See the previous article re. Prof. Mann for testimony from myself and other students who have worked with him.
As far as any of us have been able to tell, he has absolutely no medical requirement for any of his equipment, or any actual physical or psychological dependence on his equipment. He has been observed working fine without it on several occasions.
This is a publicity stunt, plain and simple. Prof. Mann has an agenda to push and is pushing it as hard as he can.
Would that be rather like neutron stars? My understanding is that current orthodox astrophysics models meutron stars as either a Bose-Einstein state, or as (in effect) a single, very big, neutron. (Or, er, are those the same things?) C'mon astrophysicists, enquiring idiots want to know!;)
Neutrons are fermions, as another poster pointed out, so they don't form a Bose-Einstein condensate.
Instead, they form a Fermi-degenerate system. No two neutrons can have the same quantum state, so they "stack up" from the lowest energy state on upwards (just as electrons "stack" to fill the different orbitals and shells in an atom).
A neutron star doesn't have many protons. Electrostatic repulsion between protons packed that densely would cause them to have a horrifically high potential energy, so instead they merge with electrons (at an energy cost) to become neutrons (or you could consider them to emit positrons; same net effect, different reaction path).
For the electrons bound to an atom, electrostatic forces are what cause the electrons to stay bound. For an atomic nucleus, the Strong force keeps the nucleons bound. For a neutron star, gravity keeps them bound. But you end up with the same kind of system in each case - particles that exclude each other filling up increasingly-energetic orbits because they're not allowed to have the same state.
A neutron star could be thought of as being similar to a giant _nucleus_, but it'll have many interesting features not found in an atomic nucleus (because it's big enough that the Strong force only affects parts that are really close to each other, and because you can get more ordinary material piling up on its surface, and because very high energy orbits many make other kinds of decay energetically favourable).
he PS3 is PowerPC-based, and won't have the wierd vector units of the PS2.
Umm, Altivec anybody?
Very different beasts.
Altivec is more or less just another functional unit tacked on to the G4 chip (like SSE or 3DNow on Intel/AMD chips). It's controlled directly by instructions in the main execution stream, like another add or divide unit would be.
The vector coprocessors on the PS2 were independent processor cores tied to the main core's co-processor bus (MIPS does funky things with coprocessors). They had their own program memory (if I recall correctly), which more or less made them big, bizzare CISC units with completely programmable microcode as far as the main processor was concerned. Program them, then send them data, trigger them, and watch them go as they run their embedded code.
For PC enthusiasts, this would be like breaking out a soldering iron and tacking a couple of completely independent DSP chips north of the northbridge in a PC. Powerful, but they'd be tricky as hell to use, because they'd have to be working on memory that the CPU wasn't modifying, and you'd have to set up the software all three chips are running so that all working data is processed and then passed on by each in turn without anyone stepping on anyone else's toes.
In theory, you can do this by pipelining your graphics data flow and having each module deal with a different part of it. This is how the PS2 was supposed to be used. In practice, this is a PITA to do, and Sony didn't pre-build developer libraries that did it (why is beyond me).
So, programming a G4 with AltiVec is much easier:).
For example, if a processor can delivery 10 IPC with a 200 MHz clock and another processor delivery 1 IPC with a 1 GHz, which one is the fastest?
If you can get an IPC average of 10, sell your idea to Intel before they send a group of thugs over to torture it out of you.
In the real world, the best sustained IPC you'll ever get is in the 2-2.5 range. The universe conspires against you - control hazards, cache and memory latencies, dependencies, misspeculation penalties, and all other brands of evil.
Trust me. I've been studying this for years, and will be making a career out of teaching it in a few more.
A 300 MHz modified MIPS core and a 200 MHz SH-4 core could go toe to toe with each other, but neither holds a candle to a current desktop.
All of the heavy lifting in *any* console produced in the past 5-10 years is done by the graphics chipset, and that gets stale pretty quickly too.
Imagine how many MIPS 4K cores you can fit in 300mm^2 in 4-5 years. That's a lot of power. Sure, they might only run at 1-2Ghz, but there will be 64 of them on a die. If you can harness that power, it might give your game developers much of that huge performance boost they want.
Two reasons why this doesn't work so well in practice:
Amdahl's Law.
Short version: 64 processors is never 64x faster than 1 processor. Things usually start to suck at 4 or 8.
Long version: As you try to parallelize to more processors, the time spent on serial (unparallelizable) parts of the task and on communication starts to limit your performance.
Communications bandwidth.
Think about graphics cards for a minute. Remember how much arse two SLI-rigged Voodoo 2 cards kicked back in days of yore? Ever wonder why 3dfx didn't just put 16 of them on a chip a few years down the road?
Evolution of graphics cards' feature sets is only part of the answer. The other reason is memory bandwidth.
Most graphics cards are memory-bound under common conditions. Clocking the GPU twice as fast, or having two GPUs, would accomplish little if you still have to wait for texel fetches from card memory. This is especially bad if you're trying to build a 64-core rendering engine. To send all of that texture data and all of that triangle data you're going to need not just completely insane bandwidth, but 64 or *more* (if multitexturing) _independent_ memory channels, each with a silly amount of bandwidth.
Texel caches on-die don't help you. They'll be too small to do any good. What you need is 1000-way interleaved DRAM on your console's board, and a bus running fast enough to radiate leakage in far-infrared to transfer all of that data.
In summary, while multiple cores are a good idea, and help up to around 4 or 8 cores, a massively parallel on-chip solution won't help you for game rendering, because the working set is too large to be cached per-core.
The only approach I've heard of that even *might* reduce this problem is the PixelFusion approach, and I strongly suspect that that smacks into memory bandwidth and Amdahl's Law problems too (it renders many pixels in a tile in parallel, but you still need texel information for all pixels for blending; z filtering doesn't save you if you still need 4 texels per pixel and don't have very bad overdraw).
So this stuff can conduct electricity, meaning it could generate a magnetic field, right? So you could theoretically generate a magnetic field to hold a concrete structure made from this stuff in the air. Does this mean that my goal of making a floating castle like all the bad guys in RPG video games have is finally within reach?
Certainly, until the rebels cut power to or carve up the plate on the ground you're pushing against, or until your castle melts to slag from resistive heating due to the vast currents required;).
Thought about the same thing a while back, and concluded that the Koopa Ship method is more practical;).
I have come to believe the following: No matter how technologically superior your product may be, if you compete directly with Microsoft you will lose (i.e. you will make less money, and have less market share). Why this is true for OS's and not for microprocessors I'm not sure.
It's because it's a lot easier to make a fully-compatible chip clone than it is to make a fully-compatible OS clone.
A chip's instruction set, bus interface, etc. are well-documented and relatively simple. An OS's API is far more complex, and can much more easily have a cloud of NDAs overshadow the dissemination of its documentation.
You can get around this with radio telescopes because you can sample and timestamp the received signals with timing resolution much finer than the period of the radio waves.
Measuring the wave collapses the wave function. Measuring then combining is not the same as combining THEN measuring. Those radio telescope arrays can't be called true interferometers.
Come again?
An interferometer is a classical device. I only have to worry about waveform collapses if I'm counting individual photons.
If you can make a 430-meter diameter telescope in this way, why can't a 10,000-meter diameter telescope be made by doing the same thing with hundreds or thousands of telescopes all across a hemisphere?
Because you need a direct optical link between the telescopes, and because you want the mirror setup you use for this to be as stable as possible (relative motion will change path lengths and muck up your image reconstruction).
You can get around this with radio telescopes because you can sample and timestamp the received signals with timing resolution much finer than the period of the radio waves. To do this with light, we'd need light sensors and electronics at least a million times faster than we have now, and atomic clocks based on mid-UV light instead of microwaves).
A giant interferometer could be built in space, of course.
This is exactly where I feel Linux should be used. The idea of dumb terminals and a central server has proven to be the most cost effective way for companies to implement computer technology. [...] In the future most companies will have dumb terminals and a server room with racks of cheap intel boxes. The OS on the server will be fault tolerant to the max, oh I lost a node ahh well only 255 left.
I'm trying to figure out what the benefit of this is. You'd have to maintain the user clients - which will still break down - and the server nodes on top of this.
You get fault tolerance - but user terminals don't need uptimes of years with transparent failover. You get centralized administration - but there are many ways of making this happen with user workstations too (witness the NT systems here that re-image their own drives every week).
Performance will always be worse with a centralized solution than with user workstations, because you have no local disk for fast scratch space (used by many applications in the environments I've worked in).
If computers cost $10k apiece, I can see cost being an issue, but if the cost of hardware and maintenance for a user's machine is much, much less than the cost of the user sitting at the machine, I don't see any justification on the basis of cost either.
How is this supposed to be a "most cost-effective" solution, again?
[Disclaimer: I think dumb terminal systems are nifty; I just don't think they're useful under most business conditions.]
Oh good, a Steve Mann expert. Could you clarify for us these last two tantalizing paragraphs of the report:
Since losing the use of his vision system and computer memory several weeks ago, he said, he cannot concentrate and is behaving differently. He is now undergoing tests to determine whether his brain has been affected by the sudden detachment from the technology.
My personal opinion?
This has "publicity stunt" written all over it. My impression (admittedly from relatively little experience) is that he has an agenda to push and actively enjoys being a sh*t disturber to do it, as that gets more coverage.
IMO, the most likely case is that he had a very stressful time before boarding the plane (understandably), didn't have the foresight to drink a lot of water on the plane trip, and ended up stressed and dehydrated at the airport on the way back. Maybe he fainted, or maybe he just felt a bit dizzy, but either way playing it up when reporting it was a golden media opportunity.
At the time, I believe he was trying to recruit people to do heavy-duty graphics work (i.e. when he moves his head side to side, his camera is taking discrete pictures of a room/building/whatever at different angles. He was working on algorithms to put them all together and make them coherent).
He's still working on that. He has pretty panoramic pictures printed up and taped to cabinets around the lab. It takes a cluster of computers to do the image calculations (yes, a beowolf - Prof. Mann would love it here if he isn't slashdotting already).
I am deeply skeptical that any real-time reality modification is going on with his glasses. The most he's ever done here is show pictures of a text overlay that tracks targets, and write a laser tag variant that's supposed to paint surfaces colours when you "shoot" them. I have no idea if the latter works in real-time, as I haven't used one of his rigs for that (I strongly suspect it doesn't).
Anyhow, the point is, I distinctly remember him saying that he got nauseous when he removed his visor. The reason was very simple. He spent all of his waking life (outside of the shower) in a 2D world. His body was so used to it, that living in 3D took some serious getting used to, and he would feel sick.
That sounds like something Prof. Mann would say.
It also contradicts everything I've seen him do. Like I said, he can perceive the world just fine without an overlay on his vision.
In one fell swoop they cut him off from his augmented memory and processing, and then threw his visual system for a loop, hence the need for a wheelchair.
I'm a graduate student at the University of Toronto, and interact with Prof. Mann on an intermittent basis (did a project under him a few years back, meet him in the lab whenever I'm borrowing his soldering equipment).
He can see fine without his HUD. It's not a complete visual transformation overlay - it's a wearable computer display, functionally equivalent to most of the other wearable displays you can buy. He's been working on information-overlay projects for years, many of them successful, but to say that he has "vital" vision-enhancement programs running at all times is a drastic overstatement.
Likewise, "augmented memory" consists of him either teleconferencing with someone or doing a Google lookup. He's perfectly capable of finding his way through this university, or an airport, without augmentation.
Use common sense, people. If he was disoriented, I'd suspect it to be the result of a many-hour delay with inadequate food/water or of an overly-zealous search as opposed to loss of any electronics.
I believe there is not only the problem of how much energy is needed to attain altitude, but the exit velocity is a factor. To leave the atmosphere, a certain velocity is needed to prevent either:
a) bouncing off the edge of the atmosphere or b) burning up during exit
You actually want to be moving as slowly as possible while you're in dense atmosphere, because you shed energy as the cube of your airspeed.
Burning up and bouncing are mainly concerns on re-entry, because you start off going at a far greater speed than would be practial to reach at low altitudes. While you're climbing to orbit, wasting your fuel by pushing against air is the main concern (well, that and pressure changes mucking with your rocket nozzle efficiency).
For perfect efficiency, build a magnetic launcher on a 50-mile-high track;). I'm not claiming it's easy to reach perfect efficiency - just that very low costs to orbit are theoretically possible.
It would seem that your source is in error about a few of these points:
Sure, there's a big cost--that can be brought down (though not free, it takes a minimum of 400 gigajoules to lift 100 kg to 100 km above Earth).
Energy to lift something to a given altitude (not orbit) is force * distance. 100 kg feels 1000 N from the Earth's pull. 100 km is 1.0e5 metres. Energy required to lift 100 kg to 100km (and stationary above Earth's surface) is 100 megajoules - or what you'd get from about $5 US worth of gasoline (or less) at perfect efficiency.
Energy to put something into orbit fairly close to Earth's surface (LEO) is the binding energy (half the gravitational potential energy of an object on the Earth's surface). GPE is -m1m2G/r, or 6e24 * 6.7e-11 / 6.5e6 = 62 MJ/kg for an object sitting on the Earth's surface. This gives a theoretical minimum of 31 MJ/kg to put something in low earth orbit, or 3.1 gigajoules for a 100kg object.
You'd get this by burning around $150 US worth of gasoline at perfect efficiency and magically imparting all of the resulting energy to the cargo.
Space travel is expensive because our rockets a) lift their fuel with themselves and b) impart a lot of energy to the outgoing exhaust instead of to the craft itself. At perfect efficiency, getting into space would be quite cheap.
The telephone companies use tower to tower microwave communication all the time. They seem to prefer these little high gain directionall dishes over laying wires. Looking at the increasing simplicity of their towers over the years gives me hope. Gone are the massive magnetic feedhorns and bulky waveguides. They are now lightweight wire mesh parabolic dishes fed by coaxial cable.
Telephone companies are routing a lot less data:).
I agree that this could be a good city-level solution, though.
just because todays wireless technology only supports a handful of users at a time (say 20 to 30 comfortably) it doesn't mean that tomorrow, we won't have standards that can route wirelessly and support gigabits of traffic at a time, enough for global internet wirelessly.
I'm afraid there are hard limits on how much you'll ever be able to route with a (broadcast) wireless scheme.
The window of frequencies you can use is limited. Above a certain frequency range, your signal will be blocked by things like rain or fog (and of course, walls and windows). This limit is probably in the 10-20 GHz range. No amount of technological development will change this - it's a physical limit.
This places an upper limit on the bandwidth that any given "cell" (broadcast region) can support (no matter how many base stations you put in that cell).
Divide the bandwidth available per cell by the bandwidth a user wants, and you have the maximum number of users per cell. This means your cell must be small enough to have *only* that number of users in it.
This ends up being about 100 people/cell, if they each want 100 kbytes/second access and your broadcasting can handle 100 Gbit (20 GHz of spectrum at 10 bits/Hz with half the bandwidth upstream and half the bandwidth downstream).
This gives a maximum cell size of maybe a hundred metres or so.
When you start to route traffic, things get _much_ worse. If the area you're trying to cover is only a kilometre wide (part of a city's core), the average path length will be on the order of 5-8 hops. If you're distributing *only* through broadcast wireless, your bandwidth use goes up by a factor of 5-8 on average, because you have that many more repeated messages flying through the mesh.
So either everyone gets 10 kbytes/sec, or you shrink your cells, which makes the number of hops needed larger, which means that even *more* of the traffic you're routing is other peoples' messages in flight...
Summary: You need high-capacity point-to-point links. You're not getting around this.
And fiber's a whole lot more reliable than microwave for this (no rain), and can carry a whole lot more (gain-bandwidth product for erbium-doped fiber is in the 100-gigahertz range if I remember correctly, and maximum theoretical limit for optical communication is around 1 petahertz).
Microwave is especially bad for between-city communication, as your range is limited by atmospheric quality and curvature of the earth (and you still need a big expensive tower, which means you still have the ISP problem).
Not really a big deal when you consider that there is *no* way for them to know this is occuring... [...] None. Zip. Zilch. ZeRo.
Unless they just sniff packet headers and notice that you're web surfing while you're playing Quake. Kind of difficult to do that with only two hands and one pair of eyes.
Or unless they notice that you're viewing dozens of web pages per second.
Either way, they'd have a hard time *proving* you're up to something, but they can jerk your connection around under any number of pretenses on their end. If this becomes a big problem, believe me, they'll start squashing people who try this.
Wake up dude. Your missing the boat. We don't need ISPs anymore.
That's true if your traffic is local to your neighbourhood.
If you want to route traffic through more than your neighbourhood, though, you're going to run into problems. If the area you're routing traffic in is more than a few hops wide, you'll either be spending most of your bandwidth routing other peoples' messages, or you'll have to set up dedicated high-bandwidth links to let long routes bypass most users' nodes. Now if you have a network of these links... you have something that looks a lot like the existing backbone.
If you have a backbone to maintain, you have to charge for use of the backbone to amortize building and maintenance costs. This gives you a multi-level system where the people running the backbone sell bandwidth to people who locally redistribute the bandwidth.
Which looks a lot like the current system of multiple levels of ISPs.
ISPs exist for a reason. If you try to do away with them, you'll just end up having to reinvent them.
Re:How did they come by that estimate?
on
Every Species on Earth
·
· Score: 3, Insightful
I read the article and it doesn't seem to offer any evidence other than speculation as to where this number comes from. It seems kinda large to me. I know humans don't occupy *every* place on the planet, but there are very few areas within the top 10,000 feet of the Earth's crust that aren't accessible to humans already. Are they suggesting that life is blossoming in the mantle?
If I understand correctly, most of the species are of things we see every day but don't bother classifying - insects, fungi, bacteria, and so forth. There's a lot of space down at the bottom of the pyramid, and with short generations and (for bacteria, at least) a high mutation rate, species differentiate a lot faster than at the top.
If every given hundred-kilometre-radius area has a hundred local subspecies of bug or bacterium, it'll take quite a while and a lot of manpower to catalogue them all.
How exactly did scientists come upon this number?
My guess: By taking a really thorough survey of *all* life within test areas in various countries, and checking to see what fraction of the distinct species found were ones we knew about.
There's quite a bit of uncertainty in the figures you'll arrive at from this, but you can certainly get a ballpark estimate. So far, the estimate is that we don't know about most species.
It will be interesting to see what this finds in the way of brown dwarfs (failed stars/really big planets).
Some models of dark matter say that much of it is in the form of "MAssive Compact Halo Objects", of which brown dwarfs would be one type. While a brown dwarf is the next best thing to undetectable at visible wavelengths, a brown dwarf might have a detectable heat signature from warming processes similar to those at work within Jupiter. A big IR telescope could help determine if there are enough brown dwarfs to account for some of the missing mass.
An IR telescope would also be good at finding stars that are masked by dense dust clouds. Starlight would heat up the clouds near the star.
And if anyone's built a Dyson sphere, that will have a pretty obvious heat signature too [unlikely, but we'd certainly see it].
Slashdot Mirror
See the previous article re. Prof. Mann for testimony from myself and other students who have worked with him.
As far as any of us have been able to tell, he has absolutely no medical requirement for any of his equipment, or any actual physical or psychological dependence on his equipment. He has been observed working fine without it on several occasions.
This is a publicity stunt, plain and simple. Prof. Mann has an agenda to push and is pushing it as hard as he can.
My god, there is so much less crap in the threads, even in spite of all the whinging-on about the so-called jokes.
I *like* Slashdot non-anon.
Unfortunately, I give it about 10 hours before about a hundred thousand "amusingly vulgar" random nicks fill the gap.
Score filtering works adequately for me.
Would that be rather like neutron stars? My understanding is that current orthodox astrophysics models meutron stars as either a Bose-Einstein state, or as (in effect) a single, very big, neutron. (Or, er, are those the same things?) C'mon astrophysicists, enquiring idiots want to know! ;)
:).
Neutrons are fermions, as another poster pointed out, so they don't form a Bose-Einstein condensate.
Instead, they form a Fermi-degenerate system. No two neutrons can have the same quantum state, so they "stack up" from the lowest energy state on upwards (just as electrons "stack" to fill the different orbitals and shells in an atom).
A neutron star doesn't have many protons. Electrostatic repulsion between protons packed that densely would cause them to have a horrifically high potential energy, so instead they merge with electrons (at an energy cost) to become neutrons (or you could consider them to emit positrons; same net effect, different reaction path).
For the electrons bound to an atom, electrostatic forces are what cause the electrons to stay bound. For an atomic nucleus, the Strong force keeps the nucleons bound. For a neutron star, gravity keeps them bound. But you end up with the same kind of system in each case - particles that exclude each other filling up increasingly-energetic orbits because they're not allowed to have the same state.
A neutron star could be thought of as being similar to a giant _nucleus_, but it'll have many interesting features not found in an atomic nucleus (because it's big enough that the Strong force only affects parts that are really close to each other, and because you can get more ordinary material piling up on its surface, and because very high energy orbits many make other kinds of decay energetically favourable).
I hope this helps
he PS3 is PowerPC-based, and won't have the wierd vector units of the PS2.
:).
Umm, Altivec anybody?
Very different beasts.
Altivec is more or less just another functional unit tacked on to the G4 chip (like SSE or 3DNow on Intel/AMD chips). It's controlled directly by instructions in the main execution stream, like another add or divide unit would be.
The vector coprocessors on the PS2 were independent processor cores tied to the main core's co-processor bus (MIPS does funky things with coprocessors). They had their own program memory (if I recall correctly), which more or less made them big, bizzare CISC units with completely programmable microcode as far as the main processor was concerned. Program them, then send them data, trigger them, and watch them go as they run their embedded code.
For PC enthusiasts, this would be like breaking out a soldering iron and tacking a couple of completely independent DSP chips north of the northbridge in a PC. Powerful, but they'd be tricky as hell to use, because they'd have to be working on memory that the CPU wasn't modifying, and you'd have to set up the software all three chips are running so that all working data is processed and then passed on by each in turn without anyone stepping on anyone else's toes.
In theory, you can do this by pipelining your graphics data flow and having each module deal with a different part of it. This is how the PS2 was supposed to be used. In practice, this is a PITA to do, and Sony didn't pre-build developer libraries that did it (why is beyond me).
So, programming a G4 with AltiVec is much easier
For example, if a processor can delivery 10 IPC with a 200 MHz clock and another processor delivery 1 IPC with a 1 GHz, which one is the fastest?
If you can get an IPC average of 10, sell your idea to Intel before they send a group of thugs over to torture it out of you.
In the real world, the best sustained IPC you'll ever get is in the 2-2.5 range. The universe conspires against you - control hazards, cache and memory latencies, dependencies, misspeculation penalties, and all other brands of evil.
Trust me. I've been studying this for years, and will be making a career out of teaching it in a few more.
A 300 MHz modified MIPS core and a 200 MHz SH-4 core could go toe to toe with each other, but neither holds a candle to a current desktop.
All of the heavy lifting in *any* console produced in the past 5-10 years is done by the graphics chipset, and that gets stale pretty quickly too.
Two reasons why this doesn't work so well in practice:
Short version: 64 processors is never 64x faster than 1 processor. Things usually start to suck at 4 or 8.
Long version: As you try to parallelize to more processors, the time spent on serial (unparallelizable) parts of the task and on communication starts to limit your performance.
Think about graphics cards for a minute. Remember how much arse two SLI-rigged Voodoo 2 cards kicked back in days of yore? Ever wonder why 3dfx didn't just put 16 of them on a chip a few years down the road?
Evolution of graphics cards' feature sets is only part of the answer. The other reason is memory bandwidth.
Most graphics cards are memory-bound under common conditions. Clocking the GPU twice as fast, or having two GPUs, would accomplish little if you still have to wait for texel fetches from card memory. This is especially bad if you're trying to build a 64-core rendering engine. To send all of that texture data and all of that triangle data you're going to need not just completely insane bandwidth, but 64 or *more* (if multitexturing) _independent_ memory channels, each with a silly amount of bandwidth.
Texel caches on-die don't help you. They'll be too small to do any good. What you need is 1000-way interleaved DRAM on your console's board, and a bus running fast enough to radiate leakage in far-infrared to transfer all of that data.
In summary, while multiple cores are a good idea, and help up to around 4 or 8 cores, a massively parallel on-chip solution won't help you for game rendering, because the working set is too large to be cached per-core.
The only approach I've heard of that even *might* reduce this problem is the PixelFusion approach, and I strongly suspect that that smacks into memory bandwidth and Amdahl's Law problems too (it renders many pixels in a tile in parallel, but you still need texel information for all pixels for blending; z filtering doesn't save you if you still need 4 texels per pixel and don't have very bad overdraw).
Would having conducting concrete make it easier or harder to prevent electrochemical corrosion in reinforced concrete?
Intuition is telling me "yes to both", but I'm not a Civ...
So this stuff can conduct electricity, meaning it could generate a magnetic field, right? So you could theoretically generate a magnetic field to hold a concrete structure made from this stuff in the air. Does this mean that my goal of making a floating castle like all the bad guys in RPG video games have is finally within reach?
;).
;).
Certainly, until the rebels cut power to or carve up the plate on the ground you're pushing against, or until your castle melts to slag from resistive heating due to the vast currents required
Thought about the same thing a while back, and concluded that the Koopa Ship method is more practical
I have come to believe the following: No matter how technologically superior your product may be, if you compete directly with Microsoft you will lose (i.e. you will make less money, and have less market share). Why this is true for OS's and not for microprocessors I'm not sure.
It's because it's a lot easier to make a fully-compatible chip clone than it is to make a fully-compatible OS clone.
A chip's instruction set, bus interface, etc. are well-documented and relatively simple. An OS's API is far more complex, and can much more easily have a cloud of NDAs overshadow the dissemination of its documentation.
I know which I'd try to clone.
You can get around this with radio telescopes because you can sample and timestamp the received signals with timing resolution much finer than the period of the radio waves.
Measuring the wave collapses the wave function. Measuring then combining is not the same as combining THEN measuring. Those radio telescope arrays can't be called true interferometers.
Come again?
An interferometer is a classical device. I only have to worry about waveform collapses if I'm counting individual photons.
If you can make a 430-meter diameter telescope in this way, why can't a 10,000-meter diameter telescope be made by doing the same thing with hundreds or thousands of telescopes all across a hemisphere?
Because you need a direct optical link between the telescopes, and because you want the mirror setup you use for this to be as stable as possible (relative motion will change path lengths and muck up your image reconstruction).
You can get around this with radio telescopes because you can sample and timestamp the received signals with timing resolution much finer than the period of the radio waves. To do this with light, we'd need light sensors and electronics at least a million times faster than we have now, and atomic clocks based on mid-UV light instead of microwaves).
A giant interferometer could be built in space, of course.
In other words, this technology is faster than anything that can be or is likely to bedone on silicon.
;).
...For problems that are easily parallelized to order 1e24 nodes
On the other hand, rendering Quake is massively parallelizable...
This is exactly where I feel Linux should be used. The idea of dumb terminals and a central server has proven to be the most cost effective way for companies to implement computer technology.
[...]
In the future most companies will have dumb terminals and a server room with racks of cheap intel boxes. The OS on the server will be fault tolerant to the max, oh I lost a node ahh well only 255 left.
I'm trying to figure out what the benefit of this is. You'd have to maintain the user clients - which will still break down - and the server nodes on top of this.
You get fault tolerance - but user terminals don't need uptimes of years with transparent failover. You get centralized administration - but there are many ways of making this happen with user workstations too (witness the NT systems here that re-image their own drives every week).
Performance will always be worse with a centralized solution than with user workstations, because you have no local disk for fast scratch space (used by many applications in the environments I've worked in).
If computers cost $10k apiece, I can see cost being an issue, but if the cost of hardware and maintenance for a user's machine is much, much less than the cost of the user sitting at the machine, I don't see any justification on the basis of cost either.
How is this supposed to be a "most cost-effective" solution, again?
[Disclaimer: I think dumb terminal systems are nifty; I just don't think they're useful under most business conditions.]
Oh good, a Steve Mann expert. Could you clarify for us these last two tantalizing paragraphs of the report:
Since losing the use of his vision system and computer memory several weeks ago, he said, he cannot concentrate and is behaving differently. He is now undergoing tests to determine whether his brain has been affected by the sudden detachment from the technology.
My personal opinion?
This has "publicity stunt" written all over it. My impression (admittedly from relatively little experience) is that he has an agenda to push and actively enjoys being a sh*t disturber to do it, as that gets more coverage.
IMO, the most likely case is that he had a very stressful time before boarding the plane (understandably), didn't have the foresight to drink a lot of water on the plane trip, and ended up stressed and dehydrated at the airport on the way back. Maybe he fainted, or maybe he just felt a bit dizzy, but either way playing it up when reporting it was a golden media opportunity.
At the time, I believe he was trying to recruit people to do heavy-duty graphics work (i.e. when he moves his head side to side, his camera is taking discrete pictures of a room/building/whatever at different angles. He was working on algorithms to put them all together and make them coherent).
He's still working on that. He has pretty panoramic pictures printed up and taped to cabinets around the lab. It takes a cluster of computers to do the image calculations (yes, a beowolf - Prof. Mann would love it here if he isn't slashdotting already).
I am deeply skeptical that any real-time reality modification is going on with his glasses. The most he's ever done here is show pictures of a text overlay that tracks targets, and write a laser tag variant that's supposed to paint surfaces colours when you "shoot" them. I have no idea if the latter works in real-time, as I haven't used one of his rigs for that (I strongly suspect it doesn't).
Anyhow, the point is, I distinctly remember him saying that he got nauseous when he removed his visor. The reason was very simple. He spent all of his waking life (outside of the shower) in a 2D world. His body was so used to it, that living in 3D took some serious getting used to, and he would feel sick.
That sounds like something Prof. Mann would say.
It also contradicts everything I've seen him do. Like I said, he can perceive the world just fine without an overlay on his vision.
In one fell swoop they cut him off from his augmented memory and processing, and then threw his visual system for a loop, hence the need for a wheelchair.
I'm a graduate student at the University of Toronto, and interact with Prof. Mann on an intermittent basis (did a project under him a few years back, meet him in the lab whenever I'm borrowing his soldering equipment).
He can see fine without his HUD. It's not a complete visual transformation overlay - it's a wearable computer display, functionally equivalent to most of the other wearable displays you can buy. He's been working on information-overlay projects for years, many of them successful, but to say that he has "vital" vision-enhancement programs running at all times is a drastic overstatement.
Likewise, "augmented memory" consists of him either teleconferencing with someone or doing a Google lookup. He's perfectly capable of finding his way through this university, or an airport, without augmentation.
Use common sense, people. If he was disoriented, I'd suspect it to be the result of a many-hour delay with inadequate food/water or of an overly-zealous search as opposed to loss of any electronics.
I believe there is not only the problem of how much energy is needed to attain altitude, but the exit velocity is a factor. To leave the atmosphere, a certain velocity is needed to prevent either:
;). I'm not claiming it's easy to reach perfect efficiency - just that very low costs to orbit are theoretically possible.
a) bouncing off the edge of the atmosphere
or
b) burning up during exit
You actually want to be moving as slowly as possible while you're in dense atmosphere, because you shed energy as the cube of your airspeed.
Burning up and bouncing are mainly concerns on re-entry, because you start off going at a far greater speed than would be practial to reach at low altitudes. While you're climbing to orbit, wasting your fuel by pushing against air is the main concern (well, that and pressure changes mucking with your rocket nozzle efficiency).
For perfect efficiency, build a magnetic launcher on a 50-mile-high track
It would seem that your source is in error about a few of these points:
Sure, there's a big cost--that can be brought down (though not free, it takes a minimum of 400 gigajoules to lift 100 kg to 100 km above Earth).
Energy to lift something to a given altitude (not orbit) is force * distance. 100 kg feels 1000 N from the Earth's pull. 100 km is 1.0e5 metres. Energy required to lift 100 kg to 100km (and stationary above Earth's surface) is 100 megajoules - or what you'd get from about $5 US worth of gasoline (or less) at perfect efficiency.
Energy to put something into orbit fairly close to Earth's surface (LEO) is the binding energy (half the gravitational potential energy of an object on the Earth's surface). GPE is -m1m2G/r, or 6e24 * 6.7e-11 / 6.5e6 = 62 MJ/kg for an object sitting on the Earth's surface. This gives a theoretical minimum of 31 MJ/kg to put something in low earth orbit, or 3.1 gigajoules for a 100kg object.
You'd get this by burning around $150 US worth of gasoline at perfect efficiency and magically imparting all of the resulting energy to the cargo.
Space travel is expensive because our rockets a) lift their fuel with themselves and b) impart a lot of energy to the outgoing exhaust instead of to the craft itself. At perfect efficiency, getting into space would be quite cheap.
But really, is this news? Maybe you think I'm a troll, but we can't get better stuff to discuss than this?
It's news to anyone who's interested in pulsars. There seem to be enough armchar (and non-armchair) astronomers here to justify it.
The only flaw that I see here is that this is filed under the "news" category instead of "space".
The telephone companies use tower to tower microwave communication all the time. They seem to prefer these little high gain directionall dishes over laying wires. Looking at the increasing simplicity of their towers over the years gives me hope. Gone are the massive magnetic feedhorns and bulky waveguides. They are now lightweight wire mesh parabolic dishes fed by coaxial cable.
:).
Telephone companies are routing a lot less data
I agree that this could be a good city-level solution, though.
just because todays wireless technology only supports a handful of users at a time (say 20 to 30 comfortably) it doesn't mean that tomorrow, we won't have standards that can route wirelessly and support gigabits of traffic at a time, enough for global internet wirelessly.
I'm afraid there are hard limits on how much you'll ever be able to route with a (broadcast) wireless scheme.
The window of frequencies you can use is limited. Above a certain frequency range, your signal will be blocked by things like rain or fog (and of course, walls and windows). This limit is probably in the 10-20 GHz range. No amount of technological development will change this - it's a physical limit.
This places an upper limit on the bandwidth that any given "cell" (broadcast region) can support (no matter how many base stations you put in that cell).
Divide the bandwidth available per cell by the bandwidth a user wants, and you have the maximum number of users per cell. This means your cell must be small enough to have *only* that number of users in it.
This ends up being about 100 people/cell, if they each want 100 kbytes/second access and your broadcasting can handle 100 Gbit (20 GHz of spectrum at 10 bits/Hz with half the bandwidth upstream and half the bandwidth downstream).
This gives a maximum cell size of maybe a hundred metres or so.
When you start to route traffic, things get _much_ worse. If the area you're trying to cover is only a kilometre wide (part of a city's core), the average path length will be on the order of 5-8 hops. If you're distributing *only* through broadcast wireless, your bandwidth use goes up by a factor of 5-8 on average, because you have that many more repeated messages flying through the mesh.
So either everyone gets 10 kbytes/sec, or you shrink your cells, which makes the number of hops needed larger, which means that even *more* of the traffic you're routing is other peoples' messages in flight...
Summary: You need high-capacity point-to-point links. You're not getting around this.
And fiber's a whole lot more reliable than microwave for this (no rain), and can carry a whole lot more (gain-bandwidth product for erbium-doped fiber is in the 100-gigahertz range if I remember correctly, and maximum theoretical limit for optical communication is around 1 petahertz).
Microwave is especially bad for between-city communication, as your range is limited by atmospheric quality and curvature of the earth (and you still need a big expensive tower, which means you still have the ISP problem).
Not really a big deal when you consider that there is *no* way for them to know this is occuring... [...] None. Zip. Zilch. ZeRo.
Unless they just sniff packet headers and notice that you're web surfing while you're playing Quake. Kind of difficult to do that with only two hands and one pair of eyes.
Or unless they notice that you're viewing dozens of web pages per second.
Either way, they'd have a hard time *proving* you're up to something, but they can jerk your connection around under any number of pretenses on their end. If this becomes a big problem, believe me, they'll start squashing people who try this.
Wake up dude. Your missing the boat. We don't need ISPs anymore.
That's true if your traffic is local to your neighbourhood.
If you want to route traffic through more than your neighbourhood, though, you're going to run into problems. If the area you're routing traffic in is more than a few hops wide, you'll either be spending most of your bandwidth routing other peoples' messages, or you'll have to set up dedicated high-bandwidth links to let long routes bypass most users' nodes. Now if you have a network of these links... you have something that looks a lot like the existing backbone.
If you have a backbone to maintain, you have to charge for use of the backbone to amortize building and maintenance costs. This gives you a multi-level system where the people running the backbone sell bandwidth to people who locally redistribute the bandwidth.
Which looks a lot like the current system of multiple levels of ISPs.
ISPs exist for a reason. If you try to do away with them, you'll just end up having to reinvent them.
I read the article and it doesn't seem to offer any evidence other than speculation as to where this number comes from. It seems kinda large to me. I know humans don't occupy *every* place on the planet, but there are very few areas within the top 10,000 feet of the Earth's crust that aren't accessible to humans already. Are they suggesting that life is blossoming in the mantle?
If I understand correctly, most of the species are of things we see every day but don't bother classifying - insects, fungi, bacteria, and so forth. There's a lot of space down at the bottom of the pyramid, and with short generations and (for bacteria, at least) a high mutation rate, species differentiate a lot faster than at the top.
If every given hundred-kilometre-radius area has a hundred local subspecies of bug or bacterium, it'll take quite a while and a lot of manpower to catalogue them all.
How exactly did scientists come upon this number?
My guess: By taking a really thorough survey of *all* life within test areas in various countries, and checking to see what fraction of the distinct species found were ones we knew about.
There's quite a bit of uncertainty in the figures you'll arrive at from this, but you can certainly get a ballpark estimate. So far, the estimate is that we don't know about most species.
It will be interesting to see what this finds in the way of brown dwarfs (failed stars/really big planets).
Some models of dark matter say that much of it is in the form of "MAssive Compact Halo Objects", of which brown dwarfs would be one type. While a brown dwarf is the next best thing to undetectable at visible wavelengths, a brown dwarf might have a detectable heat signature from warming processes similar to those at work within Jupiter. A big IR telescope could help determine if there are enough brown dwarfs to account for some of the missing mass.
An IR telescope would also be good at finding stars that are masked by dense dust clouds. Starlight would heat up the clouds near the star.
And if anyone's built a Dyson sphere, that will have a pretty obvious heat signature too [unlikely, but we'd certainly see it].