Anyone else have a problem with studies that simply study the results of other studies? It would be like a news channel that only reports the news stories that other news channels provide, then based on analysis of trends seen offers an opinion on what is fact or fiction. While convenient, as one does not have to get out and do actual reporting, (or research in this case), I find it leaves much to be desired. It would be better in my opinion to research available data, summarize it, offer an opinion as to its cause, develop a methodology to test this opinion, and perform a carefully thought out study to test if this opinion is supported or refuted by this study.
Obviously you need primary data, but, if done carefully and well, this type of metastudy can be very powerful. Note that they are taking raw data from the previous studies, not just recording their conclusions. For one thing, any study in this area is at risk of statistical error, and the metastudy effectively has a much larger sample size, making this much less severe. Since the studies were performed by many different groups with different funding and at different times, they also reduce the risk of many kinds of systematic error.
They don't go on to suggest hypotheses and test them by new primary studies, because that's not what they are good at. They publish their data and statistical conclusions from the data so that others can do those things. Also, it took them six years to get this lot together, if they wait for follow-on studies before publishing it will take many more years.
In many areas of modern science, collection of data and investigation of hypotheses using the data are becoming separated and this is a good thing! Firstly, if your project is simply to collect a large high quality data set, you are much more motivated to worry about quality of the data, and you can't bias ti towards the conclusions you want, because you don't have any. Secondly the data is open, so others can repeat the later steps, even if regathering the data is too expensive or impossible (you can't go back and observe the same supernova twice).
- between chronic depression, caused by imbalances in neurotransmitter production or reuptake, and depression caused by living in depressing circumstances.
This distinction is not as clear-cut as you would like. Either can cause the other. Being persistently unhappy for valid reasons develops emotional habits which are encoded in neurotransmitter patterns, while neurological malfunctions can easily cause your life to fall apart. Many patients will need treatment for both.
There are a number of projects out there to develop specialised telescopes that will be able to take quite high resolution photos of unprecedentedly large areas of the sky at once, and big enough to gather enough light to show reasonably faint objects without needing too long an exposure. Look at the Large Synoptic Survey Telescope, for instance https://www.lsst.org/. This aims to photograph all of the "available" sky (it's in Chile, so it never sees the stars around the North celestial pole) every few nights for 10 years. There's lots of infomation on their site and in their papers, but a few numbers that jumped out at me: 8.4m primary mirror, 3.2 GPixel camera, 15 TB of data each night!
Even this would have to get moderately lucky to see a supernova as young as this one, which was captured in it's first minutes or hours. It would also, ideally, need to identify what it was seeing almost instantly, so that it (or another telescope) could start a follow-up within seconds or minutes.
Read the article, or even the linked Lancet article, which says:
Funding National Institute for Health Research Oxford Health Biomedical Research Centre and the Japan Society for the Promotion of Science.
Which I think means 100% government funding. It also gives the exact methology -- a systematic search for, and integration of data from, all published studies that met stated criteria.
The most likely site for a Moon base is the South pole. There are crater rims there that are in permanent sunlight, which the crater bottoms are in permanent shadow (and appear to have significant quantities of moisture frozen out there.
All of which said, until you have in situ manufacturing, a compact fission reactor will give you a lot more power per ton hauled up from Earth.
I must admit I'd rather that the pizzas were delivered to someone a few blocks away, while I just watched. The collateral damage from actually getting pizza this way is a bit much
I would guess it'll go like chip making technology. Basically all the players own at least some patents that all the other players need (or at least really want) so they all license them to each other. If your R&D did really well, you own more (or better patents) you make some money off this process, if your R&D didn't luck out, it costs you some money. But it doesn't really stop anyone getting access to anuything
I;ve seen convincing arguments that airline pilots have overall negative safety value. More accidents are caused by pilots messing up or suiciding han are prevented by pilots being awesome.
So set up "Fred's privacy-enhanced vehicle service" and compete. It'll cost a little more (because that information is actually useful for something), but you can get the EFF or whoever to certify your privacy policy and see how much people actually care
This sort of taste is surprisingly volatile People used to like owning CDs and DVDs, but now we have spotify and netflix. My guess is you will have a period when people still own cars, but ALSO subscribe to transportation-as-a-service services (for the odd occasion when you need a bigger car, or two cars at once, or whatever) and then it will just be more convenient to use the service than try to remember where your car is and have it drive to where you need in time for whatever you need it for and pretty quickly ownership will fall away to become a niche thing, for a small number of enthusiasts or people who live in the middle of huge deserts.
Can't wait to see this in practice. Without manual controls, how do you adjust it's position in a parking space or garage? How would you go thru a typical drive up window? How would you maneuver around a barricade, accident or temporary police detour? How about going around and around a parking lot or underground garage?
At a guess, you select the appropriate view on the dasboard console and drag the rectangle that denotes the car to where you want it to go. A bit like the current systems that show you a rear view camera with the path of the car marked on it, but with the screen controlling the steering instead of vice versa. These cars will have radar or lidar and lots of cameras, as well as GPS.
I don't know the exact cost of this, but given it's an addon to an existing launch anyway, probably only tens of millions. Any serious program of manned exploration beyond Earth orbit is going to cost tens of billions at the very least, probably more.
I'm not sure what you regard as the difference between a fix and a mitigation. If intel changed the processor so that no memory accesses were launched until the CPU knew for sure if the memory management was going to permit it execute them, they would be slower for legitimate workloads because more time would be spent waiting for data from memory after the accesses were launched later. Is that a fix or a mitigation? If you think that i a mitigation, then no fix may be possible at all. If you think it's a fix then why is swapping in a OS kernel that avoid sthe problem equally well with the same performance cost not a fix?
No it isn't. One of the key features of the processor was memory access management. This is a HUGE bug.
It's a bit subtle. The architectural requirements of the memory access management are entirely being met. The values in every register (etc) are exactly what they should be according to the ISA, The bugs work by causing a change to what data is cached, based on memory values they cannot actually access and then measuring load times for the data to see what is in the cache. There is, I suspect nothing in the ISA specification which says that this shouldn't happen.
It also seems that there are genuine software solutions to most of this (modifiying the programs/OS holding the secure data so that it cannot be read this way), in which case they probably have an argument that software should have been written that way in the first place.
Are we sure he didn't say "Mars transfer orbit" and get edited down the reporters. He could aerobrake and enter an elliptical Mars orbit, but it would brake further on each orbit until it crashed.
Incidentally, their new engine is fueled by liquid methane. It should be possible to keep oxygen and methane liquid on a Mars transfer just by carefully managing where the sunlight falls on the ship and what colour you paint it.
Even project Orion doesn't get you to 10% of lightspeed -- there isn't that much energy in plutonium at any even slightly reasonable mass ration. You need an efficient fusion rocket or antimatter to get a rocket up to that speed, especially if you want to stop again. More plausible are beam-rider designs where the "engine" is left at home.
What I'd really like is a "high" power ion rocket that could use rocks for exhaust. This, though, is a big problem because rocks aren't a simple element, but a complex mix that varies. (By "high power" I'm thinking of about 30 pounds thrust, but that's probably dreaming.)
You're looking for a mass driver. You sit your rock on top of a superconducting magnet and accelerate in a large EM cannon. When it's nearly at the end you let the rock fly on and decelerate the magnet for reuse. Exhaust velocity is as high as your power supply and the length of your cannon allows. Thrust depends on rate of fire.
Each individual hyperloop pod has far fewer people in it than a train, so the most important consideration is to limit damage to one pod. I guess in an emergency you can brake a pod at maybe 2g which means you'd want them at least 2500m apart. Probably round up to 2 miles. You can get braking either from ablative skids on the bottom of the pod (I saw this design in an other context) or by releasing the vacuum and air-braking.
So compared to a train crash, there may be more damage, but to fewer people.
As to what happens to an individual pod, it depends what the problem is. Non-explosive loss of vacuum, or loss of magnetic levitation should result in the pod braking to a stop -- might be bumpy, but shouldn't be fatal. The pod coming apart in some way, or getting spun around so it "wedges" against the tube would be pretty ugly, as would any large foreign object managing to arrive in the tube or anything that knocks the tube seriously off straight just before a pod arrives, but these don't seem too likely.
A couple of obvious options -- use fibreglass instead of steel and build long slightly curving sections that can deal with expansion by just becoming slightly more, or less, curved. You're thinking too much like railway track.
We can already build and maintain long oil pipelines without having this problem, so how is hyperloop different?
But advancements in mechanical technology, combined with standards, enforcement, liability etc. have led to a massive reduction in the number of random dangerous mechanical failures in correctly maintained vehicles.
In most places the computers controlling the car have maps that already tell them the location of all the roadsigns to within a few millimeters. They'll stop because they haven't been told there isn't a stop sign there, and just use the cameras to look out for unexpected new stop signs for the few minutes from when they are put in until the map updates propagate.
If all cars are self-driving you don't need the signs at all, just a shared map annotated with traffic flow restrictions. If all cars are guaranteed to be continuously in communication you can go further and redesign the traffic flow across each junction continuously in real time.
Slashdot Mirror
Anyone else have a problem with studies that simply study the results of other studies? It would be like a news channel that only reports the news stories that other news channels provide, then based on analysis of trends seen offers an opinion on what is fact or fiction. While convenient, as one does not have to get out and do actual reporting, (or research in this case), I find it leaves much to be desired. It would be better in my opinion to research available data, summarize it, offer an opinion as to its cause, develop a methodology to test this opinion, and perform a carefully thought out study to test if this opinion is supported or refuted by this study.
Obviously you need primary data, but, if done carefully and well, this type of metastudy can be very powerful. Note that they are taking raw data from the previous studies, not just recording their conclusions. For one thing, any study in this area is at risk of statistical error, and the metastudy effectively has a much larger sample size, making this much less severe. Since the studies were performed by many different groups with different funding and at different times, they also reduce the risk of many kinds of systematic error.
They don't go on to suggest hypotheses and test them by new primary studies, because that's not what they are good at. They publish their data and statistical conclusions from the data so that others can do those things. Also, it took them six years to get this lot together, if they wait for follow-on studies before publishing it will take many more years.
In many areas of modern science, collection of data and investigation of hypotheses using the data are becoming separated and this is a good thing! Firstly, if your project is simply to collect a large high quality data set, you are much more motivated to worry about quality of the data, and you can't bias ti towards the conclusions you want, because you don't have any. Secondly the data is open, so others can repeat the later steps, even if regathering the data is too expensive or impossible (you can't go back and observe the same supernova twice).
- between chronic depression, caused by imbalances in neurotransmitter production or reuptake, and depression caused by living in depressing circumstances.
This distinction is not as clear-cut as you would like. Either can cause the other. Being persistently unhappy for valid reasons develops emotional habits which are encoded in neurotransmitter patterns, while neurological malfunctions can easily cause your life to fall apart. Many patients will need treatment for both.
There are a number of projects out there to develop specialised telescopes that will be able to take quite high resolution photos of unprecedentedly large areas of the sky at once, and big enough to gather enough light to show reasonably faint objects without needing too long an exposure. Look at the Large Synoptic Survey Telescope, for instance https://www.lsst.org/. This aims to photograph all of the "available" sky (it's in Chile, so it never sees the stars around the North celestial pole) every few nights for 10 years. There's lots of infomation on their site and in their papers, but a few numbers that jumped out at me: 8.4m primary mirror, 3.2 GPixel camera, 15 TB of data each night!
Even this would have to get moderately lucky to see a supernova as young as this one, which was captured in it's first minutes or hours. It would also, ideally, need to identify what it was seeing almost instantly, so that it (or another telescope) could start a follow-up within seconds or minutes.
Read the article, or even the linked Lancet article, which says:
Funding
National Institute for Health Research Oxford Health Biomedical Research Centre and the Japan Society for the Promotion of Science.
Which I think means 100% government funding. It also gives the exact methology -- a systematic search for, and integration of data from, all published studies that met stated criteria.
The most likely site for a Moon base is the South pole. There are crater rims there that are in permanent sunlight, which the crater bottoms are in permanent shadow (and appear to have significant quantities of moisture frozen out there.
All of which said, until you have in situ manufacturing, a compact fission reactor will give you a lot more power per ton hauled up from Earth.
I must admit I'd rather that the pizzas were delivered to someone a few blocks away, while I just watched. The collateral damage from actually getting pizza this way is a bit much
I would guess it'll go like chip making technology. Basically all the players own at least some patents that all the other players need (or at least really want) so they all license them to each other. If your R&D did really well, you own more (or better patents) you make some money off this process, if your R&D didn't luck out, it costs you some money. But it doesn't really stop anyone getting access to anuything
I;ve seen convincing arguments that airline pilots have overall negative safety value. More accidents are caused by pilots messing up or suiciding han are prevented by pilots being awesome.
So set up "Fred's privacy-enhanced vehicle service" and compete. It'll cost a little more (because that information is actually useful for something), but you can get the EFF or whoever to certify your privacy policy and see how much people actually care
This sort of taste is surprisingly volatile People used to like owning CDs and DVDs, but now we have spotify and netflix. My guess is you will have a period when people still own cars, but ALSO subscribe to transportation-as-a-service services (for the odd occasion when you need a bigger car, or two cars at once, or whatever) and then it will just be more convenient to use the service than try to remember where your car is and have it drive to where you need in time for whatever you need it for and pretty quickly ownership will fall away to become a niche thing, for a small number of enthusiasts or people who live in the middle of huge deserts.
Can't wait to see this in practice. Without manual controls, how do you adjust it's position in a parking space or garage? How would you go thru a typical drive up window? How would you maneuver around a barricade, accident or temporary police detour? How about going around and around a parking lot or underground garage?
At a guess, you select the appropriate view on the dasboard console and drag the rectangle that denotes the car to where you want it to go. A bit like the current systems that show you a rear view camera with the path of the car marked on it, but with the screen controlling the steering instead of vice versa. These cars will have radar or lidar and lots of cameras, as well as GPS.
I don't know the exact cost of this, but given it's an addon to an existing launch anyway, probably only tens of millions. Any serious program of manned exploration beyond Earth orbit is going to cost tens of billions at the very least, probably more.
I'm not sure what you regard as the difference between a fix and a mitigation. If intel changed the processor so that no memory accesses were launched until the CPU knew for sure if the memory management was going to permit it execute them, they would be slower for legitimate workloads because more time would be spent waiting for data from memory after the accesses were launched later. Is that a fix or a mitigation? If you think that i a mitigation, then no fix may be possible at all. If you think it's a fix then why is swapping in a OS kernel that avoid sthe problem equally well with the same performance cost not a fix?
I suspect HPC will take little or no performance hit. The hit comes on workloads that do LOTS of system calls. HPC does hardly any.
Citation please
No it isn't. One of the key features of the processor was memory access management. This is a HUGE bug.
It's a bit subtle. The architectural requirements of the memory access management are entirely being met. The values in every register (etc) are exactly what they should be according to the ISA, The bugs work by causing a change to what data is cached, based on memory values they cannot actually access and then measuring load times for the data to see what is in the cache. There is, I suspect nothing in the ISA specification which says that this shouldn't happen.
It also seems that there are genuine software solutions to most of this (modifiying the programs/OS holding the secure data so that it cannot be read this way), in which case they probably have an argument that software should have been written that way in the first place.
Are we sure he didn't say "Mars transfer orbit" and get edited down the reporters. He could aerobrake and enter an elliptical Mars orbit, but it would brake further on each orbit until it crashed.
Incidentally, their new engine is fueled by liquid methane. It should be possible to keep oxygen and methane liquid on a Mars transfer just by carefully managing where the sunlight falls on the ship and what colour you paint it.
Even project Orion doesn't get you to 10% of lightspeed -- there isn't that much energy in plutonium at any even slightly reasonable mass ration.
You need an efficient fusion rocket or antimatter to get a rocket up to that speed, especially if you want to stop again. More plausible are beam-rider designs where the "engine" is left at home.
What I'd really like is a "high" power ion rocket that could use rocks for exhaust. This, though, is a big problem because rocks aren't a simple element, but a complex mix that varies. (By "high power" I'm thinking of about 30 pounds thrust, but that's probably dreaming.)
You're looking for a mass driver. You sit your rock on top of a superconducting magnet and accelerate in a large EM cannon. When it's nearly at the end you let the rock fly on and decelerate the magnet for reuse. Exhaust velocity is as high as your power supply and the length of your cannon allows. Thrust depends on rate of fire.
Each individual hyperloop pod has far fewer people in it than a train, so the most important consideration is to limit damage to one pod. I guess in an emergency you can brake a pod at maybe 2g which means you'd want them at least 2500m apart. Probably round up to 2 miles. You can get braking either from ablative skids on the bottom of the pod (I saw this design in an other context) or by releasing the vacuum and air-braking.
So compared to a train crash, there may be more damage, but to fewer people.
As to what happens to an individual pod, it depends what the problem is. Non-explosive loss of vacuum, or loss of magnetic levitation should result in the pod braking to a stop -- might be bumpy, but shouldn't be fatal. The pod coming apart in some way, or getting spun around so it "wedges" against the tube would be pretty ugly, as would any large foreign object managing to arrive in the tube or anything that knocks the tube seriously off straight just before a pod arrives, but these don't seem too likely.
The difference is that a hyperloop pod is MUCH lighter than a train. The level of engineering required for the track is massively less.
A couple of obvious options -- use fibreglass instead of steel and build long slightly curving sections that can deal with expansion by just becoming slightly more, or less, curved. You're thinking too much like railway track.
We can already build and maintain long oil pipelines without having this problem, so how is hyperloop different?
But advancements in mechanical technology, combined with standards, enforcement, liability etc. have led to a massive reduction in the number of random dangerous mechanical failures in correctly maintained vehicles.
In most places the computers controlling the car have maps that already tell them the location of all the roadsigns to within a few millimeters. They'll stop because they haven't been told there isn't a stop sign there, and just use the cameras to look out for unexpected new stop signs for the few minutes from when they are put in until the map updates propagate.
If all cars are self-driving you don't need the signs at all, just a shared map annotated with traffic flow restrictions.
If all cars are guaranteed to be continuously in communication you can go further and redesign the traffic flow across each junction continuously in real time.
What are you guys driving?
It's 20000 miles between oil changes on my current car. And the car tells me when that's coming up!