tags do not require any batteries and can harvest energy from ambient RF signals" they make no attempt to back up this claim with measured or estimated energy efficiency of this transmitter. The standard metric for high efficiency transceivers is joules per bit, because low bitrate communication always consumes less energy than high speed, but the only useful way to compare it to another high efficiency transmitter is to see if it can transmit a certain amount of data for less energy.
While I don't expect every paper to address every aspect of a technology, they should not then turn around and make baseless claims like "We believe that this new capability is critical for the commercial adoption of RF-powered Internet of Things." in a length 12 page paper that fails to address the one metric which would allow them to make such a claim.
Competition is also good for business because it increases the total size of the market, increases availability and reduces costs from 3rd party vendors (batteries, motor drives, charging stations, etc), and reduces costs of specialized labor (eventually) by growing the labor pool of engineers who design electric vehicles.
Having a big part of a small market is not a good way to go. I think Elon's goal (in addition to any altruism on his part) is to grow the entire market and ecosystem for electric vehicles while holding onto a decent (but perhaps smaller) market share.
This is China we're talking about, with 1.3 billion people. I'm sure that people have already strapped filters onto fans, this one guy just happens to be getting some feel-good-DIY publicity because he's marketing his product. Kudos to him because it seems like a good thing that he's doing this, but I would be amazed if this is actually considered novel in China.
Are you sure about that? Other than this one article what makes you think that all or most Chinese citizens with air filtering fans were paying $1000 for them?
I agree that this doesn't pass the sniff test. A foam has a low ratio of metal to air, it's the cross-sectional area of the copper that allow vertical conduction of the heat from the heat spreader plate. This has little copper area so conduction is limited. It's further worsened by the random nature of the strands so the heat is conducted laterally as well as vertically and so the conductive thermal resistance is increased because the heat needs to travel a longer path.
So while I'm not saying it doesn't work, I suspect that this foam is more of a gimmick than a properly optimized heatsinking solution.
I get that the meteoric rise in online video streaming by customers puts pressure on ISPs because if affects the oversubsribe ratio that they can use (which is required to turn a profit) while still providing a good user experience.
But what I don't get is why you can possibly blame Netflix. Your customer requested 100GB from Netflix last month. Netflix supplied it based on your customer's request. If you think 100GB (or however much) data in one month is too much then throttle your customer, but do it fairly based on each customer's usage and don't play favorites with which companies you allow your customers access to. 100GB of Netflix traffic should be treated the same as 100GB of porn, or whatever else your customer is getting up to.
My university's residential internet connection started undergoing major strain several years ago, primarily due to online video. So they implemented traffic throttling. I don't remember the precise details, but it was along the lines of a daily 1GB of unthrottled data between 4PM-1AM after which speed was reduced , and no throttling from 1AM-4PM. This was a completely fair and balanced way of providing a pretty good user experience while limiting traffic during peak hours to avoid congesting the network.
How does settlement usually work for unbalanced connections? My thought is that since the Verizon customers request data from Netflix servers, and Level 3 delivers that data to the Verizon network, then if anybody paid settlement it should be Verizon.
But since this is an everybody-wins situation (Netflix pays level 3, Customers pay Verizon, all customers are happy when the pipes run smoothly) then settlement free peering makes sense.
Slashdot Mirror
And this is a part of why aerospace/defense contractors are so expensive for the government to employ, and I say that having worked at some of them. Stringing together multiple contract awards to try to get to a larger goal, and continually going through the proposal writing process. A commercial company with a goal can strive continuously towards it in a more efficient manner, and the goal is success, not hours billed on a cost plus contract.
The nanoparticles are magnetic, not magnets, which is an important different. It means that the nanoparticles will be attracted to an external magnetic field when it is applied, but they will not be attracted to each other.
Buckyballs were banned because if you swallow permanent magnets they can attract each other and could potentially pinch two parts of your intestine together, or other such unpleasant things which would be bad for you.
Swallowing permanent magnets: Bad idea.
Swallowing magnetic nanoparticles: Good idea assuming it passes the relevant medical trials for safety and effectiness.
If San Fran owns the power plant it means that they have paid for the cost of construction and ongoing operating and staffing costs, and by using the power internally rather than selling it they are paying the opportunity cost of not selling it and making money.
Free in this case just means a cost that appears in a different ledger.
Well in addition to falling apart in your hands there are some concerns that free graphene can be carcinogenic in a similar fashion to asbestos... so I'll let you try the hammock first ;)
So I did. Thanks for the correction.
If it's a fiction it's a brilliant one which helps this story stand out from all of the other "Fantastic New Innovation in Graphene Which Will Lead to Large Scale Production" press releases that get put out every year.
The summary was not clear so here is my version based on my understanding of the work:
My personal thought is that while this is scientifically interesting and could lead to some engineering benefits down the road this will not lead to large scale production of graphene since it is just splitting apart graphene sheets from graphite, and these sheets are generally quite small. Large scale production to me would be getting fairly good uniform growth or deposition of graphene over large areas of a substrate which is of the order of 1cm or larger so that it can eventually be scaled up to the 30cm and 45cm wafers in the silicon industry. So unless other researchers come up with a way to make a graphene boule composed of decent sized graphene sheets this technique does not seem useful for commercial electronics.
One things which makes graphene research expensive is that most growth methods end up with just little flecks of graphene material in random locations on a substrate, so a researcher or technician has to manually search for these and place contacts and gates on them using a manual lithography tool. It could even be automated but this would still be orders or magnitude slower than competing technologies.
Whenever a press release uses language like this I am forced to point out that graphene so far has had zero compelling results for electronics applications. It is soundly beat by silicon and III-V semiconductors in terms of speed and dynamic range. Graphene transistors can be made reasonably fast (for certain but not all definitions of fast) but even so the signals that they can handle are only very tiny because of the lack of a bandgap. It has some wonderful properties but also some terrible ones which make its applicability suspect.
Media are jumping all over Tesla already.
It's happened with the Tesla car fires, which are still less common per mile driven than gasoline car fires. New York Times published a likely faked review where Tesla's data logging disagrees with the reporters account, and shows that the reporter only charged the battery to a small fraction of the capacity and ignored the low battery warnings while driving past charging stations. Top Gear also faked battery trouble on their show (it was apparently in the script before they even received the car from Tesla).
Same here, many engineering classes I had were open book, but a few profs had closed books and some didn't even allow calculators.
I had one excellent engineering prof who allowed one page of notes to exams, but no calculator. Most of the questions were asked in a way such that there was little to no numeric computation required, and when he did request a numeric answer it was simple stuff that he expected us to be able to simply and quickly do in our head or on paper. His philosophy was that you need to understand the problem well enough and break it down to analyzable pieces sometimes with approximations so that you can get within ~5% of the correct numeric answer by hand analysis, and if higher accuracy is required it will generally be optimized on the computer with software you bought or wrote.
The homework problems would cover more rigorous computations and computer simulations, the tests were designed to see if you truly understood the problems at hand.
Are there e-paper screens that are fast enough? On my Kindle there is a noticeable lag for any screen update, and periodic total screen refreshes are required. That lag would be very annoying when entering formulas, numbers, or programming.
Indeed. Using a CPU as a heater is just silly, it would be like frying an egg on the engine block of your Ferrari: technically possible but there are better and cheaper ways to get the same thing done. It wouldn't be that hard to hack together a decent computer controlled heater and sample holder/heat exchanger from a few hundred to a thousand bucks in parts (depending on how much labor you want to put into it).
Instead they have this fiddly system where they have to load samples onto the heatsink of a running computer with the motherboard out in the open, and they must be careful not to drop anything onto the motherboard which could damage it, there are ESD damage concerns, and liquid damage concerns. The temperature control cannot be that good, given the vertical temperature gradient along the heatsink base to top.
It's also disingenuous (but common) for them to merely compare the raw part cost of a DIY system to a pro system. The Pro system is much more capable and robust, support is available, and it includes the labor cost of designing and assembling it. Too many DIY projects claim a "cost" which treats labor as a free commodity.
You seem older than I and so our perspectives are a bit different. I started cooking my own food around 10 years ago. I always got the impression that vegan/vegetarian lifestyles *could* be quite healthy, but that this had more to do with the fact that these folks are more likely to put more thought and planning into their diet than the average america, eating more fruit and veg, and not eating as many empty calories in general.
Seriously, the Streisand Effect? This is already all over the internet and news, going after those responsible for it won't change that in the slightest.
Not to be pedantic, but from the same paper abstract but two sentences ahead of what you quoted:
Which shows that they describe the full device including multiple atomic layers as a 2D plane.
Now the researchers obviously know that this is not a 2D device, in the same way that graphene researchers know that graphene is not 2D (put a gate dielectric and a gate on it and you have a very much 3D transistor). Except at least in graphene the transport is 2D, for the device is this paper there is vertical transport as well. These researchers are simply jumping on the "2D" buzzword bandwagon because it's a hot research topic.
I happen to dislike a lot of what the 2D folks are claiming in the media because they tend to make very hyped claims about their device performance. And most people understand that 2D materials are not really enablers for flexible electronics or transparent electronics because you can take humble silicon or other semiconductors down to similar thicknesses and achieve similar flexibility and transparency. Or you can make a really tiny silicon device that doesn't need to be flexible even when mounted on a flexible board. Flexible electronics is really all about packaging and interconnect (not to mention a killer app).
Real engineers know that reality is rarely a linear system represented only by poles and zeros. And they also know that complex systems can have multiple locally stable points, and that even stable systems may "ring" when perturbed with potentially catastrophic consequences.
Consider the humble electronic oscillator which most certainly has a right half plane zero, does its output rise over time until it blows up the universe (aka turns into Venus)? No it doesn't. And to describe the magnitude of the oscillation generally requires the nonlinear transistor model which includes things like the maximum output current of the transistor, it is not a simple linear model like the one you propose.
Earth is quite complex and while an engineer could model it as a control system it would be much more complex than the one you have proposed.
TL;DR Engineers are just as capable of being incorrect as climate scientists are.
Actually Mikey gives four reasons for the healthcare.gov problems, but the summary just focuses on the last one, probably because it sounds funny:
The original points (as summed up by me in a few words) were (1) Fragmentation of implementation, (2) Lack of monitoring of system, (3) Lack of experience by the companies building it, and (4) workplace culture clash.
Yes and if you read their paper an RF switch is used to adjust the terminating impedance of this antenna which perturbs the reflectivity. That RF switch will consume a finite energy and power. It is an active component and not a passive component. A passive component would not accomplish the modulation.
I think it sounds reasonable that a reflector like this might be able to use less energy per bit than a conventional high-efficiency transmitter, since the RF power is being provided elsewhere, this is effectively just a modulator inserted between (or to the side of) the transmitter and receiver. However the study authors do not perform any theoretical analysis or measurement in their paper to support this notion.
That's incorrect, a switch always requires energy to change state, as well as a usually very small amount of power to maintain at least one of the two states. Quite a bit of a CPU's power dissipation comes from the energy consumed by CMOS switches switching from a one state to another.
It may well be that this is more power efficient than other methods of transmitting information but that has yet to be backed up by theory or demonstration.
It only works because it has a very low bitrate of 1kbps:
Although the authors claim that "The Wi-Fi Backscatter tags do not require any batteries and can harvest energy from ambient RF signals" they make no attempt to back up this claim with measured or estimated energy efficiency of this transmitter. The standard metric for high efficiency transceivers is joules per bit, because low bitrate communication always consumes less energy than high speed, but the only useful way to compare it to another high efficiency transmitter is to see if it can transmit a certain amount of data for less energy.
While I don't expect every paper to address every aspect of a technology, they should not then turn around and make baseless claims like "We believe that this new capability is critical for the commercial adoption of RF-powered Internet of Things." in a length 12 page paper that fails to address the one metric which would allow them to make such a claim.
Competition is also good for business because it increases the total size of the market, increases availability and reduces costs from 3rd party vendors (batteries, motor drives, charging stations, etc), and reduces costs of specialized labor (eventually) by growing the labor pool of engineers who design electric vehicles.
Having a big part of a small market is not a good way to go. I think Elon's goal (in addition to any altruism on his part) is to grow the entire market and ecosystem for electric vehicles while holding onto a decent (but perhaps smaller) market share.
Probably because he doesn't live in China.
This is China we're talking about, with 1.3 billion people. I'm sure that people have already strapped filters onto fans, this one guy just happens to be getting some feel-good-DIY publicity because he's marketing his product. Kudos to him because it seems like a good thing that he's doing this, but I would be amazed if this is actually considered novel in China.
Are you sure about that? Other than this one article what makes you think that all or most Chinese citizens with air filtering fans were paying $1000 for them?
I agree that this doesn't pass the sniff test. A foam has a low ratio of metal to air, it's the cross-sectional area of the copper that allow vertical conduction of the heat from the heat spreader plate. This has little copper area so conduction is limited. It's further worsened by the random nature of the strands so the heat is conducted laterally as well as vertically and so the conductive thermal resistance is increased because the heat needs to travel a longer path.
So while I'm not saying it doesn't work, I suspect that this foam is more of a gimmick than a properly optimized heatsinking solution.
I get that the meteoric rise in online video streaming by customers puts pressure on ISPs because if affects the oversubsribe ratio that they can use (which is required to turn a profit) while still providing a good user experience.
But what I don't get is why you can possibly blame Netflix. Your customer requested 100GB from Netflix last month. Netflix supplied it based on your customer's request. If you think 100GB (or however much) data in one month is too much then throttle your customer, but do it fairly based on each customer's usage and don't play favorites with which companies you allow your customers access to. 100GB of Netflix traffic should be treated the same as 100GB of porn, or whatever else your customer is getting up to.
My university's residential internet connection started undergoing major strain several years ago, primarily due to online video. So they implemented traffic throttling. I don't remember the precise details, but it was along the lines of a daily 1GB of unthrottled data between 4PM-1AM after which speed was reduced , and no throttling from 1AM-4PM. This was a completely fair and balanced way of providing a pretty good user experience while limiting traffic during peak hours to avoid congesting the network.
How does settlement usually work for unbalanced connections? My thought is that since the Verizon customers request data from Netflix servers, and Level 3 delivers that data to the Verizon network, then if anybody paid settlement it should be Verizon.
But since this is an everybody-wins situation (Netflix pays level 3, Customers pay Verizon, all customers are happy when the pipes run smoothly) then settlement free peering makes sense.