where a logical person could deduce a result of said existing technologies
In the U.S., the standard used is a "practitioner having ordinary skill in the art", (US Code Title 35, Section 103) not a "logical person".
To counter the other usual nonsense propagated by Slashdot headlines, claims may be dependent, in that some claims state what is conventional or known, and then a later claim describes a (novel and non-obvious) extension or improvement that constitues the grounds for novelty and non-obviousness. The common technique in Slashdot articles is to mention only the general claim, such as Claim 1 in this case, and crying "how obvious! That's been around since the stone age!" without mentioning the dependent claims 6, or 9, where the non-obvious part is stated.
The relevant date is not shipment but first silicon.
That's when you finally get the chance to run test cases at full speed instead of in an extremely slow software simulation.
It's not surprising that you will find additional bugs. The number of bugs found between first silicon and shipment is also a function of how large your team is and how developed your test cases are, not just the quality of your chip.
There isn't some giant cycle of design-build-test-change design; there's that kind of cycle (practically daily) in the software model for design and test, but once you get to tapeout and fab, that's pretty much a singular event. There isn't time to change the design and fab again. You just verify what you thought you knew from design simulations, run new tests that happen at full speed, document what new things you can find, decide if it is good enough, and ship; or, you wait for rev B to tape out and have first silicon. But that's probably multiple months, so you don't plan to do the Rev A-0 just for testing; you plan to ship it, and only a drastic failure will make you wait.
So there wasn't a Pentium III verification team? (rest of your bullshit clipped).
There almost certainly was; however, due to overlaps in development, test, and release schedules, Intel doesn't shift en masse from one chip to the next, but is forced to pipeline chip development; Pentium III test folks might have been approaching peak effort when the Pentium IV test team was just getting assembled. (I don't know the chronology).
The rest of your post implies that Intel doesn't know what the limits of verification technology are; I'm guessing that the company that basically invented the microprocessor knows a little bit about how to make sure microprocessor designs work; they were doing multi-month simulations on large scale clusters, and I'm guessing they could have saved a bunch of money if they could have done these calculations with less effort and the same level of accuracy. Maybe they are stupid, but hey, they only delivered chips that millions of people use everyday running things that range from Windows laptops to supercomputing clusters.
Maybe you work for AMD or Freescale or IBM, and can speak from experience, but from what I know of the folks who work at these places, they all have tremendous respect for the abilities of their competition, and hardly blow them off as incompetents.
The problem with *full* verification as opposed to random instruction testing is that the state space is far too large. Every instruction combined with every possible high-level register configuration combined with every possible contents of the relevant memory locations? Even on an 8-bit microprocessor like, say, the 6502, that's something like 2^21 * 2^40 * 2^8 = 6 * 10^20 combinations. At 1 GHz testing rate, that's about 6.8 million years. You have no chance of even enumerating these, much less testing them. Perhaps you can verify huge chunks of this by deduction, but that assumes your formal model is accurate and your proofs are correct.
As for modular design, in fact, in the very first page of the linked article from Intel, they talk about "cluster-level", i.e., modular, testing in the software model simulations that you claim they neglected. Perhaps you missed the whole section on "Cluster-Level Testing" talking about the six "clusters" into which the Pentium IV could be divided?
Furthermore, the architecture that allows these chips to execute the IA-32 instruction set so fast is a pipeline that overlaps simultaneous execution of multiple instructions, with a high degree of coupling, because these instructions might have interdependencies and much speculative computation is going on. Logically enumerating all the possibilities is a huge combinatorial explosion that defies analysis. Those are the kind of "side-effects" that you claimed shouldn't exist. Also keep in mind that the memory architecture is full of side-effects.
I don't know what kind of testing you do, but I'm pretty confident it has nothing to do with modern scale microprocessors, or you'd drop your posturing pretty quick.
Silicon fab facilities are extremely expensive and capital intensive, but they produce shitloads of chips. The process scales; making 1000 wafers in these fabs is as easy as making one.
Engineering analysis of complex IC designs is a perfect example of combinatorial explosion. Each bit of state in the chip doubles the state space in which bugs can exist. Yes, *most* of that state is in the cache which has regularity in its structure, but that regularity didn't happen by accident: it was *designed* that way.
You can only test to a spec, and if the spec is imperfect and has gaps, you will leave space for bugs. Given that specs are written by engineers, they cannot be nearly complete for anything other than the most trivial circuits; the infrastructure used to suppor engineering of non-trivial circuits could itself have bugs.
The part of the spec that covers the cache is simple, and can conceivably be error-free and well-tested, and perhaps with methods that are amenable to mathematical proof. But that's not where the errors crop up. The errors crop up in the hugely complex mechanisms that handle all the pipelining, branch prediction, translation to microinstructions, handling of interrupts, etc., etc., that are not highly regular and modular and are not easy to spec, and are not easy to approach with formal methods.
their core "product" still weighs in at 1.3k, fits on a 640x480 monitor, and has a single significant input field.
I think this view of their "product" is totally naive.
Google makes squat from that blank page. They make lots of money from sticking tiny, unobtrusive, but still lucrative ads on all sorts of websites, including their own.
As long as they keep finding new ways to stick their ads all over the place on pages people want to view, and the ads stay lucrative, www.google.com itself could vanish, and GOOG would keep making money.
That stonewalling sure is mysterious, and cheers up those who want to stick it to the current Justice Department, but I think the real problem for Google is they have *shitloads* of data on what web sites folks visit, and this is possibly fundamental to their ability to offer relevant ads that are their revenue source.
Letting anybody with a subpoena go trawling through this ocean of data could really open Google up to having to support a lot of litigation, and probably opens up serious liability issues for Google itself, although IANAL, so I couldn't really say.
That hit on their stock price could be due to the market realizing "hey, Google isn't using magic powers, but is dependent on some huge, mysterious apparatus that might be outlawed or sued into oblivion." Sounds like a real issue to me.
Google has to deal with the Web as it is right now, not as it was in the ancient past of 1999; that's not to excuse them, but crap results could mean the Web has a lot more crap now. It's very hard for a computer to make these decisions, but I'm sure Google has lots of smart people trying to fight off the hordes of people who want to pillage the clicks of the billions of visits to Google.
In any case, I believe if you are logged into your Google account, you get a link "Remove Result" on your searches, which probably fees back in some way to improve future search results.
To clarify, now that I have a bit more time to elaborate, all information-theoretic calculations assume some underlying probability distribution for the elements of the information sequence.
In a chemical distribution, for example, if you are going to make some kind of information-content calculation, you are simply calculating a measure based on a probability weight for each molecule in the distribution.
There are, however, serious constraints on the ability of a chemical reaction to occur; each input reactant has a certain chemical energy, and the various output products have a certain chemical energy, and the good old conservation of energy demands that the input energy equal the output energy. There are additional thermodynamic constraints, because, for instance the concentration of products changes the "chemical potential" in a way to favor the reverse over the forward reaction, etc. That's why chemists introduce other quantities such as entropy (the chemical/thermodynamic/physical entropy, not simply information theory entropy) and enthalpy, to capture the effect of these macroscopic constraints on the microscopic probability that a reaction will happen.
Faced with this complexity (and, remember, a catalyst or enzyme does not cause reactions that are thermodynamically forbidden, just changes the *rate* of the reaction, because it offers a kinetic pathway that is more favorable; again, involving microscopic probabilities), what does your theory offer?
(e.g. log n each element is equally probable).
So this is the first problem: by wandering off into information theory, we have yet another way in which ID'ers can mess up the probabilities; classically, they came up with the fallacy of the "747 assembled by tornado", because they used an irrelevant uniform probability distribution for the various possible arrangements of, depending on their context, proteins in a flagellum, or whatever. Natural selection and inheritance, however, provides a probability distribution that depends on the parent, and strongly favors, of course, proteins resembling flagellar structures, but having some other function, if one exists. When you start with a 747 and a tornado sweeps through, you get a 747 with some stuff moved around. The distribution "747 parts lying around" is of very *low* prior probability, so it does not strongly enter into the probability of the outcome of a tornado.
Now, by adding another layer of mumbo-jumbo, you get to screw up the probabilities in a fresh-sounding way. Dembski does a lot of this, and this is the basis for devastating critiques of his CSI theory. The article you linked to had a false calculation of evolution probabilities, assuming multiple mutations occuring simultaneously, when they could (much more likely) occur sequentially.
Information theoretic entropy is just a measure of probability. Given the repeated failure of IDers to get probabilistic calculations right, this is just the same game by a different name.
Fine, you decide to take the perfectly fine physical term "entropy" and use it to describe some measure of a chemical distribution. Warning, you are in information theory territory, where it is exceedingly hard to connect back to the *physical* world, where biology happens.
There is still no connection between "specified complexity" and anything else relevant to evolution. In particular, it places no constraints on genetic sequences and the production of those through reproduction or mutation.
The problem is that accountants (and business types in general) as opposed to economic theorists, don't have the luxury of going out and selling something when they have to estimate its value. Yet, as conservative professionals, they must still come up with a "realistic" estimate; i.e. "as they really are" given that the sale is not going to happen as opposed to "as they would turn out if we sold this whole company tomorrow".
Theoretically, every accountant would come up with the same estimates, so when a transaction did happen, the buyer's and seller's teams of accountants would both come up with the same Platonic Value for the enterprise, and the negotiations would be very quick indeed. The fact that investment bankers get paid lots of money when these things happen means it isn't so simple.
This is all very tricky terminology. "Realistic" in your sense depends on whose perspective you adopt: the seller "how much can I sucker the buyer into paying", the buyer "how low can I go and still have this transaction happen" and the actual transaction that occured "where in the space between these two prices did the accident of one-time negotiations arrive". The first two depend at least partly on psychological factors. The third is the classic market view, where the equilibrium point leaves no surplus for either side in that most marginal transaction; but most huge acquisitions don't happen in a very liquid way that corresponds to a market. Then, once the transaction happens, the buyer turns into a potential seller, and his "realistic" price goes up, because he thinks he got something worth more than he paid for, unless he finds out he made a bad deal.
I would say "realistic" depends on an objective observer, with access to all the possible information, cooly evaluating all the factors, and coming up with a true value. But that's what you call "ideal."
As they say, in theory there is no difference between theory and reality; in reality, there is.
The whole notion of the "output distribution" of an enzyme determining something about the "information content" of the enzyme is just silly.
It means several things that make the "information content" concept useless to reason with
1) if I take away the reactants, there's less output; hence, less entropy change per enzyme molecule per time. Does it therefore have "less information"? If so, where is the information, in the reactants or in the enzyme?
2) similarly, if I lower the temperature or change the pH of the environment, to reduce the reaction rate, either of the catalyzed reaction, or some reaction that is providing the reactants, the reaction rate changes. Where's the information going now?
3) Entropy is an extensive quantity, proportional to the size of the system. If I simply increase the quantity of enzyme, am I increasing the "information content" of the enzyme? Or is the "information content" normalized per molecule? In which case a genetic change that increases the production of the enzyme would increase activity without increasing the "information content" of the enzyme.
4) If the enzyme keeps reacting for a long period of time, the entropy change increases over time (as does the energy flow that allows the reaction to happen). Does the "information content" of the enzyme depend on the lifetime of the cell?
A quantity you call "specified complexity" which depends
- on the environment (say, if the pub is open and you can drink alcohol, which an enzyme in your body must break down, or the pub is closed on Sundays, and the enzyme isn't causing a reaction) - on the simple length of time the enzyme is working - on the simple bulk amount of the enzyme (a 200 pound adult vs. a 10 pound infant have the same gene, but much different quantities of the enzyme)
doesn't seem to say much about the actual genetic content of the organism, which is the most compelling description of the "complexity" of an organism.
The only "realistic" price for anything is what it sells for.
Yes, in some theoretical sense, but not in the sense that the transaction is a rational one to make. Buyers can make mistakes. Anyway, once you own it, there's no guarantee that someone else would pay the same price, so this theoretical price has changed already; without a buyer ready to shake on the deal, you have to come up with a value through another analysis.
There is an intrinsic value in an enterprise, which is the present value of the stream of revenues that the enterprise will produce in the future. Nobody has a crystal ball to know the future stream of revenues in advance, nor do they even necessarily have a good value for the appropriate discount rate to apply. So the true value is shrouded in mystery.
Sigh.. There is no entropy change due to enzyme action when you include the enzyme itself, since the information that the product distribution has gained is of course already present in the enzyme! (Hence: "[...] the information content of the enzyme itself is at least the maximum information gained in transforming the substrate distribution into the product distribution".) The relevant entropy difference is between two such enzymes of which one has mutated.
If you think this is accurate, you are using a definition for entropy that is incompatible with the usual scientific one.
The whole point of an enzyme is that it acts as a catalyst; that is, after the reaction, the enzyme returns to the same state as it was before the reaction, and is able to catalyze the reaction for more reactants to produce more products.
If the enzyme returns to the same state, it has the same entropy, because the entropy of the enzyme is a function of the state.
I suppose you think this is bullshit, that I'm just saying because I hope to convince you; probably, if I claim 1+1=2, you think I am making an arrogant claim, I don't know. Maybe you want to use perfectly good scientific terms such as "entropy" in some strange way.
In any case the mutation does NOT happen in any particular copy of the enzyme. It happens in the genetic sequence that codes for the production of the enzyme, which is then created in a different form in the mutated organism.
But perhaps you are using some unusual meaning for the term "enzyme" and "mutation." Whatever. If you insist on these basic facts which are in contradiction to the basic definitions of the terms, you are hopeless, and it is no wonder you are persuaded by ID bullshit. Believe whatever you want, just don't claim you have scientific reasons for doing so.
The fuzziness, however, is that no one can really be sure that the stock or acquisition price is realistic. All you can count is the tangible assets (and, any goodwill left over from past acquisitions by the company, if you believe it is still accurate). If you grossly overpay for a company at the height of some stock market frenzy, you could be giving up cold, hard cash for speculative froth by the name of "goodwill".
Furthermore, you can't practically acquire Google for its quoted market cap. That market cap is based on the latest market transaction of a few thousand shares, not on what the price would be to buy 295 million shares in a single transaction.
Once again you think that saying something makes it so. Your ego must be the size of a small country.
No, my saying something doesn't make it so.
REALITY makes it so, I'm just reporting it. If you don't comprehend REALITY, that is your problem not mine. If you can't realize meaningless bullshit, that's your loss.
I went back and double-checked the "information content of proteins" section of that long screed, and it was, as far as I could tell, either BS, or trivial thermodynamics, depending on how those symbols are supposed to be interpreted.
The entropy change due to enzyme action is simple physical chemistry.
The only possible physical argument would be that the physical processes of life are forbidden by thermodynamics, but, since I'm living and breathing, evidently enzymes can and do work.
THERE IS NO ARGUMENT that leads from there to saying anything about evolution, and in fact the screed makes none.
It just denotes that information is measured as specified complexity instead of the more commonly (at least in comp.sci.) used definition of complexity by improbability (by Claude Shannon).
Whatever. You give some link to a anti-evolution web site to define your "specified complexity" quantity. That's your arbitrary definition. The arguments made on the basis of these numbers are bullshit, irrelevant, or both. The whole screed you link to is a bunch of strawmen.
1) No one claims that evolution only happens by transcription errors in the genome. So all the astronomical probabilities and timescales quoted are irrelevant. 2) The "requirements" stated for evolution, requiring a single step-by-step "mutate then select" is false. There's plenty of genetic drift going on without selection that can cause variation without selective pressure. There are also very fast ways, involving mutations in particular regulatory genes that can cause huge changes with very small mutations. 3) The math behind the odds of modification assume statistical independence between generations and between mutations. This is obviously wrong; once a change happens, it doesn't revert back immediately; it's "sticky", and persists in the descendants of the mutated form with HIGH probability. Plus, it ignores the effect of selection entirely!
I could go on and on. But, rest assured, trueorigins.org is hardly a fount of solid logical argument.
1) I hold there is no meaningful definition of "semantic characteristics" in DNA
2) EVEN IF you choose some arbitrary definition that allows you to measure the "semantic characteristics" in a sample or instance of DNA, this has nothing to do with the possibility of evolution. There is NO PHYSICAL LAW that applies to biological organisms which constrains the amount of "information", however you define it, that offspring have in their DNA compared to their ancestors.
Your arguments are fluff, bogus nonsense, crap, pseudo-intellectual ravings, whatever.
That's why nobody bothers to address them.
If you read those evolution books, and still think ID has something meaningful to say, you just haven't understood evolution, or you just want to believe ID because evolution bothers you in some psychological way.
"[...] claims there are factual and logical errors but doesn't provide anything what so ever to back it up, even when being asked to do so several times".
Listen very carefully, shithead.
GO TO THE LIBRARY. FIND A BOOK WRITTEN BY SCIENTISTS (not a dumbed-down textbook) ON EVOLUTION. OPEN IT. READ IT. YOU WILL FIND THE ANSWER THERE.
That's my back up. You blabber about "loss of information" in evolution which is a MEANINGLESS PHRASE, as I have said before. It doesn't take much to refute something that has no meaningful content.
By the way, how do you think the bacteria GOT a gene for penicillinase, and why do you think a fungus got the ability to synthesize penicillin? Because some designer thought it would be fun to include the ability for a fungus to kill a bacterium, but then thought that he better design the bacterium so that the fungus mechanism doesn't work so good? Doesn't sound so intelligent to me.
Haeckel's embryo drawings are hardly some crucial link in evolutionary theory that, if disproven, take down the entire structure. Textbooks generally *suck*, not least because *any* treatment of evolution gets weeded out by conservatively biased state textbook committees, and textbook authors typically base their presentation on older, out-of-date textbooks, and only rarely on the latest research results.
The presence of misleading or inaccurate information in textbooks does not invalidate a scientific theory, it says our process for writing and selecting textbooks is crappy, and ought to do a better job of presenting the science accurately. That goal of accuracy has absolutely NOTHING to do with intelligent design, which is not a part of any scientifically-based theory.
At the very basic level, evolution by natural selection is *almost* at the level of a mathematical theorem; heritable variation causing variation in reproductive success (given the environment) + reproductive capacity much larger than the ability of the environment to sustain it = evolution of population toward increased "fitness" for the environment. Essentially, adaptation to a mode of existence can arise without the guidance of a designer.
It becomes more complicated when that evolution of a genetic population is applied to the question of speciation. Then, you have to consider various ways in which sub-groups of populations can achieve reproductive isolation from each other, allowing one species to differentiate into multiple distinct species. Then, you have to consider various "model assumptions", and, in specific cases, try to validate that the historical reality corresponded to one or another model.
There are tons of scholarly monographs (not high school or college textbooks) and journal articles which discuss all of these issues in a context of evolutionary theory, and succeed in doing so. *THAT* is the basis for believing evolution to be a theory supported by the facts.
You want evidence for evolution? Go to the fucking science library at a university, you dumbshit, instead of asking some guy named "OOGG_THE_CAVEMAN" to do the work for you.
Are you by any chance a 13 year old brat that have watched a few documentaries on TV and now think you've figured everything out?
Can't you even read my name?
I'm OOGG_THE_CAVEMAN. I'm a CAVEMAN.
I'm actually thousands of years old, and I've seen people complain about Darwin ever since he published Origin of Species. No matter how much scientists discover about the world that agrees with Darwin, some nutcases continue to make many of the same arguments that were made decades ago.
They didn't make sense then, and they make less sense now.
And, along the way, I've got a Ph.D. from an Ivy League university.
Given that the Earth is far from the largest body in the Solar System, if the materials exist here on Earth, they MUST exist elsewhere in the Solar System.
Except that there is no guarantee the larger bodies have the minerals in ores which ended up in places that are economical to extract.
In some sense, we're very lucky that the Earth is geologically active and interesting enough that heavy junk like gold and uranium ores and significant amounts of iron, etc., are actually near enough to the surface that we can mine them. It could very easily have turned out differently, and all the iron, uranium, etc., ended up in the core (where most of the Earth's iron is believed to be), leaving us on the surface pounding sand.
Jupiter is a freaking big planet, but if all the "good" stuff is way down in the core, well then, it doesn't help us very much.
Slashdot Mirror
I suppose you are right. I was confusing the Jepson claim style with the more usual kind of claim.
where a logical person could deduce a result of said existing technologies
In the U.S., the standard used is a "practitioner having ordinary skill in the art", (US Code Title 35, Section 103) not a "logical person".
To counter the other usual nonsense propagated by Slashdot headlines, claims may be dependent, in that some claims state what is conventional or known, and then a later claim describes a (novel and non-obvious) extension or improvement that constitues the grounds for novelty and non-obviousness. The common technique in Slashdot articles is to mention only the general claim, such as Claim 1 in this case, and crying "how obvious! That's been around since the stone age!" without mentioning the dependent claims 6, or 9, where the non-obvious part is stated.
The relevant date is not shipment but first silicon.
That's when you finally get the chance to run test cases at full speed instead of in an extremely slow software simulation.
It's not surprising that you will find additional bugs. The number of bugs found between first silicon and shipment is also a function of how large your team is and how developed your test cases are, not just the quality of your chip.
There isn't some giant cycle of design-build-test-change design; there's that kind of cycle (practically daily) in the software model for design and test, but once you get to tapeout and fab, that's pretty much a singular event. There isn't time to change the design and fab again. You just verify what you thought you knew from design simulations, run new tests that happen at full speed, document what new things you can find, decide if it is good enough, and ship; or, you wait for rev B to tape out and have first silicon. But that's probably multiple months, so you don't plan to do the Rev A-0 just for testing; you plan to ship it, and only a drastic failure will make you wait.
So there wasn't a Pentium III verification team? (rest of your bullshit clipped).
There almost certainly was; however, due to overlaps in development, test, and release schedules, Intel doesn't shift en masse from one chip to the next, but is forced to pipeline chip development; Pentium III test folks might have been approaching peak effort when the Pentium IV test team was just getting assembled. (I don't know the chronology).
The rest of your post implies that Intel doesn't know what the limits of verification technology are; I'm guessing that the company that basically invented the microprocessor knows a little bit about how to make sure microprocessor designs work; they were doing multi-month simulations on large scale clusters, and I'm guessing they could have saved a bunch of money if they could have done these calculations with less effort and the same level of accuracy. Maybe they are stupid, but hey, they only delivered chips that millions of people use everyday running things that range from Windows laptops to supercomputing clusters.
Maybe you work for AMD or Freescale or IBM, and can speak from experience, but from what I know of the folks who work at these places, they all have tremendous respect for the abilities of their competition, and hardly blow them off as incompetents.
The problem with *full* verification as opposed to random instruction testing is that the state space is far too large. Every instruction combined with every possible high-level register configuration combined with every possible contents of the relevant memory locations? Even on an 8-bit microprocessor like, say, the 6502, that's something like 2^21 * 2^40 * 2^8 = 6 * 10^20 combinations. At 1 GHz testing rate, that's about 6.8 million years. You have no chance of even enumerating these, much less testing them. Perhaps you can verify huge chunks of this by deduction, but that assumes your formal model is accurate and your proofs are correct.
As for modular design, in fact, in the very first page of the linked article from Intel, they talk about "cluster-level", i.e., modular, testing in the software model simulations that you claim they neglected. Perhaps you missed the whole section on "Cluster-Level Testing" talking about the six "clusters" into which the Pentium IV could be divided?
Furthermore, the architecture that allows these chips to execute the IA-32 instruction set so fast is a pipeline that overlaps simultaneous execution of multiple instructions, with a high degree of coupling, because these instructions might have interdependencies and much speculative computation is going on. Logically enumerating all the possibilities is a huge combinatorial explosion that defies analysis. Those are the kind of "side-effects" that you claimed shouldn't exist. Also keep in mind that the memory architecture is full of side-effects.
I don't know what kind of testing you do, but I'm pretty confident it has nothing to do with modern scale microprocessors, or you'd drop your posturing pretty quick.
I think your estimates are *way* off.
Silicon fab facilities are extremely expensive and capital intensive, but they produce shitloads of chips. The process scales; making 1000 wafers in these fabs is as easy as making one.
Engineering analysis of complex IC designs is a perfect example of combinatorial explosion. Each bit of state in the chip doubles the state space in which bugs can exist. Yes, *most* of that state is in the cache which has regularity in its structure, but that regularity didn't happen by accident: it was *designed* that way.
You can only test to a spec, and if the spec is imperfect and has gaps, you will leave space for bugs. Given that specs are written by engineers, they cannot be nearly complete for anything other than the most trivial circuits; the infrastructure used to suppor engineering of non-trivial circuits could itself have bugs.
The part of the spec that covers the cache is simple, and can conceivably be error-free and well-tested, and perhaps with methods that are amenable to mathematical proof. But that's not where the errors crop up. The errors crop up in the hugely complex mechanisms that handle all the pipelining, branch prediction, translation to microinstructions, handling of interrupts, etc., etc., that are not highly regular and modular and are not easy to spec, and are not easy to approach with formal methods.
their core "product" still weighs in at 1.3k, fits on a 640x480 monitor, and has a single significant input field.
I think this view of their "product" is totally naive.
Google makes squat from that blank page. They make lots of money from sticking tiny, unobtrusive, but still lucrative ads on all sorts of websites, including their own.
As long as they keep finding new ways to stick their ads all over the place on pages people want to view, and the ads stay lucrative, www.google.com itself could vanish, and GOOG would keep making money.
That stonewalling sure is mysterious, and cheers up those who want to stick it to the current Justice Department, but I think the real problem for Google is they have *shitloads* of data on what web sites folks visit, and this is possibly fundamental to their ability to offer relevant ads that are their revenue source.
Letting anybody with a subpoena go trawling through this ocean of data could really open Google up to having to support a lot of litigation, and probably opens up serious liability issues for Google itself, although IANAL, so I couldn't really say.
That hit on their stock price could be due to the market realizing "hey, Google isn't using magic powers, but is dependent on some huge, mysterious apparatus that might be outlawed or sued into oblivion." Sounds like a real issue to me.
Google has to deal with the Web as it is right now, not as it was in the ancient past of 1999; that's not to excuse them, but crap results could mean the Web has a lot more crap now. It's very hard for a computer to make these decisions, but I'm sure Google has lots of smart people trying to fight off the hordes of people who want to pillage the clicks of the billions of visits to Google.
In any case, I believe if you are logged into your Google account, you get a link "Remove Result" on your searches, which probably fees back in some way to improve future search results.
To clarify, now that I have a bit more time to elaborate, all information-theoretic calculations assume some underlying probability distribution for the elements of the information sequence.
In a chemical distribution, for example, if you are going to make some kind of information-content calculation, you are simply calculating a measure based on a probability weight for each molecule in the distribution.
There are, however, serious constraints on the ability of a chemical reaction to occur; each input reactant has a certain chemical energy, and the various output products have a certain chemical energy, and the good old conservation of energy demands that the input energy equal the output energy. There are additional thermodynamic constraints, because, for instance the concentration of products changes the "chemical potential" in a way to favor the reverse over the forward reaction, etc. That's why chemists introduce other quantities such as entropy (the chemical/thermodynamic/physical entropy, not simply information theory entropy) and enthalpy, to capture the effect of these macroscopic constraints on the microscopic probability that a reaction will happen.
Faced with this complexity (and, remember, a catalyst or enzyme does not cause reactions that are thermodynamically forbidden, just changes the *rate* of the reaction, because it offers a kinetic pathway that is more favorable; again, involving microscopic probabilities), what does your theory offer?
(e.g. log n each element is equally probable).
So this is the first problem: by wandering off into information theory, we have yet another way in which ID'ers can mess up the probabilities; classically, they came up with the fallacy of the "747 assembled by tornado", because they used an irrelevant uniform probability distribution for the various possible arrangements of, depending on their context, proteins in a flagellum, or whatever. Natural selection and inheritance, however, provides a probability distribution that depends on the parent, and strongly favors, of course, proteins resembling flagellar structures, but having some other function, if one exists. When you start with a 747 and a tornado sweeps through, you get a 747 with some stuff moved around. The distribution "747 parts lying around" is of very *low* prior probability, so it does not strongly enter into the probability of the outcome of a tornado.
Now, by adding another layer of mumbo-jumbo, you get to screw up the probabilities in a fresh-sounding way. Dembski does a lot of this, and this is the basis for devastating critiques of his CSI theory. The article you linked to had a false calculation of evolution probabilities, assuming multiple mutations occuring simultaneously, when they could (much more likely) occur sequentially.
Information theoretic entropy is just a measure of probability. Given the repeated failure of IDers to get probabilistic calculations right, this is just the same game by a different name.
Fine, you decide to take the perfectly fine physical term "entropy" and use it to describe some measure of a chemical distribution. Warning, you are in information theory territory, where it is exceedingly hard to connect back to the *physical* world, where biology happens.
There is still no connection between "specified complexity" and anything else relevant to evolution. In particular, it places no constraints on genetic sequences and the production of those through reproduction or mutation.
The problem is that accountants (and business types in general) as opposed to economic theorists, don't have the luxury of going out and selling something when they have to estimate its value. Yet, as conservative professionals, they must still come up with a "realistic" estimate; i.e. "as they really are" given that the sale is not going to happen as opposed to "as they would turn out if we sold this whole company tomorrow".
Theoretically, every accountant would come up with the same estimates, so when a transaction did happen, the buyer's and seller's teams of accountants would both come up with the same Platonic Value for the enterprise, and the negotiations would be very quick indeed. The fact that investment bankers get paid lots of money when these things happen means it isn't so simple.
Blargh.
This is all very tricky terminology. "Realistic" in your sense depends on whose perspective you adopt: the seller "how much can I sucker the buyer into paying", the buyer "how low can I go and still have this transaction happen" and the actual transaction that occured "where in the space between these two prices did the accident of one-time negotiations arrive". The first two depend at least partly on psychological factors. The third is the classic market view, where the equilibrium point leaves no surplus for either side in that most marginal transaction; but most huge acquisitions don't happen in a very liquid way that corresponds to a market. Then, once the transaction happens, the buyer turns into a potential seller, and his "realistic" price goes up, because he thinks he got something worth more than he paid for, unless he finds out he made a bad deal.
I would say "realistic" depends on an objective observer, with access to all the possible information, cooly evaluating all the factors, and coming up with a true value. But that's what you call "ideal."
As they say, in theory there is no difference between theory and reality; in reality, there is.
The whole notion of the "output distribution" of an enzyme determining something about the "information content" of the enzyme is just silly.
It means several things that make the "information content" concept useless to reason with
1) if I take away the reactants, there's less output; hence, less entropy change per enzyme molecule per time. Does it therefore have "less information"? If so, where is the information, in the reactants or in the enzyme?
2) similarly, if I lower the temperature or change the pH of the environment, to reduce the reaction rate, either of the catalyzed reaction, or some reaction that is providing the reactants, the reaction rate changes. Where's the information going now?
3) Entropy is an extensive quantity, proportional to the size of the system. If I simply increase the quantity of enzyme, am I increasing the "information content" of the enzyme? Or is the "information content" normalized per molecule? In which case a genetic change that increases the production of the enzyme would increase activity without increasing the "information content" of the enzyme.
4) If the enzyme keeps reacting for a long period of time, the entropy change increases over time (as does the energy flow that allows the reaction to happen). Does the "information content" of the enzyme depend on the lifetime of the cell?
A quantity you call "specified complexity" which depends
- on the environment (say, if the pub is open and you can drink alcohol, which an enzyme in your body must break down, or the pub is closed on Sundays, and the enzyme isn't causing a reaction)
- on the simple length of time the enzyme is working
- on the simple bulk amount of the enzyme (a 200 pound adult vs. a 10 pound infant have the same gene, but much different quantities of the enzyme)
doesn't seem to say much about the actual genetic content of the organism, which is the most compelling description of the "complexity" of an organism.
The only "realistic" price for anything is what it sells for.
Yes, in some theoretical sense, but not in the sense that the transaction is a rational one to make. Buyers can make mistakes. Anyway, once you own it, there's no guarantee that someone else would pay the same price, so this theoretical price has changed already; without a buyer ready to shake on the deal, you have to come up with a value through another analysis.
There is an intrinsic value in an enterprise, which is the present value of the stream of revenues that the enterprise will produce in the future. Nobody has a crystal ball to know the future stream of revenues in advance, nor do they even necessarily have a good value for the appropriate discount rate to apply. So the true value is shrouded in mystery.
Sigh.. There is no entropy change due to enzyme action when you include the enzyme itself, since the information that the product distribution has gained is of course already present in the enzyme! (Hence: "[...] the information content of the enzyme itself is at least the maximum information gained in transforming the substrate distribution into the product distribution".) The relevant entropy difference is between two such enzymes of which one has mutated.
If you think this is accurate, you are using a definition for entropy that is incompatible with the usual scientific one.
The whole point of an enzyme is that it acts as a catalyst; that is, after the reaction, the enzyme returns to the same state as it was before the reaction, and is able to catalyze the reaction for more reactants to produce more products.
If the enzyme returns to the same state, it has the same entropy, because the entropy of the enzyme is a function of the state.
I suppose you think this is bullshit, that I'm just saying because I hope to convince you; probably, if I claim 1+1=2, you think I am making an arrogant claim, I don't know. Maybe you want to use perfectly good scientific terms such as "entropy" in some strange way.
In any case the mutation does NOT happen in any particular copy of the enzyme. It happens in the genetic sequence that codes for the production of the enzyme, which is then created in a different form in the mutated organism.
But perhaps you are using some unusual meaning for the term "enzyme" and "mutation." Whatever. If you insist on these basic facts which are in contradiction to the basic definitions of the terms, you are hopeless, and it is no wonder you are persuaded by ID bullshit. Believe whatever you want, just don't claim you have scientific reasons for doing so.
The fuzziness, however, is that no one can really be sure that the stock or acquisition price is realistic. All you can count is the tangible assets (and, any goodwill left over from past acquisitions by the company, if you believe it is still accurate). If you grossly overpay for a company at the height of some stock market frenzy, you could be giving up cold, hard cash for speculative froth by the name of "goodwill".
Furthermore, you can't practically acquire Google for its quoted market cap. That market cap is based on the latest market transaction of a few thousand shares, not on what the price would be to buy 295 million shares in a single transaction.
Once again you think that saying something makes it so. Your ego must be the size of a small country.
No, my saying something doesn't make it so.
REALITY makes it so, I'm just reporting it. If you don't comprehend REALITY, that is your problem not mine. If you can't realize meaningless bullshit, that's your loss.
I went back and double-checked the "information content of proteins" section of that long screed, and it was, as far as I could tell, either BS, or trivial thermodynamics, depending on how those symbols are supposed to be interpreted.
The entropy change due to enzyme action is simple physical chemistry.
The only possible physical argument would be that the physical processes of life are forbidden by thermodynamics, but, since I'm living and breathing, evidently enzymes can and do work.
THERE IS NO ARGUMENT that leads from there to saying anything about evolution, and in fact the screed makes none.
It just denotes that information is measured as specified complexity instead of the more commonly (at least in comp.sci.) used definition of complexity by improbability (by Claude Shannon).
Whatever. You give some link to a anti-evolution web site to define your "specified complexity" quantity. That's your arbitrary definition. The arguments made on the basis of these numbers are bullshit, irrelevant, or both. The whole screed you link to is a bunch of strawmen.
1) No one claims that evolution only happens by transcription errors in the genome. So all the astronomical probabilities and timescales quoted are irrelevant.
2) The "requirements" stated for evolution, requiring a single step-by-step "mutate then select" is false. There's plenty of genetic drift going on without selection that can cause variation without selective pressure. There are also very fast ways, involving mutations in particular regulatory genes that can cause huge changes with very small mutations.
3) The math behind the odds of modification assume statistical independence between generations and between mutations. This is obviously wrong; once a change happens, it doesn't revert back immediately; it's "sticky", and persists in the descendants of the mutated form with HIGH probability. Plus, it ignores the effect of selection entirely!
I could go on and on. But, rest assured, trueorigins.org is hardly a fount of solid logical argument.
information in DNA has semantic characteristics
OK, the answer to this is simply "so what."
Let me break this down for you
1) I hold there is no meaningful definition of "semantic characteristics" in DNA
2) EVEN IF you choose some arbitrary definition that allows you to measure the "semantic characteristics" in a sample or instance of DNA, this has nothing to do with the possibility of evolution. There is NO PHYSICAL LAW that applies to biological organisms which constrains the amount of "information", however you define it, that offspring have in their DNA compared to their ancestors.
Your arguments are fluff, bogus nonsense, crap, pseudo-intellectual ravings, whatever.
That's why nobody bothers to address them.
If you read those evolution books, and still think ID has something meaningful to say, you just haven't understood evolution, or you just want to believe ID because evolution bothers you in some psychological way.
"[...] claims there are factual and logical errors but doesn't provide anything what so ever to back it up, even when being asked to do so several times".
Listen very carefully, shithead.
GO TO THE LIBRARY. FIND A BOOK WRITTEN BY SCIENTISTS (not a dumbed-down textbook) ON EVOLUTION. OPEN IT. READ IT. YOU WILL FIND THE ANSWER THERE.
That's my back up. You blabber about "loss of information" in evolution which is a MEANINGLESS PHRASE, as I have said before. It doesn't take much to refute something that has no meaningful content.
By the way, how do you think the bacteria GOT a gene for penicillinase, and why do you think a fungus got the ability to synthesize penicillin? Because some designer thought it would be fun to include the ability for a fungus to kill a bacterium, but then thought that he better design the bacterium so that the fungus mechanism doesn't work so good? Doesn't sound so intelligent to me.
Haeckel's embryo drawings are hardly some crucial link in evolutionary theory that, if disproven, take down the entire structure. Textbooks generally *suck*, not least because *any* treatment of evolution gets weeded out by conservatively biased state textbook committees, and textbook authors typically base their presentation on older, out-of-date textbooks, and only rarely on the latest research results.
The presence of misleading or inaccurate information in textbooks does not invalidate a scientific theory, it says our process for writing and selecting textbooks is crappy, and ought to do a better job of presenting the science accurately. That goal of accuracy has absolutely NOTHING to do with intelligent design, which is not a part of any scientifically-based theory.
At the very basic level, evolution by natural selection is *almost* at the level of a mathematical theorem; heritable variation causing variation in reproductive success (given the environment) + reproductive capacity much larger than the ability of the environment to sustain it = evolution of population toward increased "fitness" for the environment. Essentially, adaptation to a mode of existence can arise without the guidance of a designer.
It becomes more complicated when that evolution of a genetic population is applied to the question of speciation. Then, you have to consider various ways in which sub-groups of populations can achieve reproductive isolation from each other, allowing one species to differentiate into multiple distinct species. Then, you have to consider various "model assumptions", and, in specific cases, try to validate that the historical reality corresponded to one or another model.
There are tons of scholarly monographs (not high school or college textbooks) and journal articles which discuss all of these issues in a context of evolutionary theory, and succeed in doing so. *THAT* is the basis for believing evolution to be a theory supported by the facts.
You want evidence for evolution? Go to the fucking science library at a university, you dumbshit, instead of asking some guy named "OOGG_THE_CAVEMAN" to do the work for you.
Are you by any chance a 13 year old brat that have watched a few documentaries on TV and now think you've figured everything out?
Can't you even read my name?
I'm OOGG_THE_CAVEMAN. I'm a CAVEMAN.
I'm actually thousands of years old, and I've seen people complain about Darwin ever since he published Origin of Species. No matter how much scientists discover about the world that agrees with Darwin, some nutcases continue to make many of the same arguments that were made decades ago.
They didn't make sense then, and they make less sense now.
And, along the way, I've got a Ph.D. from an Ivy League university.
It is as pointless to point out missing or incorrect facts to someone who is claiming "1+1=3, show me the facts that prove otherwise!!!!"
Naturalists have been collecting facts for over a hundred years, all of which make sense in the context of evolution.
Given that the Earth is far from the largest body in the Solar System, if the materials exist here on Earth, they MUST exist elsewhere in the Solar System.
Except that there is no guarantee the larger bodies have the minerals in ores which ended up in places that are economical to extract.
In some sense, we're very lucky that the Earth is geologically active and interesting enough that heavy junk like gold and uranium ores and significant amounts of iron, etc., are actually near enough to the surface that we can mine them. It could very easily have turned out differently, and all the iron, uranium, etc., ended up in the core (where most of the Earth's iron is believed to be), leaving us on the surface pounding sand.
Jupiter is a freaking big planet, but if all the "good" stuff is way down in the core, well then, it doesn't help us very much.
You are simply ignorant of the facts.