Exemplary Technologies: Chapter 7, CEO did time in Club Fed
Network Associates: the division I was at got annihilated
... I'm currently in graduate school because I've come to the belief that the Industry is just not worth it. Whenever I hear HR reps talking about how they're irked that young engineers have no company loyalty and will abandon ship for the next good offer to come along, I want to shake them vigorously and shout that we've got no loyalty to employers because it's been proven the employer has no loyalty to us.
You know, Chris, I could just drop an email off to you and ask this, but... what the hell: you might get some more campaign contributions this way.
What are the California campaign contribution laws like? Can out-of-staters contribute, or does it have to be through a PAC or the like? Can you give us a campaign headquarters address for mailing checks, etc., stuff like that?
(I worked with Chris a few years ago at a failed dot-com startup. Let me tell you, you think your dot-com stories are bad...:) )
but screw up big time when you confuse energy requirements with some sort of cataclysmic power surge.
It's not a power argument; it's purely an energy argument. That much energy in our cosmos would lead to utter catastrophe, which shouldn't be surprising given that it's orders of magnitude greater than the total energy content of the universe. I literally cannot imagine the scope of the damage, and I'm generally pretty good at imagining spacetime deformations.
w.r.t. solving chess, there are such things as tree pruning and of course not all states have to be stored, they are merely searched.
Ah, I see. So we get around the storage requirements by dynamically generating our states as we search through them... which will lead to us regenerating states over and over again, thus increasing the energy requirements over that of the naieve method. Sorry; this doesn't work, either.
Re: tree pruning, that's not what the original poster was talking about. The OP was talking about fully mapping the chess space, which is simply not going to happen.
Not entirely certain, but it's easy enough to come up with some guesstimates.
A Go board is 19x19 and a game is finished only when the entire board is covered. Each point on the board can, in the endgame state, possess either of two states: black stone or white stone.
So take 19*19 and you get 361. Since each of the stones can be in two states, we come up with an upper bound of 2^361 (or 4.7 * 10^108) different endgame states. Some of these states will be unreachable and would later be pruned, so this number is best used as an upper bound on the states and not an accurate count.
Given that there are only 10^35 different states a chess board can be in (as someone corrected me on; the 10^150 number is closer to the amount of different moves, not the number of different states; it's possible to get to the same state by many different paths), we can see the upper bound for the different endgame states of Go just clobbers the upper bound of the different endgame states of chess.
It's the actual solution of the problem that requires the tree, not the part about writing a program which implements the solution once it is known. Basically solving this type of an optimization problem involves calculating a "score" for each of the states that tells you the relative value of being in a given state versus another.
First, I think neurodynamic programming is very cool; we're covering it right now in one of my algorithmic analysis courses.
That said, I think we're at loggerheads here; you're talking about sensible and smart ways to approach the chess game computationally, while I'm talking about how the naieve way to approach the chess game (a conventional decision tree and some kind of depth-first or breadth-first algorithm) is totally impractical.
I don't disagree with anything you've written; I think it's pretty interesting, honestly. I just think I'm trying to say something different.:)
Please note how I phrased things: it requires the Landauer Bound of energy to set a bit (i.e., to discard the previous state and to give it a new state). This is correct.
The way Feynman's reversible computers evade the Landauer Bound is that they never discard state. They manage to beat the Landauer Bound, but the Second Law is still upheld.
I'll also say that the first difficulty in your analysis comes in by stating how much energy it takes to flip a bit.
It's called the Landauer Bound. It's the thermodynamic minimum required to erase one bit of information (or, for any von Neumann computer, the energy required to flip, set or erase a bit). If you can go below this level, then you've just broken the Second Law of Thermodynamics. You suspect that my energy estimate will go down over time? I suspect the Second Law will prevail.
All right, so you have a huge amount of numbers storable on a relatively small amount of space, from what I'm gathering. You know, as a layman. And it goes on to say that operations can be perfomed simultaneously on all these states.
Yes. Now sit down and do the math. Figure out how many qubits you need to solve a problem of that magnitude. Figure out the minimum energy required for those qubits. Figure out whether or not the decoherence problem can be resolved for that many qubits (this one is completely unknown). And don't forget to prove that the chess space lends itself to a quantum solution; one of the things we've formally proven about quantum computing is that there are many, many problems which cannot be efficiently solved on quantum hardware.
All of these are mathematical and/or physical limitations, not technological or engineering limitations.
You say that, no matter what, there can be no possible situation where we could work out the entire tree.
Not until computers are made of something other than matter and occupy something other than space and run on something other than energy. Every model we currently have says it's not going to happen. If you want to say "well, it will happen!", then you need to first present a different model.
It's like saying that I'm "anti-scientific-revolutions" because I believe in the Second Law of Thermodynamics. I'm not anti-revolution. I'm anti-idiocy. Ignoring the Second Law just because "well, someday we'll figure out how to get around it" is a fool's errand.
For now, for the forseeable future, and very possibly for all time, the entire chess space will be an intractable problem. This is not anti-science. This is not prejudice. This is postjudice; taking a look at the corpus of knowledge and making decisions about what avenues of research are likely to bear fruit and which are likely to be barren.
If you want to sit there and say "hey, you're a stick-in-the-mud who doesn't believe in scientific progress," go right ahead. As for me, I'm going to take a look down one road and say "as near as I can tell, that's a dead end" and look down another road and say "hey, I don't know what's at the end there, but it looks pretty cool."
Carl Sagan, himself a great advocate of both skepticism and open-mindedness, once came up with a remark for people like you. "They laughed at Fulton. They laughed at the Wright brothers. They laughed at Einstein. But they also laughed at Bozo the Clown."
Just to nitpick a little (since you're a math person, I thought you'd appreciate having your terms right):
Absolutely. Thank you for this correction; I was going from a recollection of Norvig's number. On the other hand, the 10^120 sequences of moves is also a killer; sure, you only have to store 10^35 boards at 256 bits each (which is still an outrageously large number), but since you're turning it into a decision tree you need to keep track of which moves take you into which other states. Storing 10^120 pointers for the tree leads you into a similar thermodynamic catastrophe.
I'll happily concede the 10^35 different boards, but I think the tree links themselves are enough to keep my original line of reasoning intact.
However, noting that the state-space size is large isn't really a very useful observation, since chess programs these days don't try to map out the entire tree of possible outcomes.
I'm aware of this. My response was meant only as a refutation of the (incorrect, IMO) notion that the game space of chess would eventually be fully mapped.
I have very little doubt that the ChessBase article is spectacularly wrong. When the author says that so far "both sides are holding their own", he's not exactly on the same planet the rest of us are. In the 1970s, chess programs were trivial toys that children could beat. In thirty years they've ascended to the very top rank. When the gap between man and machine closes that much in just a few years, you don't say "both sides are holding their own"; you say "the machines are really moving up".
The first mistake is in assuming that you have to store board state in its entierty. We've invented this thing called 'compression.'
I'm trying to keep from laughing, but I'm mostly failing.:) It's just funny as hell is all.
In order to save on thermodynamic costs, you're going to do more processing--i.e., statistical analysis to remove redundancy--to reduce the number of bits required per board. But in the process of doing this, you're going to be setting and flipping orders of magnitude more bits than you would just by a naieve algorithm. So instead of 10^126 joules, now we're looking at (conservatively) 10^130 joules. Sure, you save space in your decision tree, but you do so at a significant thermodynamic cost, and it's the thermodynamic cost which you've got to minimize in the first place.
Do you see why this is so funny?
Then, let's consider advances in technology that aren't merely software. Let's say...Quantum Computing. Ah, those clever Quanta. You could store multiple bits of information on a single atom, hypothetically.
Please, please, please, take a class in algorithmic analysis or formal theories of computation. You are making many errors here, the first of which is you don't understand the computational theory behind quantum computation. You are turning QC into a magical talisman capable of solving anything, when the reality is quite different (and quite well-known to algorithmic researchers and computational theorists).
SOOOooo, what you're saying is that before we can even begin to build that game tree, we'd have to become a Type IV civilization first?
Not necessarily. I don't want to predict what the future will hold; it's possible we'll see reversible computers take hold, in which case we can compute with infinitesimal energy. A reversible computer could compute the entire decision tree using a nine-volt battery for a power supply, although you'd still have the problem of finding a hard drive larger than the universe on which to write your answers, and you'd have to wait a potentially infinite amount of time to get your answers, since reversible computing isn't exactly snappy.
Superpositional ("quantum") computing could potentially reduce the size requirements by a large degree, but reducing 10^150 states by a "large degree" still means leaving you with a Godawful number of entries. If you reduce the total number of states by 10^149 (a very large reduction), you still have approximately 10^150 states. Exponential growth is interesting that way.
I cannot imagine any way--even positing reversible computing and superpositional (i.e., quantum) computing--to reduce it to a solvable problem space. But then again, the universe is weirder than I can imagine, so I'll hedge my bet a little and just say "I cannot imagine it", not "it can't be done".
While the total number of states in Tic Tac Toe is a boringly small finite number, the total number of states in chess is rather amusingly large. And by "amusingly large", I should point out that I'm a large number theorist.
How large is "amusingly large"? Around 10^150, if I remember my AI class correctly. Discarding entirely the problem of how you'd create a game tree of that size (given the cosmos has about 10^77 particles), let's just address the energy required to compute the table.
It requires an absolute minimum of kT*ln2, or about 3*10^-26 Joules, of energy to set a bit. Each cell on a chess board requires a minimum of four bits to store its state (it has to store a three-bit enum { PAWN, ROOK, KNIGHT, BISHOP, QUEEN, KING } and a one-bit enum { BLACK, WHITE }). So for a 64-block chess grid, you're looking at 256 bits just to store state.
256 * 3*10^-26 = 7.7*10^-24
7.7 * 10^-24 * 10^150 = 7.7 * 10^126
Do you have any freaking clue how much energy 10^126 Joules is? It's frickin' huge. Like enough to cause a symmetry-breaking event which would propagate through the universe at the speed of light and utterly annihilate everything in its path, including the computer churning out the complete decision tree for chess.
I can see it now. When Judgment Day comes, it's all going to be because of a Slashdotter who thinks he knows a lot more about what computers can and can't do than he really does, and goes off to solve unsolvable problems without considering the thermodynamic consequences of his actions.
Is there some reason we are talking about bolt-action rifles and handguns, not semi-automatics?
We have been talking about automatic rifles. Or were you not paying attention to the discussion about the weight of the M-16A2, G3A3, etc.? I'm bringing in other weapons and other mechanisms because it shows the trend is not limited to assault and battle rifles.
Good grief, man, think about it: if weapons are made heavier to handle more energetic rounds (more precisely, as you say, rounds with more momentum), then why is the M-16 as light as it is? You say it's because the M-16 was specifically designed for the.223 Remington, but that's not true; Eugene Stoner designed the M-16 based off the AR-15, and the original AR-15 was just a.308 AR-10 rechambered for.221 Fireball. (Once the Air Force agreed to purchase AR-15s, the AR-15 chambering was changed again to.223 Remington.) The AR-10 was the breakthrough lightweight assault rifle, not the AR-15 or M-16. So if every weapon designer "knows" that you always make weapons heavier to offset momentum, why is it the most brilliant weaponsmith since John Moses Browning (i.e., Eugene Stoner) had his biggest breakthrough slashing the weight of a full-caliber battle rifle by several pounds?
If what you say is true, then the M-16A2 should be heavier than it is by several pounds, since it's originally designed to chamber a cartridge with far, far more momentum.
And where did you get the nine-and-a-half-pound rifle figure you mention later on?
Jane's Weapons of the World lists it as between 4.3 and 4.6 kilograms, depending on its particular type.
It is the momentum of the round leaving the barrel that is the main consideration for how heavy a rifle should be. Yes, caliber alone is not the deciding factor, but in order to make an accurate rifle that uses a heavy round as light as one that does not one must reduce the velocity of the round.
You've clearly never done any serious shooting before.
For instance, a.45 Glock 21 is a very, very light weapon (about 1.7 pounds) due to its extensive use of synthetics--a clear exception to your "weapons of larger caliber are heavier" rule, considering that a Beretta M92 weighs considerably more despite the 9mm round's lower momentum.
Why and how can Glock get away with this? Because of synthetics; the Glock uses a lot of high-impact polymers, whereas the Beretta is solid steel. Why is the Beretta solid steel and the Glock polymer? Because of economics.
In other words, it's economic forces which dominate a weapon's weight--not caliber.
As another example, a.408 CheyTac costs about five grand and packs a massive.408 round into a weapon that's lighter than a Remington 700. More momentum, less weight--but the CheyTac is far more expensive.
In firearms, you pay for what you don't get. In this case, weight.
Where? I am forced to admire your facilities for invention. My actual point was that the particular woman pictured was too small for the particular weapon pictured.
You're claiming that a fit woman is incapable of carrying a nine-and-a-half-pound rifle, when other women of far less fitness routinely carry fifteen- and twenty-pound children around all day.
Whatever it is that you're smoking, do you mind if I get some? It seems to have some truly wonderful mind-altering effects.
Actually, I did. But you're arguing trivialties here, and more than that, incorrect trivialties. A Remington 700 chambered in.223 Remington weighs 11 pounds; an HK G3K in.308 weighs eight. You keep on trying to argue that "women are too small to handle full-caliber weapons", and completely ignore the fact that most weapons don't weigh all that much.
When you're called on the subject, you hide behind trivially-refuted tripe like "the weight of a rifle is determined primarily by its ammunition type", which is a non-sequitur (it doesn't have anything to do with your original statement) as well as incorrect (weapon weight is determined by many factors, and caliber does not dominate the weight consideration).
The weight of a rifle is determined mainly by the ammunition used.
M-16A2; 8.5 pounds loaded, with a magazine which weighs just about a pound. In other words, the weight of the rifle is determined only 12% by the ammunition used.
A G3A3 battle rifle; about 11 pounds loaded, with a magazine which weighs about two pounds. In other words, for a battle rifle chambered in the very hefty 7.62mm NATO round, the weight is determined only 20% by the ammunition used.
Notice how the lady obviously does not have the upper body strength required to even hold up the weapon
A fully-loaded M-16A2 with undermounted grenade launcher weighs a total of about twelve pounds. My mother could carry that today, and she's not exactly young.
A Barrett M82A1--which is a big honkin' piece of hardware (click here)--weighs about thirty pounds. So as a rough guess, I'd say the hardware she's carrying there in the screenshot (a rifle I can't place, although it looks vaguely Dragunovish, which would make it nine and a half pounds fully loaded) weighs no more than twelve pounds.
So either you don't know jack about women, if you think they can't carry a twelve-pound weapon, or you don't know jack about firearms. Take your pick which.:)
(Up until her 50th birthday, my mother could beat me in skeet shooting. Her preferred rig was a twelve-gauge firing three-inch magnum rounds. So please don't tell me "real women can't handle hardware like that." Real women, especially real women who come from Alabama and grow up hunting and fishing with their fathers, most definitely do handle hardware like that.)
In military formalisms, a rifle is an assault rifle if and only if it (a) is a rifle which (b) fires an intermediate-power cartridge (c) selectively (d) from a magazine.
Rifles which fire full-power cartridges, like the M-14 and G3, are correctly termed "battle rifles", as are intermediate-caliber weapons which only fire semiautomatically.
This is really fundamental and amazing stuff, if it turns out to be true. (Every decade or so someone discovers monopoles, then six months later, oops!, guess we didn't.) The reason why it's fundamental and amazing is the conflict between relativity and quantum mechanics.
Relativity and quantum mechanics currently give physicists nightmares. As near as we can tell, both are fantastically accurate descriptions of the world, and both are fundamentally at odds with each other. If we could find a flaw in one or the other, that would potentially open the door to a new and better theory, one which might allow us to reconcile these differences without resorting to theology [*].
Now, when Einstein devised relativity, he based it very heavily on Maxwell's Laws. The Laws are a set of four equations which describe pretty much all electromagnetic phenomena out there. It was the world's first Grand Unified Theory (GUT), in that it unified electricity and magnetism into one package. And one of Maxwell's Laws ("the divergence of the magnetic field equals zero") has, as a direct consequence, an absolute law: NO MAGNETIC MONOPOLES EXIST. NO OPEN MAGNETIC FIELD LINES EXIST.
So if Maxwell's Laws are wrong and relativity is built heavily on Maxwell's Laws, then there's a tantalizing chance relativity is wrong.
Or if Maxwell's Laws are right and monopoles are conclusively proven not to exist, then there's a tantalizing chance quantum mechanics is wrong.
Either way, physics wins: no matter what happens, we get to see a flaw in our current theories. And seeing that flaw is the first step to coming up with better, more accurate theories.
What's the worst-case scenario? The worst-case scenario is these guys are wrong, just like every other monopole researcher before, and the "do magnetic monopoles exist?" question remains unresolved for the next few hundred years.
Warning: I'm not a professional physics geek. In fact, I may be stark barking mad wrong here.:)
[*] Theology, aka string theory. Sorry, but any theory which literally cannot be experimentally tested at any realistic energy level isn't a theory at all. It's an article of faith.
No, it's not. There are several areas of the HTML spec which conflict with the XML spec; for instance, all attributes must be enclosed in quotation marks in XML, whereas HTML allows naked attributes. XML isn't a superset of HTML if there are things HTML allows which XML doesn't.
Admittedly, I'm no XML or HTML guru; but from reading HTML to XHTML transition guidelines, it certainly does appear that there are unavoidable problems with trying to treat normal HTML as a subset of XML.
XHTML, on the other hand, is a subset of XML (or, rather, it's just XML with a specific DTD). That's why I use XHTML for all my own web pages.
And then when XML-only browsers pop up, all these old pages become unviewable.
Yes. Just like you can't view a WordStar 6.0 document in a Web browser.
Free hint: XML is not HTML. It's close, but it's not the same. Any HTML document that is conformant to a given HTML specification can be rendered by any competent HTML browser that's conformant to that specification. If you don't believe me, I can find some very, very old web pages that far predate the 4.01 standard, yet are conformant to the standard of their time, and Firebird renders them perfectly.
Saying "yeah, and when XML-only browsers pop up, all these old pages become unviewable" is a trivial statement. If it's XML-only, then it's not a freaking HTML browser, and it makes no sense to complain that an XML-only browser can't grok HTML. Just like it makes no sense to complain that Firebird can't render WordStar 6.0 documents.
if the only people you play against are approximately equal in skill level with you, how are any of you going to ever get better?
Pet peeve: people who think that the amount of fun you have in a game must be directly proportional to how good you are at the game. Why should I have to devote myself to getting better at a game in order to have fun?
I'm a moderately good chess player. I enjoy playing chess. It's fun. I could be a hell of a lot better if I were to take the game more seriously, but I want the game to be fun, not a job.
I'm a poor Q3 player. But I enjoy playing regardless, even if I usually do wind up in the lower half of the scores at local LANfests.
Nobody should have to get better at a game as a prerequisite for having fun.
Period.
And people who don't understand this simply don't understand the most important thing about game design. Namely--it's not whether you win or lose, or how skilled or unskilled you are. It's about fun. Everything else is just candy.
I doubt a single manager said "Hey, let's kill some astronaunts."
In many jurisdictions, acting with a "reckless disregard for human life" is enough to warrant second-degree murder.
I.e., if you're doing 10 over the speed limit and you hit a pedestrian and kill them, that's manslaughter (or vehicular manslaughter, or what-have-you). Were you reckless? Yeah. Could you reasonably expect someone would die as a result? No, because we all do 10 over the limit all the time and nobody gets hurt.
Doing 40 over the limit in a school zone ten minutes after school's out, hitting and killing a kid, is a different story. Were you reckless? Yeah. Could you reasonably expect someone would die as a result? Yeah. Did someone die? Yeah. Presto: murder two.
Examining all the risks of landing and coming to an informed decision to take the risks, and seven astronauts dying as a result, is a tragedy.
Not examining the risks--and denying well-grounded requests to examine the risks--and seven astronauts dying as a result, could very well be "reckless disregard for human life" and thus murder.
Slashdot Mirror
- MCI-WorldCom: in Chapter 11
- McLeodUSA: in Chapter 11
- Exemplary Technologies: Chapter 7, CEO did time in Club Fed
- Network Associates: the division I was at got annihilated
... I'm currently in graduate school because I've come to the belief that the Industry is just not worth it. Whenever I hear HR reps talking about how they're irked that young engineers have no company loyalty and will abandon ship for the next good offer to come along, I want to shake them vigorously and shout that we've got no loyalty to employers because it's been proven the employer has no loyalty to us.I'm 29, and I want out.
You know, Chris, I could just drop an email off to you and ask this, but... what the hell: you might get some more campaign contributions this way.
:) )
What are the California campaign contribution laws like? Can out-of-staters contribute, or does it have to be through a PAC or the like? Can you give us a campaign headquarters address for mailing checks, etc., stuff like that?
(I worked with Chris a few years ago at a failed dot-com startup. Let me tell you, you think your dot-com stories are bad...
Every user of the Internet.
but screw up big time when you confuse energy requirements with some sort of cataclysmic power surge.
It's not a power argument; it's purely an energy argument. That much energy in our cosmos would lead to utter catastrophe, which shouldn't be surprising given that it's orders of magnitude greater than the total energy content of the universe. I literally cannot imagine the scope of the damage, and I'm generally pretty good at imagining spacetime deformations.
w.r.t. solving chess, there are such things as tree pruning and of course not all states have to be stored, they are merely searched.
Ah, I see. So we get around the storage requirements by dynamically generating our states as we search through them... which will lead to us regenerating states over and over again, thus increasing the energy requirements over that of the naieve method. Sorry; this doesn't work, either.
Re: tree pruning, that's not what the original poster was talking about. The OP was talking about fully mapping the chess space, which is simply not going to happen.
Not entirely certain, but it's easy enough to come up with some guesstimates.
A Go board is 19x19 and a game is finished only when the entire board is covered. Each point on the board can, in the endgame state, possess either of two states: black stone or white stone.
So take 19*19 and you get 361. Since each of the stones can be in two states, we come up with an upper bound of 2^361 (or 4.7 * 10^108) different endgame states. Some of these states will be unreachable and would later be pruned, so this number is best used as an upper bound on the states and not an accurate count.
Given that there are only 10^35 different states a chess board can be in (as someone corrected me on; the 10^150 number is closer to the amount of different moves, not the number of different states; it's possible to get to the same state by many different paths), we can see the upper bound for the different endgame states of Go just clobbers the upper bound of the different endgame states of chess.
It's the actual solution of the problem that requires the tree, not the part about writing a program which implements the solution once it is known. Basically solving this type of an optimization problem involves calculating a "score" for each of the states that tells you the relative value of being in a given state versus another.
:)
First, I think neurodynamic programming is very cool; we're covering it right now in one of my algorithmic analysis courses.
That said, I think we're at loggerheads here; you're talking about sensible and smart ways to approach the chess game computationally, while I'm talking about how the naieve way to approach the chess game (a conventional decision tree and some kind of depth-first or breadth-first algorithm) is totally impractical.
I don't disagree with anything you've written; I think it's pretty interesting, honestly. I just think I'm trying to say something different.
Please note how I phrased things: it requires the Landauer Bound of energy to set a bit (i.e., to discard the previous state and to give it a new state). This is correct.
The way Feynman's reversible computers evade the Landauer Bound is that they never discard state. They manage to beat the Landauer Bound, but the Second Law is still upheld.
I'll also say that the first difficulty in your analysis comes in by stating how much energy it takes to flip a bit.
It's called the Landauer Bound. It's the thermodynamic minimum required to erase one bit of information (or, for any von Neumann computer, the energy required to flip, set or erase a bit). If you can go below this level, then you've just broken the Second Law of Thermodynamics. You suspect that my energy estimate will go down over time? I suspect the Second Law will prevail.
All right, so you have a huge amount of numbers storable on a relatively small amount of space, from what I'm gathering. You know, as a layman. And it goes on to say that operations can be perfomed simultaneously on all these states.
Yes. Now sit down and do the math. Figure out how many qubits you need to solve a problem of that magnitude. Figure out the minimum energy required for those qubits. Figure out whether or not the decoherence problem can be resolved for that many qubits (this one is completely unknown). And don't forget to prove that the chess space lends itself to a quantum solution; one of the things we've formally proven about quantum computing is that there are many, many problems which cannot be efficiently solved on quantum hardware.
All of these are mathematical and/or physical limitations, not technological or engineering limitations.
You say that, no matter what, there can be no possible situation where we could work out the entire tree.
Not until computers are made of something other than matter and occupy something other than space and run on something other than energy. Every model we currently have says it's not going to happen. If you want to say "well, it will happen!", then you need to first present a different model.
It's like saying that I'm "anti-scientific-revolutions" because I believe in the Second Law of Thermodynamics. I'm not anti-revolution. I'm anti-idiocy. Ignoring the Second Law just because "well, someday we'll figure out how to get around it" is a fool's errand.
For now, for the forseeable future, and very possibly for all time, the entire chess space will be an intractable problem. This is not anti-science. This is not prejudice. This is postjudice; taking a look at the corpus of knowledge and making decisions about what avenues of research are likely to bear fruit and which are likely to be barren.
If you want to sit there and say "hey, you're a stick-in-the-mud who doesn't believe in scientific progress," go right ahead. As for me, I'm going to take a look down one road and say "as near as I can tell, that's a dead end" and look down another road and say "hey, I don't know what's at the end there, but it looks pretty cool."
Carl Sagan, himself a great advocate of both skepticism and open-mindedness, once came up with a remark for people like you. "They laughed at Fulton. They laughed at the Wright brothers. They laughed at Einstein. But they also laughed at Bozo the Clown."
Just to nitpick a little (since you're a math person, I thought you'd appreciate having your terms right):
Absolutely. Thank you for this correction; I was going from a recollection of Norvig's number. On the other hand, the 10^120 sequences of moves is also a killer; sure, you only have to store 10^35 boards at 256 bits each (which is still an outrageously large number), but since you're turning it into a decision tree you need to keep track of which moves take you into which other states. Storing 10^120 pointers for the tree leads you into a similar thermodynamic catastrophe.
I'll happily concede the 10^35 different boards, but I think the tree links themselves are enough to keep my original line of reasoning intact.
However, noting that the state-space size is large isn't really a very useful observation, since chess programs these days don't try to map out the entire tree of possible outcomes.
I'm aware of this. My response was meant only as a refutation of the (incorrect, IMO) notion that the game space of chess would eventually be fully mapped.
I have very little doubt that the ChessBase article is spectacularly wrong. When the author says that so far "both sides are holding their own", he's not exactly on the same planet the rest of us are. In the 1970s, chess programs were trivial toys that children could beat. In thirty years they've ascended to the very top rank. When the gap between man and machine closes that much in just a few years, you don't say "both sides are holding their own"; you say "the machines are really moving up".
The first mistake is in assuming that you have to store board state in its entierty. We've invented this thing called 'compression.'
:) It's just funny as hell is all.
I'm trying to keep from laughing, but I'm mostly failing.
In order to save on thermodynamic costs, you're going to do more processing--i.e., statistical analysis to remove redundancy--to reduce the number of bits required per board. But in the process of doing this, you're going to be setting and flipping orders of magnitude more bits than you would just by a naieve algorithm. So instead of 10^126 joules, now we're looking at (conservatively) 10^130 joules. Sure, you save space in your decision tree, but you do so at a significant thermodynamic cost, and it's the thermodynamic cost which you've got to minimize in the first place.
Do you see why this is so funny?
Then, let's consider advances in technology that aren't merely software. Let's say...Quantum Computing. Ah, those clever Quanta. You could store multiple bits of information on a single atom, hypothetically.
Please, please, please, take a class in algorithmic analysis or formal theories of computation. You are making many errors here, the first of which is you don't understand the computational theory behind quantum computation. You are turning QC into a magical talisman capable of solving anything, when the reality is quite different (and quite well-known to algorithmic researchers and computational theorists).
SOOOooo, what you're saying is that before we can even begin to build that game tree, we'd have to become a Type IV civilization first?
Not necessarily. I don't want to predict what the future will hold; it's possible we'll see reversible computers take hold, in which case we can compute with infinitesimal energy. A reversible computer could compute the entire decision tree using a nine-volt battery for a power supply, although you'd still have the problem of finding a hard drive larger than the universe on which to write your answers, and you'd have to wait a potentially infinite amount of time to get your answers, since reversible computing isn't exactly snappy.
Superpositional ("quantum") computing could potentially reduce the size requirements by a large degree, but reducing 10^150 states by a "large degree" still means leaving you with a Godawful number of entries. If you reduce the total number of states by 10^149 (a very large reduction), you still have approximately 10^150 states. Exponential growth is interesting that way.
I cannot imagine any way--even positing reversible computing and superpositional (i.e., quantum) computing--to reduce it to a solvable problem space. But then again, the universe is weirder than I can imagine, so I'll hedge my bet a little and just say "I cannot imagine it", not "it can't be done".
While the total number of states in Tic Tac Toe is a boringly small finite number, the total number of states in chess is rather amusingly large. And by "amusingly large", I should point out that I'm a large number theorist.
How large is "amusingly large"? Around 10^150, if I remember my AI class correctly. Discarding entirely the problem of how you'd create a game tree of that size (given the cosmos has about 10^77 particles), let's just address the energy required to compute the table.
It requires an absolute minimum of kT*ln2, or about 3*10^-26 Joules, of energy to set a bit. Each cell on a chess board requires a minimum of four bits to store its state (it has to store a three-bit enum { PAWN, ROOK, KNIGHT, BISHOP, QUEEN, KING } and a one-bit enum { BLACK, WHITE }). So for a 64-block chess grid, you're looking at 256 bits just to store state.
256 * 3*10^-26 = 7.7*10^-24
7.7 * 10^-24 * 10^150 = 7.7 * 10^126
Do you have any freaking clue how much energy 10^126 Joules is? It's frickin' huge. Like enough to cause a symmetry-breaking event which would propagate through the universe at the speed of light and utterly annihilate everything in its path, including the computer churning out the complete decision tree for chess.
I can see it now. When Judgment Day comes, it's all going to be because of a Slashdotter who thinks he knows a lot more about what computers can and can't do than he really does, and goes off to solve unsolvable problems without considering the thermodynamic consequences of his actions.
Typical for Slashdot.
I'm done with this thread.
.222 Special, not .221 Fireball. My goof.
Except for a minor (pedantic) correction: the AR-15 was an AR-10 rechambered for
Is there some reason we are talking about bolt-action rifles and handguns, not semi-automatics?
.223 Remington, but that's not true; Eugene Stoner designed the M-16 based off the AR-15, and the original AR-15 was just a .308 AR-10 rechambered for .221 Fireball. (Once the Air Force agreed to purchase AR-15s, the AR-15 chambering was changed again to .223 Remington.) The AR-10 was the breakthrough lightweight assault rifle, not the AR-15 or M-16. So if every weapon designer "knows" that you always make weapons heavier to offset momentum, why is it the most brilliant weaponsmith since John Moses Browning (i.e., Eugene Stoner) had his biggest breakthrough slashing the weight of a full-caliber battle rifle by several pounds?
We have been talking about automatic rifles. Or were you not paying attention to the discussion about the weight of the M-16A2, G3A3, etc.? I'm bringing in other weapons and other mechanisms because it shows the trend is not limited to assault and battle rifles.
Good grief, man, think about it: if weapons are made heavier to handle more energetic rounds (more precisely, as you say, rounds with more momentum), then why is the M-16 as light as it is? You say it's because the M-16 was specifically designed for the
If what you say is true, then the M-16A2 should be heavier than it is by several pounds, since it's originally designed to chamber a cartridge with far, far more momentum.
And where did you get the nine-and-a-half-pound rifle figure you mention later on?
Jane's Weapons of the World lists it as between 4.3 and 4.6 kilograms, depending on its particular type.
I'm done with this thread.
It is the momentum of the round leaving the barrel that is the main consideration for how heavy a rifle should be. Yes, caliber alone is not the deciding factor, but in order to make an accurate rifle that uses a heavy round as light as one that does not one must reduce the velocity of the round.
.45 Glock 21 is a very, very light weapon (about 1.7 pounds) due to its extensive use of synthetics--a clear exception to your "weapons of larger caliber are heavier" rule, considering that a Beretta M92 weighs considerably more despite the 9mm round's lower momentum.
.408 CheyTac costs about five grand and packs a massive .408 round into a weapon that's lighter than a Remington 700. More momentum, less weight--but the CheyTac is far more expensive.
You've clearly never done any serious shooting before.
For instance, a
Why and how can Glock get away with this? Because of synthetics; the Glock uses a lot of high-impact polymers, whereas the Beretta is solid steel. Why is the Beretta solid steel and the Glock polymer? Because of economics.
In other words, it's economic forces which dominate a weapon's weight--not caliber.
As another example, a
In firearms, you pay for what you don't get. In this case, weight.
Where? I am forced to admire your facilities for invention. My actual point was that the particular woman pictured was too small for the particular weapon pictured.
You're claiming that a fit woman is incapable of carrying a nine-and-a-half-pound rifle, when other women of far less fitness routinely carry fifteen- and twenty-pound children around all day.
Whatever it is that you're smoking, do you mind if I get some? It seems to have some truly wonderful mind-altering effects.
Actually, I did. But you're arguing trivialties here, and more than that, incorrect trivialties. A Remington 700 chambered in .223 Remington weighs 11 pounds; an HK G3K in .308 weighs eight. You keep on trying to argue that "women are too small to handle full-caliber weapons", and completely ignore the fact that most weapons don't weigh all that much.
When you're called on the subject, you hide behind trivially-refuted tripe like "the weight of a rifle is determined primarily by its ammunition type", which is a non-sequitur (it doesn't have anything to do with your original statement) as well as incorrect (weapon weight is determined by many factors, and caliber does not dominate the weight consideration).
The weight of a rifle is determined mainly by the ammunition used.
M-16A2; 8.5 pounds loaded, with a magazine which weighs just about a pound. In other words, the weight of the rifle is determined only 12% by the ammunition used.
A G3A3 battle rifle; about 11 pounds loaded, with a magazine which weighs about two pounds. In other words, for a battle rifle chambered in the very hefty 7.62mm NATO round, the weight is determined only 20% by the ammunition used.
Try again.
Check another picture of the sniper rifle, here. It's clearly a Dragunov, although only God knows what specific variant it is.
Nine and a half pounds, fully loaded.
Dunno about you, but all the real women I know can carry nine and a half pound weapons without any problem.
Notice how the lady obviously does not have the upper body strength required to even hold up the weapon
:)
A fully-loaded M-16A2 with undermounted grenade launcher weighs a total of about twelve pounds. My mother could carry that today, and she's not exactly young.
A Barrett M82A1--which is a big honkin' piece of hardware (click here)--weighs about thirty pounds. So as a rough guess, I'd say the hardware she's carrying there in the screenshot (a rifle I can't place, although it looks vaguely Dragunovish, which would make it nine and a half pounds fully loaded) weighs no more than twelve pounds.
So either you don't know jack about women, if you think they can't carry a twelve-pound weapon, or you don't know jack about firearms. Take your pick which.
(Up until her 50th birthday, my mother could beat me in skeet shooting. Her preferred rig was a twelve-gauge firing three-inch magnum rounds. So please don't tell me "real women can't handle hardware like that." Real women, especially real women who come from Alabama and grow up hunting and fishing with their fathers, most definitely do handle hardware like that.)
All rifles are assault rifles.
In military formalisms, a rifle is an assault rifle if and only if it (a) is a rifle which (b) fires an intermediate-power cartridge (c) selectively (d) from a magazine.
Rifles which fire full-power cartridges, like the M-14 and G3, are correctly termed "battle rifles", as are intermediate-caliber weapons which only fire semiautomatically.
This is really fundamental and amazing stuff, if it turns out to be true. (Every decade or so someone discovers monopoles, then six months later, oops!, guess we didn't.) The reason why it's fundamental and amazing is the conflict between relativity and quantum mechanics.
:)
Relativity and quantum mechanics currently give physicists nightmares. As near as we can tell, both are fantastically accurate descriptions of the world, and both are fundamentally at odds with each other. If we could find a flaw in one or the other, that would potentially open the door to a new and better theory, one which might allow us to reconcile these differences without resorting to theology [*].
Now, when Einstein devised relativity, he based it very heavily on Maxwell's Laws. The Laws are a set of four equations which describe pretty much all electromagnetic phenomena out there. It was the world's first Grand Unified Theory (GUT), in that it unified electricity and magnetism into one package. And one of Maxwell's Laws ("the divergence of the magnetic field equals zero") has, as a direct consequence, an absolute law: NO MAGNETIC MONOPOLES EXIST. NO OPEN MAGNETIC FIELD LINES EXIST.
So if Maxwell's Laws are wrong and relativity is built heavily on Maxwell's Laws, then there's a tantalizing chance relativity is wrong.
Or if Maxwell's Laws are right and monopoles are conclusively proven not to exist, then there's a tantalizing chance quantum mechanics is wrong.
Either way, physics wins: no matter what happens, we get to see a flaw in our current theories. And seeing that flaw is the first step to coming up with better, more accurate theories.
What's the worst-case scenario? The worst-case scenario is these guys are wrong, just like every other monopole researcher before, and the "do magnetic monopoles exist?" question remains unresolved for the next few hundred years.
Warning: I'm not a professional physics geek. In fact, I may be stark barking mad wrong here.
[*] Theology, aka string theory. Sorry, but any theory which literally cannot be experimentally tested at any realistic energy level isn't a theory at all. It's an article of faith.
However, XML is a *superset* of html
No, it's not. There are several areas of the HTML spec which conflict with the XML spec; for instance, all attributes must be enclosed in quotation marks in XML, whereas HTML allows naked attributes. XML isn't a superset of HTML if there are things HTML allows which XML doesn't.
Admittedly, I'm no XML or HTML guru; but from reading HTML to XHTML transition guidelines, it certainly does appear that there are unavoidable problems with trying to treat normal HTML as a subset of XML.
XHTML, on the other hand, is a subset of XML (or, rather, it's just XML with a specific DTD). That's why I use XHTML for all my own web pages.
And then when XML-only browsers pop up, all these old pages become unviewable.
Yes. Just like you can't view a WordStar 6.0 document in a Web browser.
Free hint: XML is not HTML. It's close, but it's not the same. Any HTML document that is conformant to a given HTML specification can be rendered by any competent HTML browser that's conformant to that specification. If you don't believe me, I can find some very, very old web pages that far predate the 4.01 standard, yet are conformant to the standard of their time, and Firebird renders them perfectly.
Saying "yeah, and when XML-only browsers pop up, all these old pages become unviewable" is a trivial statement. If it's XML-only, then it's not a freaking HTML browser, and it makes no sense to complain that an XML-only browser can't grok HTML. Just like it makes no sense to complain that Firebird can't render WordStar 6.0 documents.
if the only people you play against are approximately equal in skill level with you, how are any of you going to ever get better?
Pet peeve: people who think that the amount of fun you have in a game must be directly proportional to how good you are at the game. Why should I have to devote myself to getting better at a game in order to have fun?
I'm a moderately good chess player. I enjoy playing chess. It's fun. I could be a hell of a lot better if I were to take the game more seriously, but I want the game to be fun, not a job.
I'm a poor Q3 player. But I enjoy playing regardless, even if I usually do wind up in the lower half of the scores at local LANfests.
Nobody should have to get better at a game as a prerequisite for having fun.
Period.
And people who don't understand this simply don't understand the most important thing about game design. Namely--it's not whether you win or lose, or how skilled or unskilled you are. It's about fun. Everything else is just candy.
I doubt a single manager said "Hey, let's kill some astronaunts."
In many jurisdictions, acting with a "reckless disregard for human life" is enough to warrant second-degree murder.
I.e., if you're doing 10 over the speed limit and you hit a pedestrian and kill them, that's manslaughter (or vehicular manslaughter, or what-have-you). Were you reckless? Yeah. Could you reasonably expect someone would die as a result? No, because we all do 10 over the limit all the time and nobody gets hurt.
Doing 40 over the limit in a school zone ten minutes after school's out, hitting and killing a kid, is a different story. Were you reckless? Yeah. Could you reasonably expect someone would die as a result? Yeah. Did someone die? Yeah. Presto: murder two.
Examining all the risks of landing and coming to an informed decision to take the risks, and seven astronauts dying as a result, is a tragedy.
Not examining the risks--and denying well-grounded requests to examine the risks--and seven astronauts dying as a result, could very well be "reckless disregard for human life" and thus murder.
IANAL.