In their defense, I'm sure "Rummy & Wolfie" knew there was no way, in this day and age, that we could sustain three to four hundred thousand troops in Iraq for any significant length of time. We just don't have the divisions anymore.
We'll have enough trouble sustaining 100,000+ for a couple years, and the resulting extended deployments will probably play hell with Army recruiting & retainment for years to come.
Actually, while we may have an overwhelming firepower advantage in Iraq, we hardly have overwhelming numbers.
A general rule-of-thumb for occupations is 1 soldier per 40 inhabitants. Having less means you don't have enough troops to adequately control activities on the ground.
We have something like 1 soldier for every 160 Iraqis there today.
"The US unilaterally undertook not to use, and to place in inactive stockpile status with the intent to demilitarize by the end of 1999, all nonself-destructing AP mines not needed to train personnel engaged in demining and countermine operations, and to defend the United States and its allies from armed aggression that crosses the Korean Demilitarized Zone."
So the only place we use non-self destructing AP mines is in Korea under controlled conditions.
Sampling only produces an estimation of behavior. If the sample group isn't representative of the whole, then the estimate won't be either.
What happens if I create a tune that only 100 people worldwide decide to download and listen to, but none are part of the sample group? I would get nothing.
Sampling is fine for determining what "most" people are listening to, but it fails to reward the fringes.
Besides, on the Internet, everyone participating in the licensing scheme could be part of the sample group.
I think using a Nielson-like scheme to determine the value of works is inherently flawed in this system. It relies on sampling a relatively small number of households who have access to a relatively small number of potential choices.
This will inherently concentrate value towards the handful of songs and artists preferred by the sample group.
For this to produce 'fair' results, all paying customers would have to be part of the sample group.
Instead, perhaps the distribution of money should be left up to the license purchaser. If I want my $5 this month to go to 'Ice Ice Baby', then so be it!
P2P software & media players could, by default, record downloading & listening habits to form a basic percentage allocation, which I could modify each month, if I felt like it.
In short, I don't think you'd ever tell a computer "Get me that thing I need for that meeting on that day" unless you knew that it wouldn't work and you were only trying to prove a point.
Unfortunately, I think I would make well-meaning, ambiguous requests to the computer, because I'm often distracted, lazy, or just unable to verbalize my stream of consciousness.
Personally, I'm not a huge fan of 'natural language' interfaces. My interaction with a natural language interface would be more like, "Hey, get me that thing I needed for that meeting I'm supposed to go to on that day."
Natural language is far too ambiguous, IMHO, and requires a lifetime's worth of context to be practical.
In a hundred years, I hope that we aren't using textual languages at all. Hopefully, biomechanical engineering progresses to the point where a person's thoughts can directly interact with the computer.
The 'language' used would be derived from the way our brains process information, not the way we historically communicated with other humans. Our descendants should think 'the square root of pie', and instantly 'know' the answer from a computation performed by the computer. Obviously, in this example, traditional language is still used, but other interations are possible. A person could think about their child and instantly feel their status - sleeping in the crib - and even see them projected in a mental image, like a very detailed daydream.
I do agree with the author about one thing. Programming languages will evolve from specifying how a program accomplishes the goal, to describing the goal itself. Design and development is then reduced to formulating a precise, unambiguous description of the goal.
With this in mind, we should look to logical and functional languages for guidance, and not traditional imperative languages like Java & C/C++.
I don't see the point of comparing the Q3 benchmarks for these cards any more. Q3 is an extremely dated engine. IIRC, it doesn't even use the early fixed-function hardware T&L, let alone any pixel or vertex shaders. It is, in no way, a predictor for how fast newer, heavy T&L, shader-enhanced games will run.
Indrema didn't have a $79 bill of materials. It was meant to compete with the XBoxes and PS2s of the world which are sold for a loss. This required Indrema to have a ton of cash to get the ball rolling.
This Intel device is not meant as a game console and doesn't have any of the expensive 3D hardware or CPU power. It's basically a PDA-level core with an 802.11b radio and TV out and could probably be sold for around a hundred bucks retail.
Hmm, I bet it'd make a nifty MAME-style emulator platform. Nothing like paying a hundred bucks for a device you could get on eBay for pennies..
The basic design could be useful though. I bet you could make a neat answering machine out of the core components.
Features:
-Automatically puts you messages on a secure web site or emails them for remote access.
-TeleZaps those annoying Telemarketers.
-Separate voicemail boxes for everyone in the house.
-Automatically routes calls based on Caller ID.
-Simple calendar, contact list, etc. through the phone using voice recognition/keypad and Text-To-Speech or over the Web.
-Internet telephony.
Jeez, maybe I should quit my job & make this puppy!;)
I actually went out of my way & spent over an hour watching the BMW short films on their web site (BMWFilms). I even watched the director voice-over & "making of" extras. They rock!
Slashdot Mirror
We (mkt10.com) don't require MSIE. We test with IE, Firefox and Safari. We do require Javascript and AJAX support.
Let me know if you have any specific problems.
Thx,
Brent Smith
Mkt10 Engineering Team
In their defense, I'm sure "Rummy & Wolfie" knew there was no way, in this day and age, that we could sustain three to four hundred thousand troops in Iraq for any significant length of time. We just don't have the divisions anymore.
We'll have enough trouble sustaining 100,000+ for a couple years, and the resulting extended deployments will probably play hell with Army recruiting & retainment for years to come.
So? 14-year old kids can wear suicide belts and fire RPGs & AK-47s too.
Actually, while we may have an overwhelming firepower advantage in Iraq, we hardly have overwhelming numbers.
A general rule-of-thumb for occupations is 1 soldier per 40 inhabitants. Having less means you don't have enough troops to adequately control activities on the ground.
We have something like 1 soldier for every 160 Iraqis there today.
A Proven Formula for How Many Troops We Need
So it should be obvious why we're having these problems today.
The US already does this.
From FAS
"The US unilaterally undertook not to use, and to place in inactive stockpile status with the intent to demilitarize by the end of 1999, all nonself-destructing AP mines not needed to train personnel engaged in demining and countermine operations, and to defend the United States and its allies from armed aggression that crosses the Korean Demilitarized Zone."
So the only place we use non-self destructing AP mines is in Korea under controlled conditions.
Sampling only produces an estimation of behavior. If the sample group isn't representative of the whole, then the estimate won't be either.
What happens if I create a tune that only 100 people worldwide decide to download and listen to, but none are part of the sample group? I would get nothing.
Sampling is fine for determining what "most" people are listening to, but it fails to reward the fringes.
Besides, on the Internet, everyone participating in the licensing scheme could be part of the sample group.
I think using a Nielson-like scheme to determine the value of works is inherently flawed in this system. It relies on sampling a relatively small number of households who have access to a relatively small number of potential choices. This will inherently concentrate value towards the handful of songs and artists preferred by the sample group.
For this to produce 'fair' results, all paying customers would have to be part of the sample group.
Instead, perhaps the distribution of money should be left up to the license purchaser. If I want my $5 this month to go to 'Ice Ice Baby', then so be it!
P2P software & media players could, by default, record downloading & listening habits to form a basic percentage allocation, which I could modify each month, if I felt like it.
Basic economics. Value is driven by demand.
In short, I don't think you'd ever tell a computer "Get me that thing I need for that meeting on that day" unless you knew that it wouldn't work and you were only trying to prove a point.
Unfortunately, I think I would make well-meaning, ambiguous requests to the computer, because I'm often distracted, lazy, or just unable to verbalize my stream of consciousness.
Personally, I'm not a huge fan of 'natural language' interfaces. My interaction with a natural language interface would be more like, "Hey, get me that thing I needed for that meeting I'm supposed to go to on that day."
Natural language is far too ambiguous, IMHO, and requires a lifetime's worth of context to be practical.
In a hundred years, I hope that we aren't using textual languages at all. Hopefully, biomechanical engineering progresses to the point where a person's thoughts can directly interact with the computer.
The 'language' used would be derived from the way our brains process information, not the way we historically communicated with other humans. Our descendants should think 'the square root of pie', and instantly 'know' the answer from a computation performed by the computer. Obviously, in this example, traditional language is still used, but other interations are possible. A person could think about their child and instantly feel their status - sleeping in the crib - and even see them projected in a mental image, like a very detailed daydream.
I do agree with the author about one thing. Programming languages will evolve from specifying how a program accomplishes the goal, to describing the goal itself. Design and development is then reduced to formulating a precise, unambiguous description of the goal.
With this in mind, we should look to logical and functional languages for guidance, and not traditional imperative languages like Java & C/C++.
I don't see the point of comparing the Q3 benchmarks for these cards any more. Q3 is an extremely dated engine. IIRC, it doesn't even use the early fixed-function hardware T&L, let alone any pixel or vertex shaders. It is, in no way, a predictor for how fast newer, heavy T&L, shader-enhanced games will run.
How is Linguana better than WordNet when they haven't actually done anything?
Can SD media hold 500MB and only cost a few dollars each?
Indrema didn't have a $79 bill of materials. It was meant to compete with the XBoxes and PS2s of the world which are sold for a loss. This required Indrema to have a ton of cash to get the ball rolling. This Intel device is not meant as a game console and doesn't have any of the expensive 3D hardware or CPU power. It's basically a PDA-level core with an 802.11b radio and TV out and could probably be sold for around a hundred bucks retail.
;)
Hmm, I bet it'd make a nifty MAME-style emulator platform. Nothing like paying a hundred bucks for a device you could get on eBay for pennies..
The basic design could be useful though. I bet you could make a neat answering machine out of the core components.
Features:
-Automatically puts you messages on a secure web site or emails them for remote access.
-TeleZaps those annoying Telemarketers.
-Separate voicemail boxes for everyone in the house.
-Automatically routes calls based on Caller ID.
-Simple calendar, contact list, etc. through the phone using voice recognition/keypad and Text-To-Speech or over the Web.
-Internet telephony.
Jeez, maybe I should quit my job & make this puppy!
Make better commercials!
Make me want to watch them.
I actually went out of my way & spent over an hour watching the BMW short films on their web site (BMWFilms). I even watched the director voice-over & "making of" extras. They rock!