Usually when it is said "no new inventions" they are comparing against the flurry of activity from circa 1850-1950, and then I point out how the commercial worldwide Internet (early 1990's) and affordable cell phones (mid 1990's) fundamentally changed life. Before the 1990's cell phone, women did not go out alone at night. But this BBC article defines "recent" as "21st century" and focuses on everyone's favorite non-invention whipping boy, the iPhone.
Well, 2001-2010, has been called the "lost decade" for Microsoft, and in my opinion was the lost decade in general for a lot of organizations, market sectors, and technology areas. 9-11 set in an economic conservatism and then the housing bubble that followed (resulting from the low interest rates that were instituted to counter that economic conservatism) misdirected a huge percentage of time and effort away from productive endeavors. And then of course the Great Recession. We're only now waking up from that 12-year sleep, and there is now a lot of exciting stuff going on.
The BBC's article is like writing about a 17-year-old and saying, "he hasn't even gotten a diploma yet to show for all those years of schooling."
Besides those two life-changing inventions I started out with, though, there is a third and it is more recent: the end of physical media. One could say eBooks, flash drives, DVRs, broadband, and low-priced scanners (who remembers when they used to be $5,000?) are all incremental, but taken together, they have fundamentally changed life: libraries are obsolete, and homes no longer have to devote square footage to media storage.
Post-web people may like to read what was all the rage in the early 1990's: the deep philosophy of software development. Doug Lea has a concise summary, so you don't have to read several hundred pages. And here's a quick direct quote from Christopher Alexander I just now Googled:
...it is not possible to make something beautiful, merely by combining fixed components.
The article mentions whitelist technology as the next step beyond conventional signature-based blacklist systems. But that's what I used three years ago, with RegRun. As soon as an executable is run that it doesn't recognize, RegRun pops up an alert asking you if it's legitimate. Of course, this is useful only for the technologically savvy.
But now instead of that, I employ the ultimate in virus recovery (albeit not virus control). Using the multi-boot software BootIt Bare Metal (like a commercial version of GRUB, GParted, and other utilities rolled into one), I keep a clean OS on a separate partition that I can copy over the main partition at any time. Of course, I keep data on fileservers instead of my local hard drive.
One doesn't even need to consider climate change to mitigate the Hurrican Sandys. The sea level rise is incontrovertible, and easy to extrapolate. The minimum insurance companies should be doing is accounting for sea level rise, regardless of whether it is tied to climate change, and regardless of whether that climate change is anthropogenic, and regardless of whether interventions to mitigate said climate change would be more costly or less costly than doing nothing.
Many 1970's sets did NOT include detailed instr.
on
Has Lego Sold Out?
·
· Score: 2
To address the "well, Lego sets always had detailed instructions, at least since the 1980's," well, maybe that is so. But many sets from the 1970's did not include detailed instructions. E.g., a store would typically have a bunch of general purpose sets like Set 190, plus some model-specific sets like Moon Landing and Brick Yard.
If you have messy code, it means you're not taking pride in your work, and besides, why should the rest of us have to look at that mess? Car analogy: a '68 Mustang the color of bondo. If you can't bother to paint it, it suggests maybe you didn't bother to tune it up either.
The problem is that one individual or one company doesn't have enough bargaining power to keep an API available. Collectively, many individuals and businesses would have bargaining power. Pre-Internet, government was a practical way to collectivize action. These days, it's no longer necessary, with things like the ransom model used to release Blender 3D a decade ago and Kickstarter now.
They were able obtain a longer history of telemetry data by getting it from some guy's laptop hard and finding some mag tapes under a staircase, and they reverse engineered hard-copy blueprints with the help of retired TRW engineers into modern CAD & FEA, and determined that the RTGs were bouncing thermal energy off the dish, creating recoil -- about the same amount as a car's headlights throwing photons forwards push the car backwards.
I don't even see the mouse as complementing the keyboard. The lack of accelerator keys on web sites/browsers is frustrating, the inability to alt-tab out of the various VMs and VNCs is frustrating. Touch is going in the wrong direction.
Here's the direction computers should be going in: Intelligent User Interfaces. Computers should guess the next noun/object or verb/action and list them in descending likelihood -- kind of like IntelliSense. Quick keyboard commands 1-9 or first-letter/auto-complete select out of the prioritized list. We're so far away from that that file selector dialogs don't even default -- let alone remember to! -- sort reverse chronological. (Nor do they remember last directory, cross-application)
OK, mouse is good for panning 2D (Google Maps), and zooming and sliders. Maybe there's something touch is better at than both mouse and keyboard, but I don't know what that might be.
Long before "gamification" became a buzzword half a decade ago, those developing scientific visualization or modeling & simulation software strove to make their software applications videogame-like, i.e. interactive and engaging. Now "gamification" means incorporating a Pavlovian reward system. Even "gamification" (if we take that to mean broadly the incorporation of videogame features into non-videogame software) has suffered from the declining creativity in videogames.
I wonder how much of these improvements in accuracy are due to fundamental advances
I was wondering the same thing, and just now found this interview on Google. Perhaps someone can fill in the details.
But basically, machine learning is at its heart hill-climbing on a multi-dimensional landscape, with various tricks thrown in to avoid local maxima. Usually, humans detemine the dimensions to search on -- these are called the "features". Well, philosophically, everything is ultimately created by humans because humans built the computers, but the holy grail is to minimize human invovlement -- "unsupervised learning". According to the interview, this one particular team (the one mentioned at the end of the Slashdot summary) actually rode the bicycle with no hands and to demonstrate how strong their neural network was at determining its own features, did not guide it, even though it meant their also-excellent conventional machine learning at the end of the process would be handicapped.
The last time I looked at neural networks was circa 1990, so perhaps someone writing to an audience more technically literate than the New York Times general audience could fill in the details for us on how a neural network can create features.
The comments are from India, where the software field has not been around as long as it has been in the U.S. Attitudes on age are just now (barely) starting to come around in the U.S., and I predict they will in India as well in a few years.
I love how your entire post is moaning about the loss of business class. Suck it, one-percenter!
Due to my height and previous knee injury, I take advantage of Frontier Airlines' "Economy Plus" for the extra $25. That's one of the things that will be going away.
The shortage of doctors in the U.S. is due to the AMA cartel's control over university accreditation and corresponding rent-seeking state laws requiring accreditation. The result is speed-exams when you go visit a doctor (or maybe not see the doctor at all, but rather a "nurse practitioner").
Similarly, with legislatively reduced supply of pilots, look for cattle class throughout, with even tighter row spacing. Better keep those 747's tuned up, airlines, because you're gonna need to convert them to full economy class the way Japan uses all-economy class 747's between Osaka and Tokyo.
Don't worry, even though there won't be a business class to upgrade to with your frequent flyer miles, you'll still be able to spend your miles on magazine subscriptions.
once you add in all the overhead of scheduling work, passing messages etc, you will find that you are running at a much slower speed than the raw speeds of the GPUs would have you believe
Would you happen to know how that compares to real supercomputers?
I don't have any first-hand experience with supercomputers -- only hearing about and reading about that they also struggle against Amdahl's law.
Yeah, if only that box had some other processor specializing in scalar operations and connected to the vector processors via high bandwidth low latency link.
An i5 has four cores and is connected to the Radeon via PCIe 3.0 x8.
At very least, you are probably looking at doing some networking as or more costly than a 10GbE setup
There is no networking involved in a four-Radeon setup, just a special rackmount motherboard that has a dozen PCIe slots (because each Radeon is triple-width physically).
In November, 2001, the fastest supercomputer was 12 TFlops. You can achieve that today for less than $5,000 on your desktop by ganging together four GPGPU cards (such as the 3 TFlops Radeon 7970 for less than $500 each). Go back to 1999 and it's only 3 TFlops and to match today you wouldn't even need a special motherboard.
So just wait 11 years for the prices to come down.
Slashdot Mirror
Usually when it is said "no new inventions" they are comparing against the flurry of activity from circa 1850-1950, and then I point out how the commercial worldwide Internet (early 1990's) and affordable cell phones (mid 1990's) fundamentally changed life. Before the 1990's cell phone, women did not go out alone at night. But this BBC article defines "recent" as "21st century" and focuses on everyone's favorite non-invention whipping boy, the iPhone.
Well, 2001-2010, has been called the "lost decade" for Microsoft, and in my opinion was the lost decade in general for a lot of organizations, market sectors, and technology areas. 9-11 set in an economic conservatism and then the housing bubble that followed (resulting from the low interest rates that were instituted to counter that economic conservatism) misdirected a huge percentage of time and effort away from productive endeavors. And then of course the Great Recession. We're only now waking up from that 12-year sleep, and there is now a lot of exciting stuff going on.
The BBC's article is like writing about a 17-year-old and saying, "he hasn't even gotten a diploma yet to show for all those years of schooling."
Besides those two life-changing inventions I started out with, though, there is a third and it is more recent: the end of physical media. One could say eBooks, flash drives, DVRs, broadband, and low-priced scanners (who remembers when they used to be $5,000?) are all incremental, but taken together, they have fundamentally changed life: libraries are obsolete, and homes no longer have to devote square footage to media storage.
Post-web people may like to read what was all the rage in the early 1990's: the deep philosophy of software development. Doug Lea has a concise summary, so you don't have to read several hundred pages. And here's a quick direct quote from Christopher Alexander I just now Googled:
The article mentions whitelist technology as the next step beyond conventional signature-based blacklist systems. But that's what I used three years ago, with RegRun. As soon as an executable is run that it doesn't recognize, RegRun pops up an alert asking you if it's legitimate. Of course, this is useful only for the technologically savvy.
But now instead of that, I employ the ultimate in virus recovery (albeit not virus control). Using the multi-boot software BootIt Bare Metal (like a commercial version of GRUB, GParted, and other utilities rolled into one), I keep a clean OS on a separate partition that I can copy over the main partition at any time. Of course, I keep data on fileservers instead of my local hard drive.
Would Bloomberg prefer the 70's when everyone was getting mugged for their wristwatches and cash?
One doesn't even need to consider climate change to mitigate the Hurrican Sandys. The sea level rise is incontrovertible, and easy to extrapolate. The minimum insurance companies should be doing is accounting for sea level rise, regardless of whether it is tied to climate change, and regardless of whether that climate change is anthropogenic, and regardless of whether interventions to mitigate said climate change would be more costly or less costly than doing nothing.
To address the "well, Lego sets always had detailed instructions, at least since the 1980's," well, maybe that is so. But many sets from the 1970's did not include detailed instructions. E.g., a store would typically have a bunch of general purpose sets like Set 190, plus some model-specific sets like Moon Landing and Brick Yard.
If you have messy code, it means you're not taking pride in your work, and besides, why should the rest of us have to look at that mess? Car analogy: a '68 Mustang the color of bondo. If you can't bother to paint it, it suggests maybe you didn't bother to tune it up either.
The problem is that one individual or one company doesn't have enough bargaining power to keep an API available. Collectively, many individuals and businesses would have bargaining power. Pre-Internet, government was a practical way to collectivize action. These days, it's no longer necessary, with things like the ransom model used to release Blender 3D a decade ago and Kickstarter now.
They were able obtain a longer history of telemetry data by getting it from some guy's laptop hard and finding some mag tapes under a staircase, and they reverse engineered hard-copy blueprints with the help of retired TRW engineers into modern CAD & FEA, and determined that the RTGs were bouncing thermal energy off the dish, creating recoil -- about the same amount as a car's headlights throwing photons forwards push the car backwards.
Being in print for 25 years means it's got most single-platform computer magazines beat.
I think you meant "single-manufacturer".
I don't even see the mouse as complementing the keyboard. The lack of accelerator keys on web sites/browsers is frustrating, the inability to alt-tab out of the various VMs and VNCs is frustrating. Touch is going in the wrong direction.
Here's the direction computers should be going in: Intelligent User Interfaces. Computers should guess the next noun/object or verb/action and list them in descending likelihood -- kind of like IntelliSense. Quick keyboard commands 1-9 or first-letter/auto-complete select out of the prioritized list. We're so far away from that that file selector dialogs don't even default -- let alone remember to! -- sort reverse chronological. (Nor do they remember last directory, cross-application)
OK, mouse is good for panning 2D (Google Maps), and zooming and sliders. Maybe there's something touch is better at than both mouse and keyboard, but I don't know what that might be.
First priority is to fix keyboard UI.
Long before "gamification" became a buzzword half a decade ago, those developing scientific visualization or modeling & simulation software strove to make their software applications videogame-like, i.e. interactive and engaging. Now "gamification" means incorporating a Pavlovian reward system. Even "gamification" (if we take that to mean broadly the incorporation of videogame features into non-videogame software) has suffered from the declining creativity in videogames.
I wonder how much of these improvements in accuracy are due to fundamental advances
I was wondering the same thing, and just now found this interview on Google. Perhaps someone can fill in the details.
But basically, machine learning is at its heart hill-climbing on a multi-dimensional landscape, with various tricks thrown in to avoid local maxima. Usually, humans detemine the dimensions to search on -- these are called the "features". Well, philosophically, everything is ultimately created by humans because humans built the computers, but the holy grail is to minimize human invovlement -- "unsupervised learning". According to the interview, this one particular team (the one mentioned at the end of the Slashdot summary) actually rode the bicycle with no hands and to demonstrate how strong their neural network was at determining its own features, did not guide it, even though it meant their also-excellent conventional machine learning at the end of the process would be handicapped.
The last time I looked at neural networks was circa 1990, so perhaps someone writing to an audience more technically literate than the New York Times general audience could fill in the details for us on how a neural network can create features.
At my workplace we're going even more retro. On Monday I'm going to sign up on the internal IRC network
Listserv is 1986 and IRC is 1988.
If so, may I ask Eastern-EU folks to please refrain from hacking my servers during the holiday season?
At least 10 countries have just been given the green light for hacking.
I'm surprised at the number of people who don't know the skill of dealing with telemarketers on the telephone: "I'm not interested"
Dealing with recruiters is similar and simple: "I currently make $X, and would consider a change for a 30% increase."
The comments are from India, where the software field has not been around as long as it has been in the U.S. Attitudes on age are just now (barely) starting to come around in the U.S., and I predict they will in India as well in a few years.
I love how your entire post is moaning about the loss of business class. Suck it, one-percenter!
Due to my height and previous knee injury, I take advantage of Frontier Airlines' "Economy Plus" for the extra $25. That's one of the things that will be going away.
The shortage of doctors in the U.S. is due to the AMA cartel's control over university accreditation and corresponding rent-seeking state laws requiring accreditation. The result is speed-exams when you go visit a doctor (or maybe not see the doctor at all, but rather a "nurse practitioner").
Similarly, with legislatively reduced supply of pilots, look for cattle class throughout, with even tighter row spacing. Better keep those 747's tuned up, airlines, because you're gonna need to convert them to full economy class the way Japan uses all-economy class 747's between Osaka and Tokyo.
Don't worry, even though there won't be a business class to upgrade to with your frequent flyer miles, you'll still be able to spend your miles on magazine subscriptions.
once you add in all the overhead of scheduling work, passing messages etc, you will find that you are running at a much slower speed than the raw speeds of the GPUs would have you believe
Would you happen to know how that compares to real supercomputers?
I don't have any first-hand experience with supercomputers -- only hearing about and reading about that they also struggle against Amdahl's law.
Yeah, if only that box had some other processor specializing in scalar operations and connected to the vector processors via high bandwidth low latency link.
An i5 has four cores and is connected to the Radeon via PCIe 3.0 x8.
What is feeding them work, coordinating their inputs/outputs, etc? That is where all the hard work is.
OpenCL uses C99. It's tricky, maybe even "hard", but far from impossible.
At very least, you are probably looking at doing some networking as or more costly than a 10GbE setup
There is no networking involved in a four-Radeon setup, just a special rackmount motherboard that has a dozen PCIe slots (because each Radeon is triple-width physically).
Supercomputers measure double precision FLOPS while the GPGPU vendor cheat and report single precision.
Ah, OK, Radeon is then 1 TFlop for double precision (which is new to the Radeon). So four Radeon 7970's beat the top 1999 supercomputer.
In November, 2001, the fastest supercomputer was 12 TFlops. You can achieve that today for less than $5,000 on your desktop by ganging together four GPGPU cards (such as the 3 TFlops Radeon 7970 for less than $500 each). Go back to 1999 and it's only 3 TFlops and to match today you wouldn't even need a special motherboard.
So just wait 11 years for the prices to come down.