But it's clear that Wikipedia isn't dead, not in the way Dvorak said. So why mention it, unless it's interesting that Dvorak is wrong again, fundamentally?
FWIW, the way that some authorities blacklist Wikipedia because it's "too interactive", preferring "reliable" sources like Britannica, shows how those authorities are, in some senses, dead. They'd show signs of life if instead of purely taking info from these references for their "reports", they included the reporters and validators (teachers) in the loop, and forced corrections/additions back into Wikipedia after grading them. Learning how to test references for accuracy is an essential skill in the modern infosphere. And learning how to revise references is an essential skill even more important to educational institutions.
I've seen studies showing Wikipedia more accurate overall than is Britannica. And Wikipedia will improve, while Britannica will stay the same - and never lose some of its institutional inaccuracies. That's "dead" in intellectual terms.
Besides, anyone who quotes or cites Wikipedia without checking the sources it lists, or the history of the content they're referencing, is setting themselves up for a fall. Wikipedia doesn't rely in human interactivity just to accumulate content. It requires the consumer to interact, too, to keep their confidence high and correct mistakes they can tell. It's a medium for people to interact, not just a message from some people to a consumer.
Dvorak has never understood interactivity. He doesn't seem to understand people, either. He's a good compilation of crappy wisdom that's conveniently collected in one place to easily ignore.
Threads and processes are one way to do it. But they map relatively closely to only some HW configurations. And they have overhead, in each thread/process and in managing them. While their relatively monolithic concurrency is a model that only a subset of all parallel problems fall into. Thread synchronization and IPC latency are additional problems that consume additional resources.
As mentioned elsewhere, consider processing two parallel loops like c = a + b ; e = c + d . That's hard to do with threads or processes, but is parallelizable in HW, like FPGA or ASIC. The languages that do that, like VHDL and Verilog, are very different from the procedures that are encapsulated in threads and processes.
Now we're discussing an Intel chip with very parallel HW. I'm programming its competitor, the Cell uP, which has a fast Power RISC on a bus with 6-8 very fast DSPs. The point of parallelizing the HW is large compute capacity. DSP programming strategies are defined by maximizing throughput. There's no room on the DSPs for thread or process overhead. And the Power does run Linux, which could have concurrent processing on the DSPs while the program you're programming needs them. So the programming is complex, to account for the extra complexity of the HW. Partly because the OS hasn't caught up with the use cases of the HW. Both because at the early stage of the tech, so history hasn't shown some case classes to be negligible, and because these are general purpose chips that will serve many case classes, many of them new because the HW capability is new.
Those threads and processes were new once, too. They relied on context switching features of chips to evolve. Their techniques gradually became stable through long use, and migrated into the OS and compilers. Now HW has taken another jump. We' must repeat the innovation cycle again. That's what I'm calling for, and looking to the parallel HW (FPGA and ASIC) community for ideas, now that the commodity uPs are including more of their tech. Until people build new languages and compilers, it'll be challenging and frustrating to do it right, making the new HW architectures worthwhile.
They might not have legal grounds to reopen the case. But how many of these random people, many teenagers or their parents, decided against their lawyers' advice to drop it rather than pursue the counterclaim? Very few, if any. Therefore what has changed is that their lawyers have gotten smarter. That such change isn't legal grounds for getting justice is injustice.
It's probably safe to presume that these new cases finding counterclaims allowed are of the same merit (or lack) as the ones that just totally dropped, and even found the defendants liable.
What has changed? The laws are the same. The actions are the same.
Maybe the judges are a little smarter now. Maybe the lawyers are a little smarter. But if I didn't get the same results as a defendant in the same circumstances a year or more ago, I'd want a new trial. It's not supposed to be my problem if the administrators of justice are too stupid to leave me alone with laws they don't understand.
That's the new development I want to see: a retrial on the basis that maybe the courts aren't as stupid as they were when they decided against me.
Maybe I'm not getting the joke, but I don't see how the apparently totally legit mentions of Williams in Dvorak's entry and Ballmer in Williams' bio invalidate Wikipedia.
In fact, that statement by Dvorak makes him look like a troll of all trades, not just an Apple basher, considering that the wiki lives, despite years of targeting for organized vandalism.
Thanks - it sounds like Zune beats PSP on number of songs, but PSP wins on everything else (especially games:). But PSP can take a $40 4GB stick, plus a WiFi stream, so it can hold quite a lot (or unlimited amounts) of music.
I wonder why I don't hear about more PSP vs Zune competition. Maybe if Sony would release the 1.8GB UMD blanks and recorders, there would be little competition. Maybe we need a "How Sony Can Make Zune a Success" story.
How does the Zune compare to the Sony PSP, against just the PSP portable media player features?
And overall, which one is the better buy? What if you own an XBox, or if you own a PS3? How about cross-brand, is either portable anything but useless with the cross-brand console?
Who cares what Dvorak says? What reason is there to listen to him? He isn't even the guy after whom the interesting keyboard, the obscure cryptography based on it, or anything else.
He's been totally wrong about Apple all the time, including such fundamentals of whether people would use a "mouse".
Enough of this clown. Steve Jobs should just take him behind the woodshed and spank him down once and for all.
It's not like concurrent programming these days is efficient, in either processor or programmer time.
We need better software and tools for making it that are more parallel. Software doesn't adapt, people adapt software. And I'm talking about how slowly we're doing it, while suggesting new ways to get more.
What are you talking about? Don't you know anything about how challenging and frustrating it is to program parallel HW?
There's more to Verilog than just creating functional HW from raw gates. And modern processors have some advanced modal features, like conditional execution, speculative execution, which are starting to show up in compilers. But which don't get expression in high level languages like C. As the HW features get C representations, the compilers will support them with more flexibility. That all brings more of the offered solutions closer to the basic state of reconfigurable HW that Verilog exposes. So I'm looking
This is a kind of catch-22. I can tell that the Verilog developer base has parallel techniques to teach programmers of nonreconfigurable HW. I'm not a Verilog programmer (not actively, for over a decade), so I don't know what exactly are those techniques I'd like them to teach us. I know enough about sequential and parallel processing (as an FPGA programmer back then, without even VHDL or Verilog) that today's sequential languages need to be reformed for parallelization. And I know that the way languages evolve is by borrowing from others, or just describing life in other "cultures". That's why I'm encouraging parallel programmers, mostly familiar with Verilog, to improve tools like gcc sequential programmers use as more parallel features arrive for use by sequential programmers.
Like "Conservatives" in general, their words have value only in exploiting a moment for maximum political benefit, without regard to what those words will mean later. Political rhetoric is bad enough, "Conservatives" have turned hypocrisy from a momentary contradiction into a permanent stream of propaganda, beyond inconsistencies into an ocean of lies.
Whether the Conservapedia is a parody or not is not a matter of "opinion", but of fact. The fact is that it is not a parody, but real Conservatism. Which looks like a parody when collected in one place, rather than dispensed situationally inside some medium controlled by "Conservatives", because its Conservative messages combine for a worldview undeniably absurd when collected together - as it must be, to reflect the reality it attempts to control.
Pretending that wrong facts are forgivable "opinions" is a typical "Conservative" weasel trick for having it all ways, whenever it's convenient. Calling "bigotry" my simple analysis that I have backed up with facts and logic, as well as citations of the blatant reality that is the silly Conservapedia is another weasel "Conservative" trick. It's not bigotry to treat bad people badly, when they're bad by choice.
I think your comment shows a lot of nothing, except typical "Conservative" weaselly tricks to attack people without basis.
What if the new parallel processors shipped with their interconnect buses as preconfigured FPGA, but which could be dynamically reconfigured? First as point-to-point buses on per-program basis, then with inline transformation logic. That kind of hardware marches closer since it appeared on the horizon over a decade ago. When it ships, at first compilers like gcc will just import configurations to send to the processor, probably wrapped in procedural code, much like Cell processors will use PPE procedures to load SPEs with DSP code. But integrations will start to migrate. Surely incorporating some VHDL compilation techniques into the neighboring gcc code, and from there probably some VHDL programming techniques. Especially in compiler tech like branch prediction and other "preprocessed parallelization" already used to flatten parallel operations into linear sequences.
The hardest part of parallel programming is organizing the processing tasks into coherent parallel units. The existing languages and the compiler features they trigger are defined in terms of a single-threaded processor model. As we automate, even hide, thread groups and parallel streams, we'll take advantage of the parallel HW. And that will show up in our languages and programming modes. Since VHDL is already further down that road, I expect we'll see some of the same sights VHDL has shown us.
Of course if we're going to chain revolutionary new HW to backwards compatibility, or just portable single source, we're not going to use the new parallel features.
pthreads aren't everything. Programming the new Cell uPs is hard, because each the overhead of threading is wasteful on the SPEs with their fast math and relatively slow (and simple) logic.
That's why I'm saying we need new languages (and other expression modes) that don't just reflect the old HW types, with their old limitations.
As for Verilog, I'm not sure what or how to start to import into gcc for sequential processors. But I would like to see conditional function calls, argument passing in parallel from a single register (or anything that isn't a stack), more stream processing syntax. I'm sure people who program in Verilog (I don't) have features and strategies they'd like to use in gcc. That's where the "new breed" of solutions can come from. I've got enough experience to see what the "big picture" problem is, and one source of solutions. That's more than I've seen coming out, mostly because people are full of reasons why not to target the new HW through a new model for new techniques.
I've been looking for signs that a Cell BE can have FPGA attached to its pins. What I'd really like to see would be a Cell BE attached by FPGA to an RSX graphics chip, initially configured as the bus. But reconfigurable for logic inline between those two parallel processors that's as fast CLU as is their ALU(s).
Now that would turn a PlayStation into a PrayStation.
OK, these new parallel chips aren't even out yet, and software has to get the hardware before SW can improve to exploit the HW. But the HW has all the momentum, as usual. SW for parallel computing is as rudimentary as a 16bit microprocessor.
What we need is new models of computing that programmers can use, not just new tools. Languages that specify purely sequential operations on specific virtual hardware (like scalar variables that merely represent specific allocated memory hardware), or metaphors for info management that computing killed in the last century ("file cabinets", trashcans of unique items and universal "documents" are going extinct) are like speaking Latin about quantum physics.
There's already a way forward. Compiler geeks should be incorporating features of VHDL and VeriLog, inherently parallel languages, into gcc. And better "languages", like flowchart diagrams and other modes of expressing info flow, that aren't constrained by the procedural roots of those HW synthesis old guard, should spring up on these new chips like mushrooms on dewy morning lawns.
The hardware is always ahead of the software - as instructions for hardware to do what it does, software cannot do more. But now the HW is growing capacity literally geometrically, even arguably exponentially, in power and complexity beyond our ability to even articulate what it should do within what it can. Let's see some better ways to talk the walk.
The more "Conservative" you are, the more you suffer the personality defects I mentioned. One (among many) essential characteristics of "Conservatism" that's not symmetrical in liberalism is the Conservative (actually conserving something this time) value of ideological purity. Liberals are much more heterogenous, even conflicting, in their beliefs.
Personally, I am neither. I want to conserve some things, from progressive gains through the humanist era, through some of the cumulative wisdom not proven wrong or suspect by increasing sophistication and freedom, and including the environment. To do so, I am very progressive, especially in undoing the radical revisions we've suffered in the past decade or two. I'm just a person, with my complexity. I don't have the kind of truncated political view that turns people into "*ists" of whatever denomination. But it's perfectly evident that many people do have it, and are that - or at least struggle to act like that.
So my "broad brush", with its carefully sculpted tip, covers quite the right parts of the human canvas to which I apply it.
When neurons are engineered to express the halorhodopsin gene, the researchers can inhibit their activity by shining yellow light on them.
So not only would doctors have to get light inside the brain, they'd first have to genetically engineer the neurons to include and express the halorhodopsin gene. The right neurons: the ones that will later have Parkinson's Disease or whatever is being treated.
How are they going to guess which neurons? Which healthy person is going to let them genetically engineer their neurons? Those neurons are going to behave the same, though they're now expressing proteins that make them work like retinal cells?
Installing these shutdown hooks is a neat trick. But not for neurological medicine. Maybe for some biomechanics or biocomputation. Throwing genes into neurons for probing with light so violates our most absolutely personal spaces - inside our craniums and our genomes - that the cure is worse than the disease.
The US generally lacks telecom competition, except in longdistance telephone - where the rates are rock-bottom, even compared to pricing around the world. Other countries also largely lack competition, but government policies encourage capacity and feature growth. In the US, our policies encourage telecom monopolies.
If you look at places where, say, municipal broadband (whether fiber or wireless) is even beginning, you'll see some of the best economics for consumers. And consider that government policies created the Internet, and pushed it across many milestones on its way to becoming obviously indispensible. Obvious, that is, to everyone except government policymakers and the telco monopolies that love them.
There's two different phenomena we're talking about. In NYC, I get 8Mbps down, about 0.5Mbps up over cablemodem. I pay about $60:mo. That's generally as good as it gets: most neighborhoods have more cablemodem users sharing the bandwidth, so they don't get 8Mbps, but many neighborhoods don't have it at all. DSL is 1.5/0.5Mbps at most, sometimes 1.0/0.3Mbps, anywhere from $50-100:mo; some neighborhoods not at all. There is a FiOS pilot in the Financial District, supposedly 30Mbps symmetrical, but I'll believe it when I see it, and it's in 1 building, though there's another experimental pilot covering a town in a very dense suburban county. EVDO gets maybe 1Mbps, more like 0.3Mbps, when it's useable with all the multipath and the blackouts all our cells suffer from. And some neighborhoods which can't get any broadband at all.
But that's all just the maximum that a highly motivated and funded residential can get. Businesses have options including more fiber in a controlled facility than most countries can get, at dirt cheap, but that's not the market we're talking about.
The problem is in the results: much lower penetration rates than in Europe or in Asian cities. And those competitors have higher bandwidth, at lower prices. The density is clearly not the reason. It's the policies that favor telecom monopoly instead of promoting broadband adoption. I'd like to know why NC has a better landscape, but I expect it has to do with the very policies that created RTP, and has nothing to do with the density offered as an excuse for Eurasian advantages.
Then how come American cities, even more dense than European countries, don't have as much broadband penetration as outside European and Asian cities, let alone inside them?
The NYC "service area" has about 20 million people living and working in and around it, more than nearly any European country, in an area smaller than nearly any European country. How come broadband costs more for so much less bandwidth? How come so many fewer people have it? See the previous question, I guess.
And we've got other municipal centers, like Chicago, LA, Miami, Boston, Philly, all with the supposedly critical required density.
It's clear that the difference is that those other countries have technology growth policies that encourage broadband adoption, and the US doesn't. We used to, when we invented the Internet, but we don't. Instead we have government censorship of science and favoritism for monopoly technology corporations. And new politics that cut off the brain drain we used to pull on foreign countries.
We are responsible for creating our competition and failing to compete with them. It's not geography, but economics and political science that we're failing.
Conservapedia is self-parody, but it is produced and maintained by "Conservatives" as a repository of official "Conservative" dogma. Because they think Wikipedia is "liberal", as they clearly state in their About page. Typically Conservative, they're using the Wikipedia software for free, but don't even mutter a minimal thanks to Wikipedia - they just bash it.
Anonymous Conservative Coward is a typical Conservative: trying to have it both ways, all ways, whenever it's convenient. There is no "truth" for today's "Conservatives" (What are they "conserving"? They're wasters, reckless consumers and rampant destroyers.) So whenever they dart out from behind their favorite weasel words to make a clear statement, they're usually a joke, at least because they contradict whatever other statement they made before that was once convenient then.
Slashdot Mirror
And since the optical media is so cheap, make multiple copies of each disc, with data shuffled around on the surface on each copy.
That way, any "spot defects" will be very unlikely to hit the same data on every copy. Making the whole redundant backup set last many times longer.
But it's clear that Wikipedia isn't dead, not in the way Dvorak said. So why mention it, unless it's interesting that Dvorak is wrong again, fundamentally?
FWIW, the way that some authorities blacklist Wikipedia because it's "too interactive", preferring "reliable" sources like Britannica, shows how those authorities are, in some senses, dead. They'd show signs of life if instead of purely taking info from these references for their "reports", they included the reporters and validators (teachers) in the loop, and forced corrections/additions back into Wikipedia after grading them. Learning how to test references for accuracy is an essential skill in the modern infosphere. And learning how to revise references is an essential skill even more important to educational institutions.
I've seen studies showing Wikipedia more accurate overall than is Britannica. And Wikipedia will improve, while Britannica will stay the same - and never lose some of its institutional inaccuracies. That's "dead" in intellectual terms.
Well, there you go. Dvorak: wrong again.
Besides, anyone who quotes or cites Wikipedia without checking the sources it lists, or the history of the content they're referencing, is setting themselves up for a fall. Wikipedia doesn't rely in human interactivity just to accumulate content. It requires the consumer to interact, too, to keep their confidence high and correct mistakes they can tell. It's a medium for people to interact, not just a message from some people to a consumer.
Dvorak has never understood interactivity. He doesn't seem to understand people, either. He's a good compilation of crappy wisdom that's conveniently collected in one place to easily ignore.
Threads and processes are one way to do it. But they map relatively closely to only some HW configurations. And they have overhead, in each thread/process and in managing them. While their relatively monolithic concurrency is a model that only a subset of all parallel problems fall into. Thread synchronization and IPC latency are additional problems that consume additional resources.
As mentioned elsewhere, consider processing two parallel loops like c = a + b ; e = c + d . That's hard to do with threads or processes, but is parallelizable in HW, like FPGA or ASIC. The languages that do that, like VHDL and Verilog, are very different from the procedures that are encapsulated in threads and processes.
Now we're discussing an Intel chip with very parallel HW. I'm programming its competitor, the Cell uP, which has a fast Power RISC on a bus with 6-8 very fast DSPs. The point of parallelizing the HW is large compute capacity. DSP programming strategies are defined by maximizing throughput. There's no room on the DSPs for thread or process overhead. And the Power does run Linux, which could have concurrent processing on the DSPs while the program you're programming needs them. So the programming is complex, to account for the extra complexity of the HW. Partly because the OS hasn't caught up with the use cases of the HW. Both because at the early stage of the tech, so history hasn't shown some case classes to be negligible, and because these are general purpose chips that will serve many case classes, many of them new because the HW capability is new.
Those threads and processes were new once, too. They relied on context switching features of chips to evolve. Their techniques gradually became stable through long use, and migrated into the OS and compilers. Now HW has taken another jump. We' must repeat the innovation cycle again. That's what I'm calling for, and looking to the parallel HW (FPGA and ASIC) community for ideas, now that the commodity uPs are including more of their tech. Until people build new languages and compilers, it'll be challenging and frustrating to do it right, making the new HW architectures worthwhile.
They might not have legal grounds to reopen the case. But how many of these random people, many teenagers or their parents, decided against their lawyers' advice to drop it rather than pursue the counterclaim? Very few, if any. Therefore what has changed is that their lawyers have gotten smarter. That such change isn't legal grounds for getting justice is injustice.
It's probably safe to presume that these new cases finding counterclaims allowed are of the same merit (or lack) as the ones that just totally dropped, and even found the defendants liable.
What has changed? The laws are the same. The actions are the same.
Maybe the judges are a little smarter now. Maybe the lawyers are a little smarter. But if I didn't get the same results as a defendant in the same circumstances a year or more ago, I'd want a new trial. It's not supposed to be my problem if the administrators of justice are too stupid to leave me alone with laws they don't understand.
That's the new development I want to see: a retrial on the basis that maybe the courts aren't as stupid as they were when they decided against me.
Maybe I'm not getting the joke, but I don't see how the apparently totally legit mentions of Williams in Dvorak's entry and Ballmer in Williams' bio invalidate Wikipedia.
In fact, that statement by Dvorak makes him look like a troll of all trades, not just an Apple basher, considering that the wiki lives, despite years of targeting for organized vandalism.
Thanks - it sounds like Zune beats PSP on number of songs, but PSP wins on everything else (especially games :). But PSP can take a $40 4GB stick, plus a WiFi stream, so it can hold quite a lot (or unlimited amounts) of music.
I wonder why I don't hear about more PSP vs Zune competition. Maybe if Sony would release the 1.8GB UMD blanks and recorders, there would be little competition. Maybe we need a "How Sony Can Make Zune a Success" story.
How does the Zune compare to the Sony PSP, against just the PSP portable media player features?
And overall, which one is the better buy? What if you own an XBox, or if you own a PS3? How about cross-brand, is either portable anything but useless with the cross-brand console?
Who cares what Dvorak says? What reason is there to listen to him? He isn't even the guy after whom the interesting keyboard, the obscure cryptography based on it, or anything else.
He's been totally wrong about Apple all the time, including such fundamentals of whether people would use a "mouse".
Enough of this clown. Steve Jobs should just take him behind the woodshed and spank him down once and for all.
It's not like concurrent programming these days is efficient, in either processor or programmer time.
We need better software and tools for making it that are more parallel. Software doesn't adapt, people adapt software. And I'm talking about how slowly we're doing it, while suggesting new ways to get more.
What are you talking about? Don't you know anything about how challenging and frustrating it is to program parallel HW?
Physics is Greek. That's why "quantum" physics is so confusing.
There's more to Verilog than just creating functional HW from raw gates. And modern processors have some advanced modal features, like conditional execution, speculative execution, which are starting to show up in compilers. But which don't get expression in high level languages like C. As the HW features get C representations, the compilers will support them with more flexibility. That all brings more of the offered solutions closer to the basic state of reconfigurable HW that Verilog exposes. So I'm looking
This is a kind of catch-22. I can tell that the Verilog developer base has parallel techniques to teach programmers of nonreconfigurable HW. I'm not a Verilog programmer (not actively, for over a decade), so I don't know what exactly are those techniques I'd like them to teach us. I know enough about sequential and parallel processing (as an FPGA programmer back then, without even VHDL or Verilog) that today's sequential languages need to be reformed for parallelization. And I know that the way languages evolve is by borrowing from others, or just describing life in other "cultures". That's why I'm encouraging parallel programmers, mostly familiar with Verilog, to improve tools like gcc sequential programmers use as more parallel features arrive for use by sequential programmers.
Like "Conservatives" in general, their words have value only in exploiting a moment for maximum political benefit, without regard to what those words will mean later. Political rhetoric is bad enough, "Conservatives" have turned hypocrisy from a momentary contradiction into a permanent stream of propaganda, beyond inconsistencies into an ocean of lies.
Whether the Conservapedia is a parody or not is not a matter of "opinion", but of fact. The fact is that it is not a parody, but real Conservatism. Which looks like a parody when collected in one place, rather than dispensed situationally inside some medium controlled by "Conservatives", because its Conservative messages combine for a worldview undeniably absurd when collected together - as it must be, to reflect the reality it attempts to control.
Pretending that wrong facts are forgivable "opinions" is a typical "Conservative" weasel trick for having it all ways, whenever it's convenient. Calling "bigotry" my simple analysis that I have backed up with facts and logic, as well as citations of the blatant reality that is the silly Conservapedia is another weasel "Conservative" trick. It's not bigotry to treat bad people badly, when they're bad by choice.
I think your comment shows a lot of nothing, except typical "Conservative" weaselly tricks to attack people without basis.
What if the new parallel processors shipped with their interconnect buses as preconfigured FPGA, but which could be dynamically reconfigured? First as point-to-point buses on per-program basis, then with inline transformation logic. That kind of hardware marches closer since it appeared on the horizon over a decade ago. When it ships, at first compilers like gcc will just import configurations to send to the processor, probably wrapped in procedural code, much like Cell processors will use PPE procedures to load SPEs with DSP code. But integrations will start to migrate. Surely incorporating some VHDL compilation techniques into the neighboring gcc code, and from there probably some VHDL programming techniques. Especially in compiler tech like branch prediction and other "preprocessed parallelization" already used to flatten parallel operations into linear sequences.
The hardest part of parallel programming is organizing the processing tasks into coherent parallel units. The existing languages and the compiler features they trigger are defined in terms of a single-threaded processor model. As we automate, even hide, thread groups and parallel streams, we'll take advantage of the parallel HW. And that will show up in our languages and programming modes. Since VHDL is already further down that road, I expect we'll see some of the same sights VHDL has shown us.
Of course if we're going to chain revolutionary new HW to backwards compatibility, or just portable single source, we're not going to use the new parallel features.
pthreads aren't everything. Programming the new Cell uPs is hard, because each the overhead of threading is wasteful on the SPEs with their fast math and relatively slow (and simple) logic.
That's why I'm saying we need new languages (and other expression modes) that don't just reflect the old HW types, with their old limitations.
As for Verilog, I'm not sure what or how to start to import into gcc for sequential processors. But I would like to see conditional function calls, argument passing in parallel from a single register (or anything that isn't a stack), more stream processing syntax. I'm sure people who program in Verilog (I don't) have features and strategies they'd like to use in gcc. That's where the "new breed" of solutions can come from. I've got enough experience to see what the "big picture" problem is, and one source of solutions. That's more than I've seen coming out, mostly because people are full of reasons why not to target the new HW through a new model for new techniques.
I've been looking for signs that a Cell BE can have FPGA attached to its pins. What I'd really like to see would be a Cell BE attached by FPGA to an RSX graphics chip, initially configured as the bus. But reconfigurable for logic inline between those two parallel processors that's as fast CLU as is their ALU(s).
Now that would turn a PlayStation into a PrayStation.
OK, these new parallel chips aren't even out yet, and software has to get the hardware before SW can improve to exploit the HW. But the HW has all the momentum, as usual. SW for parallel computing is as rudimentary as a 16bit microprocessor.
What we need is new models of computing that programmers can use, not just new tools. Languages that specify purely sequential operations on specific virtual hardware (like scalar variables that merely represent specific allocated memory hardware), or metaphors for info management that computing killed in the last century ("file cabinets", trashcans of unique items and universal "documents" are going extinct) are like speaking Latin about quantum physics.
There's already a way forward. Compiler geeks should be incorporating features of VHDL and VeriLog, inherently parallel languages, into gcc. And better "languages", like flowchart diagrams and other modes of expressing info flow, that aren't constrained by the procedural roots of those HW synthesis old guard, should spring up on these new chips like mushrooms on dewy morning lawns.
The hardware is always ahead of the software - as instructions for hardware to do what it does, software cannot do more. But now the HW is growing capacity literally geometrically, even arguably exponentially, in power and complexity beyond our ability to even articulate what it should do within what it can. Let's see some better ways to talk the walk.
The more "Conservative" you are, the more you suffer the personality defects I mentioned. One (among many) essential characteristics of "Conservatism" that's not symmetrical in liberalism is the Conservative (actually conserving something this time) value of ideological purity. Liberals are much more heterogenous, even conflicting, in their beliefs.
Personally, I am neither. I want to conserve some things, from progressive gains through the humanist era, through some of the cumulative wisdom not proven wrong or suspect by increasing sophistication and freedom, and including the environment. To do so, I am very progressive, especially in undoing the radical revisions we've suffered in the past decade or two. I'm just a person, with my complexity. I don't have the kind of truncated political view that turns people into "*ists" of whatever denomination. But it's perfectly evident that many people do have it, and are that - or at least struggle to act like that.
So my "broad brush", with its carefully sculpted tip, covers quite the right parts of the human canvas to which I apply it.
So not only would doctors have to get light inside the brain, they'd first have to genetically engineer the neurons to include and express the halorhodopsin gene. The right neurons: the ones that will later have Parkinson's Disease or whatever is being treated.
How are they going to guess which neurons? Which healthy person is going to let them genetically engineer their neurons? Those neurons are going to behave the same, though they're now expressing proteins that make them work like retinal cells?
Installing these shutdown hooks is a neat trick. But not for neurological medicine. Maybe for some biomechanics or biocomputation. Throwing genes into neurons for probing with light so violates our most absolutely personal spaces - inside our craniums and our genomes - that the cure is worse than the disease.
The US generally lacks telecom competition, except in longdistance telephone - where the rates are rock-bottom, even compared to pricing around the world. Other countries also largely lack competition, but government policies encourage capacity and feature growth. In the US, our policies encourage telecom monopolies.
If you look at places where, say, municipal broadband (whether fiber or wireless) is even beginning, you'll see some of the best economics for consumers. And consider that government policies created the Internet, and pushed it across many milestones on its way to becoming obviously indispensible. Obvious, that is, to everyone except government policymakers and the telco monopolies that love them.
There's two different phenomena we're talking about. In NYC, I get 8Mbps down, about 0.5Mbps up over cablemodem. I pay about $60:mo. That's generally as good as it gets: most neighborhoods have more cablemodem users sharing the bandwidth, so they don't get 8Mbps, but many neighborhoods don't have it at all. DSL is 1.5/0.5Mbps at most, sometimes 1.0/0.3Mbps, anywhere from $50-100:mo; some neighborhoods not at all. There is a FiOS pilot in the Financial District, supposedly 30Mbps symmetrical, but I'll believe it when I see it, and it's in 1 building, though there's another experimental pilot covering a town in a very dense suburban county. EVDO gets maybe 1Mbps, more like 0.3Mbps, when it's useable with all the multipath and the blackouts all our cells suffer from. And some neighborhoods which can't get any broadband at all.
But that's all just the maximum that a highly motivated and funded residential can get. Businesses have options including more fiber in a controlled facility than most countries can get, at dirt cheap, but that's not the market we're talking about.
The problem is in the results: much lower penetration rates than in Europe or in Asian cities. And those competitors have higher bandwidth, at lower prices. The density is clearly not the reason. It's the policies that favor telecom monopoly instead of promoting broadband adoption. I'd like to know why NC has a better landscape, but I expect it has to do with the very policies that created RTP, and has nothing to do with the density offered as an excuse for Eurasian advantages.
Then how come American cities, even more dense than European countries, don't have as much broadband penetration as outside European and Asian cities, let alone inside them?
The NYC "service area" has about 20 million people living and working in and around it, more than nearly any European country, in an area smaller than nearly any European country. How come broadband costs more for so much less bandwidth? How come so many fewer people have it? See the previous question, I guess.
And we've got other municipal centers, like Chicago, LA, Miami, Boston, Philly, all with the supposedly critical required density.
It's clear that the difference is that those other countries have technology growth policies that encourage broadband adoption, and the US doesn't. We used to, when we invented the Internet, but we don't. Instead we have government censorship of science and favoritism for monopoly technology corporations. And new politics that cut off the brain drain we used to pull on foreign countries.
We are responsible for creating our competition and failing to compete with them. It's not geography, but economics and political science that we're failing.
Conservapedia is self-parody, but it is produced and maintained by "Conservatives" as a repository of official "Conservative" dogma. Because they think Wikipedia is "liberal", as they clearly state in their About page. Typically Conservative, they're using the Wikipedia software for free, but don't even mutter a minimal thanks to Wikipedia - they just bash it.
Anonymous Conservative Coward is a typical Conservative: trying to have it both ways, all ways, whenever it's convenient. There is no "truth" for today's "Conservatives" (What are they "conserving"? They're wasters, reckless consumers and rampant destroyers.) So whenever they dart out from behind their favorite weasel words to make a clear statement, they're usually a joke, at least because they contradict whatever other statement they made before that was once convenient then.
"Reality has a well-known liberal bias." - Stephen Colbert
Does that mean chips are going to get just as shabby as everything else we've invented in glass, and turned into cheapo plastic versions?