Either way, the concept is flawed - if it's to prevent an action by a 35 YOWM, the action of checking an ID does nothing. If the intent is to prevent an action by a 35 YOWM, then police agents must be allowed to search, arbitrarily, all property owned by white males, and all property that white males have access to. And that's for starters. And hey, if you have nothing to hide...
Websites can post whatever TOS they want; it doesn't mean the USER visiting that site has the authority to consent to that TOS. Otherwise, some clod who merely turns french fries for a living could grant me full, unfettered access to the BurgerKing network, and there's isn't a damned thing anyone could do about it. After all, the french-fry guy said I could.
Since you went through the trouble of preventing *them* from gaining access to *your* page rendering software... can what they've done be construed as the circumvention of an access control device?
After all, they have deliberately initiated a packet stream that, in every sense of the word, has caused an unwanted and unauthorized impact on the state of your machine, and that stream was specifically crafted to defeat any measures you may have taken to prevent it.
I'd want to know how it behaves over the long term. Straw is bad enough when it decomposes, and will self-ignite. These bales theoretically mitigate that with additives, which should be fine in the short-term at least. Long term is an issue I'd be concerned about, however, and I'd want a real answer on that. Roofs leak, walls sweat, and sooner or later your toilet will overflow.
Second, I'd want to know how this stuff behaves in a fire. Sooner or later, your house will have one - and remember, "fire ratings" don't mean dick in real life... a "one hour" rating usually means 15 minutes if you are lucky, and that rating is exactly for the item in question. It means little for anything behind it that'll be exposed via conductive heat transfer.
Nah, look at the sample product lines - large segement displays. Sure, "some day" it'll be great for pr0n and what-not. Today, their immediate market seems to be call center stat boards (might be updated every 30 seconds or so), airports/train stations (updated once per 15 mins), and quasi-static signage for advertising.
These are all things that change relatively infrequently (compared to a 20 or 70 Hz clock), so power not spent on refresh is significant. Or so goes the sales pitch, at least...
Yes, would *prefer*. However, you'll have to ignore the short (and probably the mid) term cost/benefits - interoperability is going to require the (at least) partial implementation of an irrelevent feature set. This goes, in short or medium terms, against the needs of the market as that market becomes more vertical.
I'm admin at a *very* niche company, and our software needs are quite complex. In the US, there are exactly 56 companies that do what we do, typically one per state (sometimes two or three per state, and in a couple cases, 3 to 5 states per company). 56 customers, how many software vendors will that market support? Currently, the answer is five. The sofware we use is extremely customized to suit the various state and regional laws, and "interoperability" with other vendors, or even other states who share our vendor, would be a bloated nightmare.
It may seem like an irrelevent example, but it is one boundary of the general case. Interoperability is nice, but as the market becomes more vertical, interop becomes less relevent and a helluvalot more expensive. For bullshit cookie-cutter apps? Sure. But not for real stuff, it isn't worth the development costs, and it sure as hell isn't worth the lifecycle costs. I'll never need (I hope) the firmware in my car to "interoperate" with the firmware in my dishwasher, nor do I want to need to maintain that interoperability.
Then again, for bullshit apps, "prefer" is generally correct. The more bullshit the application, interop becomes less of a perference, and more of a demand - after all, the software becomes more of an expendable/replaceable commodity as the bullshit factor increases.
The Test and Measurement aspect is what I remember about HP. It carries a strong scent of credibility with it.
The "new" HP is totally divorced from that vision, to the point that I do not relate the two - instead of being focused on competent technique, the focus is "A toll on every highway, a troll on every bridge".
They are little more than patent squatters at this point, and offer no value to the market.
There is no logic in demanding things that may not be true.
To the contrary - consider the following.
x=0
if x = 0, repeat this statement.
This should loop forever, it's a nice closed system. Very predictable. There is no logic in demanding that it someday might not... so adding anything (after it) makes no sense.
Wrong. It's quite possible for x to deviate - the dreaded cat hair on the mainboard, for example. Or, Intel has another "accident". An unpredictable quirk that exists far outside the scope of the assumed model. If the above statement is mission critical, I damned well better handle the case of the loop unexpectedly ending - it means reality, as we've assumed it in our model, has a problem - namely that our state machine has lost it's marbles. Your method does not allow for coding in this regard. "There is no logic in demanding things that may not be true." Wrong, wrong, wrong... I'd suggest you download an ISO called "MemTest" to see just how wrong your assertion is. Memtest is irrelevent within the scope of the machine; inside that scope, we demand that state follows causality. Outside that scope, shit happens, and chips burn. Even VisualBasic doesn't have a "If RAM IS FUXED" statement - such a condition is not appropriate for that model. (Instead, that condition is typically handled via NMI vectored to firmware, which consequently halts the machine.)
As to the "designing software" comment - in no way does being a physicist qualify someone to develop abstract models. There is a great deal of abstraction in this endeavor, and the door of "qualification" swings both ways.
Re: obfuscation -
That's a complete obfuscation. No, it's about as up front as you can get.
First off you're introducing a red herring by suggesting that people will die if a physicist makes a mistake. No, I'm not suggesting anything. I'm stating a situation where you have ONE try, and successive approximation is NOT APPROPRIATE. Tell ya what - we'll remove the red herring. Instead, if it pukes, we lose one gazillion dollars. That should change everything, right?
Secondly, observations are what science is based on, everyone stands on the shoulders of giants. In other words, trial and error (successive approximation) is always the most appropriate technique. No, it is not. You've clearly never bothered with high speed password attacks.
Thirdly, the analogy seems to be that the server failing equates to an innacurate prediction of physics due to the fact that it didn't take into account how many dimentions (power supplies) are needed. Pretty much, yes. And therein lies the flaw in your method - yours will produce a failure. "Eventualities" will produce a cost/benefit which will yield at least two supplies. One will fail, the other(s) will hold long enough to replace the dead one(s). And meanwhile, our new server will hum along come hell or high water, while yours is flat on its face. Sure, in the long run, extra supplies that aren't used end up being wasted. Care to guess, in both our analogy and in reality, what the impact of that extra supply is on our solution?
Nothing.
The solution with the irrelevent factors will simply wash those factors out as the general case is evolved. The model will work with the extra factors, because it's designed to. I think the thing you're overlooking is that this "fat" model still has to predict the desired behavior, even with the extra garbage. That garbage (and it is garbage, you're correct) will simply come out in the wash, as opposed to your technique of not adding it in the first place, which always comes up short. You say x=5, and that's it. I say x=(5*1*1*1*1)+0+0+0+0. I don't know what those extra terms might be some day, but they just might prove to be useful - and they certainly appear to exist, because the result is correct. And meanwhile, my model produces the exact same result as yours - 5 - while being a more general case, which yours
Interesting (and thanks for keeping this topic alive, it's good to talk to you:) )
The topic isn't physics; the topic is about creating a model of an abstract process, which (hopefully) has a real instance to compare against. Physicists play a part, but it is far from their normal domain - it's all toolsets, behaviors, and devined implications upon devined implications that'll never be experiementally shown. In other words, it's damned near a software project. Physicists are still a huge part of it, but by and large their "normal" skill sets are not appropriate to the techniques that are needed. Half of them still think that inductance and impededance are two different things, for chrissakes. This is what I mean by "abstraction" - or the ability to do it, for that matter. The physicist method allows for it, but does not generally condone it. According the the razor, the results are too messy and exceed the scope.
Back to induction -
Step 1: Quark-head says "4 dimensions ain't enough. We're wrong, and need 5 for this model to work." Step 2: Quark-head says "5 dimensions ain't enough. We're wrong, and need 6 for this model to work." Step 3: Quark-head says "6 dimensions ain't enough. We're wrong, and need 7 for this model to work." Step 4: Quark-head says "7 dimensions ain't enough. We're wrong, and need 8 for this model to work."...
Step 22: Quark-head says "25 dimensions ain't enough. We're wrong, and need 26 for this model to work." Step 23: Quark-head says "26 dimensions ain't enough. We're wrong, and need 27 for this model to work."
Again, consider the diference in methods - the method you propose is one of successive approximation, using "observation" as the measure of error for the next iteration. There is a whole pile of caveats that go along with the SA method which you're quite familiar with, but probably haven't connected as being relevent. It is quite relevent... but we never associate the classic "scientific method" with being SA, so we don't consider it to have any caveats. When you abstract the method you propose, and then abstract a simple SA - the resulting processes are identical. This is key for two reasons - (a) the process is not appropriate to some applications, and (b) most people have never connected the dots and realized this. Scary, because it's blatently obvious after it's mentioned.
On the good side, I have *finally* remembered that one freakin word I've been needing. Talk about brain dead, lol.
From the view point of *eventualities* - that's the word that physicists are required to disavow, because it precludes the concept of SA - we know that our view is always incomplete, and intend for it to be. The implication that you seek will not be present in any single step of the above 25 steps - but as you scroll through that anecdotle babble, the implication emerges from the whole. You claim we must show the pattern must continue; I claim you must show that the pattern will stop. Wrong 27 times in a row, you tell me where you'd place your money. Scientific Method says that they're correct relative to the scope of their observations. Eventualities demand that once a loser, always a loser... they've repeatedly shown that they cannot determine what the scope of observation needs to be, and *that* is the implication of the next step. The cycle will continue, and the induction holds. It's not about what the quark-head observes. It's about what *we* observe in the observer, and it's recursive. It's also exactly valid in a great many models, and (in those applications) will produce models that better match observation than any SA method will.
Consider the following, which a "physicist's method" will fail miserably at. This is not because anyone is dumb, or incompetent - it's merely because your methods are not universal as you claim.
I'm a physicist. I've got 190 machines in my shop. In the many years I've had them, I've *never* had a power supply fail. Ever.
Absolutely - but you make one error. Well, "error" should be in quotes, because it's *not* an error. But it is.
You're viewing from an engineer perspective... that which isn't explicitly allowed is absolutely forbidden. I'm viewing from a software perspective - that which isn't explicitly forbidden is absolutely demanded. The differences are subtle, but the impacts are severe... I'm allowed to handle a recursive abstraction. You must not (and god help us all if you did)... and in fact, you must disclaim that such a concept cannot exist.
Your model of a theory assumes that it is complete until shown otherwise, and the Razor has a high value in making the theory manageable. "It goes against scientific principles" - no, it goes against physics-specfic scientific *method*... the introduction of seemingly spurious parameters only complicates a model, making it unmanageable. So, the razor is good, and the razor is god in that arena.
Apply that method as a software theorist, however, and you're deader than a racecar with a.com sponsorship. My models demand I treat them as incomplete, even if I cannot show it - if I solve a general case, great! But, not finding an exception to that general case does not *mean* it is *the* general case - it merely means that I might be too stupid to find that one needed exception, TODAY. You (rightly) suggest that something being "counterproductive" is sufficient reason to discard it (aka the Razor). I (rightly) assume that something that appears counter-productive will generally yield a more general (and correct) case. In fact, I demand the usage of unneeded complexity, just in case something unexpected happens. And I know it will happen, but I cannot predict what, how, or where. Buffer overflows, as a niaeve example. Unhandled exceptions as another. Your model routinely produces them, and yet cannot entertain that they even exist until *after* the box blows up. Mine assumes they will happen, always, and will often count on it. You're free to pretend that a given state will always exist at a certain line - but you're a fool if you believe that to be true. All it takes is one cat hair getting sucked into your machine and landing on your RAM, and you'll be eating crow. Cat hair does not exist within your "fundamental assumptions" - you'll not find it in any book by Booch, and until today you've probably never conceived of it as a relevent problem, and probably still discount it as irrelevent even now. That's the difference - and you are completely wrong. You cannot prove it'll ever happen, so you respect the razor's demand for removal. I cannot prove it will NOT happen, so I must demand that it will.
Physicists love to abstract a behavior... but they don't routinely abstract the abstractions, and study the abstractions of those abstractions for commonalities. And quite frankly, they should not... as you say, the "barking up the wrong tree" factor for them is potentially huge. Their work is based on the premise that knowlege (and possible behavior) is finite, ergo there are a finite number of predictions that must be made. That's fine, it makes physics viable as a research field - but that doesn't make it true, it only makes it convenient for the sake of the physicists. You perhaps should reconsider your definition of "scientific method", because it is not complete.
In other words, you've done nothing to demonstrate that there *aren't* a large number of dimensions - you said yourself, there's no evidence. The Razor demands their removal; but the software theorist views that as sheer arrogance - after all, we have plenty of GOOD evidence to predict their existence. There are at least two dimensions to space. Oh, hey... make that at least three. I'm seeing a pattern here, Errr.... ok, four. Induction, anyone? No reason to stop - not at 20, not at 30, not at a zillion. Especially not at 20 or 30 like the current batch of quark-heads use... dude, you've already got *30* freakin dimensions in your space, and you added them one at
Photon paradox: Double-slit experiment. Wave or particle, depending on which you choose to observe. The paradox exists due to the model by which you perceive it - if you use another model, it isn't a paradox at all, right? That's the idea I'm trying to get across, at least, that a paradox is only relative to the perception model and merely indicates that the model is incomplete / not a general case.
For the causality thing - the easiest way to envision the game is to consider a simple iterative function, and a surface who's coords can be mapped as inputs. The output then gets used as the next input, etc - classic chaos. There's a problem that we need to avoid, however - if at any time a full set of coords is repeated, you'll end up with a loop and it won't ever break. THAT is how I envision the dreaded "breaking causality" you mentioned... if all inputs are duplicated, you've effectively violated Pauli's exclusion in a philosphical sense and "the universe" needs a ctrl-alt-del. Bad, bad, bad.
So, to get around that - given, say, 4 parameters - if any trio of coords have already been used together, it isn't a problem if the 4th *didn't* happen with them. Simplistically, "no two objects in the same space." I put a pen on my desk - remove it, then put my keyboard in that same spot. Relative to a 3-space model, I've violated that rule. Relative to a 4-space model, I haven't - the first trio is "already used", but the 4th wasn't (with that trio, at least.) Sadly, I can't describe it much better without being incredibly long-winded.:)
Now for my challenge about breaking causality - given an input space of (n) - and we only directly perceive 3 vectors of it - we know that we cannot (trivially) differentiate the remaining (n-3) vectors from the single sum of those vectors - we'll perceive them all as a single vector, called "time". (n) might indeed be 4, or it might also be a gazillion, we cannot easily tell - from our perspective, we'll chalk them all up as a single net-effect. (I've come up with a couple of great yet simple games to demonstrate this to lay people, if you're interested. It's an easy and fun way to generate a forehead slapping "Aha!" from a victim. And when you're done, they'll never view a sheet of paper or driving a car the same way again, lol.) So... given the case that (n) might be huge (induction sort of demands that (n) might be infinite, if you put any salt into that sort of thing), and our models are at best playing with 20 or 30 of them - we're artifically capping (n) at 20 or 30, when (n) is in fact greater. In that case, having a duplicate "state" is not an issue - it's only duplicated in respect to the (20 or 30) space we're observing from, not the true (n) space. So long as (n) contains one axis more than we're aware of (and again, there's at least 4 that we're aware of... so induction says... induction says the theorists are potentially screwed trying to model it, Argh.) then causality will not be broken, because there are inputs (states) that are still arbitrary.
Now consider all of that drivel with this game - and yes, this game is an abomination against everything... but it's my game, so we'll play it just to see what happens. We're going to play the "Positional Exclusion" game. We'll use a two-axis surface that is the universe, we'll put an object on it called P. And we'll have two rules - Our rule: we view the state of the game as a sequence of instantaneous observations. THE rule we're interested in: P can never exist in the same place (state) twice in a row.
So, we take our initial look, and P is somewhere. Look again, and P moves. How far? It moves the minimum distance demanded to not violate the rule - it moves the distance of its radius, let's say. Never more because it's a lazy P, and never less because it's forbidden. Look again, and it moves again... the distance of it's radius. A simple game, totally wrong and doesn't show much - in fact, we can't even explain "why" P moves. But watch what we can do with it.
Yeah, but a paradox is usually relative to a certain perspective. Hitting one indicates nothing. Photons are a paradox, and you're suggesting that they're "not allowed" in science.
Switch to a more macro perspective, and there is no paradox. Much like being local to a recursive function - from within it, there can be a paradox with causality. From outside, however, the causality of that function is obvious. In this case, what we perceive as "breaking causality" might in fact appear linear from a larger, more "macro" view... e.g. more than just niaeve 4-space, or worse 3-space.
Regardless of what n-space you choose, however, causality only demands ONE axis to maintain unique state. The remaining (n-1) can go "back in time" all day long, it really doesn't matter - and therein lies my point (argh! bad pun, sorry.) If our "perspective" is constrained to within the (n-1) space, then our perspective is incomplete - and causality *will* appear broken. Causality isn't, however, when perceived from the full (n) view.
And since I'm pretty damned sure we haven't actually solved for (n) yet, I'm pretty sure we can appear to break causality all day long relative to our current niaeve ruleset. And no, approximations of (n) that are convenient for a specific model do not make a solution for (n). When (n) is resolved, the resulting space would appear as being static / e.g. no such thing as "time", relative to *anything*. Time, after all, may well be the perception of the sum of the remaining n's that we can't directly observe.
Fear not the paradox:) It is simply a case where both results are true, but the question is incomplete.
Dumb question, but is that related to the alg used in mrSID files? I'm grabbing 800~900K.sid files, and converted as tifs yields about 25meg per tile. Converting to.jpg, very lossy, yields about 1.5 megs per tile on average. Converting to jpeg with very low loss yields about 5 megs per tile.
I only ask because you mentioned sat imagery... I'm recently quite impressed by.sids, and equally as curious.
Slashdot Mirror
Either way, the concept is flawed - if it's to prevent an action by a 35 YOWM, the action of checking an ID does nothing. If the intent is to prevent an action by a 35 YOWM, then police agents must be allowed to search, arbitrarily, all property owned by white males, and all property that white males have access to. And that's for starters. And hey, if you have nothing to hide...
Actually, claim that you cannot remove one of your shoes for medical reasons. When they push, remind them about HIPPA.
You answered your own question -
If I have nothing to hide, then the government has no need to know. Period.
Bzzzzzt WRONG!
HBO, Skinemax, et al, ALL use airwaves.
You think they run a point-to-point line from HBO to your local catv affiliate?
No, they use... satellites. Any questions, drive past your local provider's main office, and look at all them big dishes in the back.
Hey, wow... I just got a spam from some zombie on comcast! It contains "forward looking statements", and says SCO is set to become great buy!
I think that makes it official...
Websites can post whatever TOS they want; it doesn't mean the USER visiting that site has the authority to consent to that TOS. Otherwise, some clod who merely turns french fries for a living could grant me full, unfettered access to the BurgerKing network, and there's isn't a damned thing anyone could do about it. After all, the french-fry guy said I could.
Hmm.
Since you went through the trouble of preventing *them* from gaining access to *your* page rendering software... can what they've done be construed as the circumvention of an access control device?
After all, they have deliberately initiated a packet stream that, in every sense of the word, has caused an unwanted and unauthorized impact on the state of your machine, and that stream was specifically crafted to defeat any measures you may have taken to prevent it.
And there's the other obvious thing -
I'd want to know how it behaves over the long term. Straw is bad enough when it decomposes, and will self-ignite. These bales theoretically mitigate that with additives, which should be fine in the short-term at least. Long term is an issue I'd be concerned about, however, and I'd want a real answer on that. Roofs leak, walls sweat, and sooner or later your toilet will overflow.
Second, I'd want to know how this stuff behaves in a fire. Sooner or later, your house will have one - and remember, "fire ratings" don't mean dick in real life... a "one hour" rating usually means 15 minutes if you are lucky, and that rating is exactly for the item in question. It means little for anything behind it that'll be exposed via conductive heat transfer.
Nah, look at the sample product lines - large segement displays. Sure, "some day" it'll be great for pr0n and what-not. Today, their immediate market seems to be call center stat boards (might be updated every 30 seconds or so), airports/train stations (updated once per 15 mins), and quasi-static signage for advertising.
These are all things that change relatively infrequently (compared to a 20 or 70 Hz clock), so power not spent on refresh is significant. Or so goes the sales pitch, at least...
Yes, would *prefer*. However, you'll have to ignore the short (and probably the mid) term cost/benefits - interoperability is going to require the (at least) partial implementation of an irrelevent feature set. This goes, in short or medium terms, against the needs of the market as that market becomes more vertical.
I'm admin at a *very* niche company, and our software needs are quite complex. In the US, there are exactly 56 companies that do what we do, typically one per state (sometimes two or three per state, and in a couple cases, 3 to 5 states per company). 56 customers, how many software vendors will that market support? Currently, the answer is five. The sofware we use is extremely customized to suit the various state and regional laws, and "interoperability" with other vendors, or even other states who share our vendor, would be a bloated nightmare.
It may seem like an irrelevent example, but it is one boundary of the general case. Interoperability is nice, but as the market becomes more vertical, interop becomes less relevent and a helluvalot more expensive. For bullshit cookie-cutter apps? Sure. But not for real stuff, it isn't worth the development costs, and it sure as hell isn't worth the lifecycle costs. I'll never need (I hope) the firmware in my car to "interoperate" with the firmware in my dishwasher, nor do I want to need to maintain that interoperability.
Then again, for bullshit apps, "prefer" is generally correct. The more bullshit the application, interop becomes less of a perference, and more of a demand - after all, the software becomes more of an expendable/replaceable commodity as the bullshit factor increases.
Have fun,
SBB
Yeah, well, in Soviet Russia... well, imagine an entire Beowulf cluster of them, you insensitive clod!
My impression?
The Test and Measurement aspect is what I remember about HP. It carries a strong scent of credibility with it.
The "new" HP is totally divorced from that vision, to the point that I do not relate the two - instead of being focused on competent technique, the focus is "A toll on every highway, a troll on every bridge".
They are little more than patent squatters at this point, and offer no value to the market.
Well, I was going to say that Step 2 should be "pay outstanding debts", but with the bankruptcy, you're correct- They don't need step 2.
There is no logic in demanding things that may not be true.
To the contrary - consider the following.
x=0
if x = 0, repeat this statement.
This should loop forever, it's a nice closed system. Very predictable. There is no logic in demanding that it someday might not... so adding anything (after it) makes no sense.
Wrong. It's quite possible for x to deviate - the dreaded cat hair on the mainboard, for example. Or, Intel has another "accident". An unpredictable quirk that exists far outside the scope of the assumed model. If the above statement is mission critical, I damned well better handle the case of the loop unexpectedly ending - it means reality, as we've assumed it in our model, has a problem - namely that our state machine has lost it's marbles. Your method does not allow for coding in this regard. "There is no logic in demanding things that may not be true." Wrong, wrong, wrong... I'd suggest you download an ISO called "MemTest" to see just how wrong your assertion is. Memtest is irrelevent within the scope of the machine; inside that scope, we demand that state follows causality. Outside that scope, shit happens, and chips burn. Even VisualBasic doesn't have a "If RAM IS FUXED" statement - such a condition is not appropriate for that model. (Instead, that condition is typically handled via NMI vectored to firmware, which consequently halts the machine.)
As to the "designing software" comment - in no way does being a physicist qualify someone to develop abstract models. There is a great deal of abstraction in this endeavor, and the door of "qualification" swings both ways.
Re: obfuscation -
That's a complete obfuscation. No, it's about as up front as you can get.
First off you're introducing a red herring by suggesting that people will die if a physicist makes a mistake. No, I'm not suggesting anything. I'm stating a situation where you have ONE try, and successive approximation is NOT APPROPRIATE. Tell ya what - we'll remove the red herring. Instead, if it pukes, we lose one gazillion dollars. That should change everything, right?
Secondly, observations are what science is based on, everyone stands on the shoulders of giants. In other words, trial and error (successive approximation) is always the most appropriate technique. No, it is not. You've clearly never bothered with high speed password attacks.
Thirdly, the analogy seems to be that the server failing equates to an innacurate prediction of physics due to the fact that it didn't take into account how many dimentions (power supplies) are needed. Pretty much, yes. And therein lies the flaw in your method - yours will produce a failure. "Eventualities" will produce a cost/benefit which will yield at least two supplies. One will fail, the other(s) will hold long enough to replace the dead one(s). And meanwhile, our new server will hum along come hell or high water, while yours is flat on its face. Sure, in the long run, extra supplies that aren't used end up being wasted. Care to guess, in both our analogy and in reality, what the impact of that extra supply is on our solution?
Nothing.
The solution with the irrelevent factors will simply wash those factors out as the general case is evolved. The model will work with the extra factors, because it's designed to. I think the thing you're overlooking is that this "fat" model still has to predict the desired behavior, even with the extra garbage. That garbage (and it is garbage, you're correct) will simply come out in the wash, as opposed to your technique of not adding it in the first place, which always comes up short. You say x=5, and that's it. I say x=(5*1*1*1*1)+0+0+0+0. I don't know what those extra terms might be some day, but they just might prove to be useful - and they certainly appear to exist, because the result is correct. And meanwhile, my model produces the exact same result as yours - 5 - while being a more general case, which yours
Interesting (and thanks for keeping this topic alive, it's good to talk to you :) )
...
The topic isn't physics; the topic is about creating a model of an abstract process, which (hopefully) has a real instance to compare against. Physicists play a part, but it is far from their normal domain - it's all toolsets, behaviors, and devined implications upon devined implications that'll never be experiementally shown. In other words, it's damned near a software project. Physicists are still a huge part of it, but by and large their "normal" skill sets are not appropriate to the techniques that are needed. Half of them still think that inductance and impededance are two different things, for chrissakes. This is what I mean by "abstraction" - or the ability to do it, for that matter. The physicist method allows for it, but does not generally condone it. According the the razor, the results are too messy and exceed the scope.
Back to induction -
Step 1: Quark-head says "4 dimensions ain't enough. We're wrong, and need 5 for this model to work."
Step 2: Quark-head says "5 dimensions ain't enough. We're wrong, and need 6 for this model to work."
Step 3: Quark-head says "6 dimensions ain't enough. We're wrong, and need 7 for this model to work."
Step 4: Quark-head says "7 dimensions ain't enough. We're wrong, and need 8 for this model to work."
Step 22: Quark-head says "25 dimensions ain't enough. We're wrong, and need 26 for this model to work."
Step 23: Quark-head says "26 dimensions ain't enough. We're wrong, and need 27 for this model to work."
Again, consider the diference in methods - the method you propose is one of successive approximation, using "observation" as the measure of error for the next iteration. There is a whole pile of caveats that go along with the SA method which you're quite familiar with, but probably haven't connected as being relevent. It is quite relevent... but we never associate the classic "scientific method" with being SA, so we don't consider it to have any caveats. When you abstract the method you propose, and then abstract a simple SA - the resulting processes are identical. This is key for two reasons - (a) the process is not appropriate to some applications, and (b) most people have never connected the dots and realized this. Scary, because it's blatently obvious after it's mentioned.
On the good side, I have *finally* remembered that one freakin word I've been needing. Talk about brain dead, lol.
From the view point of *eventualities* - that's the word that physicists are required to disavow, because it precludes the concept of SA - we know that our view is always incomplete, and intend for it to be. The implication that you seek will not be present in any single step of the above 25 steps - but as you scroll through that anecdotle babble, the implication emerges from the whole. You claim we must show the pattern must continue; I claim you must show that the pattern will stop. Wrong 27 times in a row, you tell me where you'd place your money. Scientific Method says that they're correct relative to the scope of their observations. Eventualities demand that once a loser, always a loser... they've repeatedly shown that they cannot determine what the scope of observation needs to be, and *that* is the implication of the next step. The cycle will continue, and the induction holds. It's not about what the quark-head observes. It's about what *we* observe in the observer, and it's recursive. It's also exactly valid in a great many models, and (in those applications) will produce models that better match observation than any SA method will.
Consider the following, which a "physicist's method" will fail miserably at. This is not because anyone is dumb, or incompetent - it's merely because your methods are not universal as you claim.
I'm a physicist. I've got 190 machines in my shop. In the many years I've had them, I've *never* had a power supply fail. Ever.
Next week, I'm b
Absolutely - but you make one error. Well, "error" should be in quotes, because it's *not* an error. But it is.
.com sponsorship. My models demand I treat them as incomplete, even if I cannot show it - if I solve a general case, great! But, not finding an exception to that general case does not *mean* it is *the* general case - it merely means that I might be too stupid to find that one needed exception, TODAY. You (rightly) suggest that something being "counterproductive" is sufficient reason to discard it (aka the Razor). I (rightly) assume that something that appears counter-productive will generally yield a more general (and correct) case. In fact, I demand the usage of unneeded complexity, just in case something unexpected happens. And I know it will happen, but I cannot predict what, how, or where. Buffer overflows, as a niaeve example. Unhandled exceptions as another. Your model routinely produces them, and yet cannot entertain that they even exist until *after* the box blows up. Mine assumes they will happen, always, and will often count on it. You're free to pretend that a given state will always exist at a certain line - but you're a fool if you believe that to be true. All it takes is one cat hair getting sucked into your machine and landing on your RAM, and you'll be eating crow. Cat hair does not exist within your "fundamental assumptions" - you'll not find it in any book by Booch, and until today you've probably never conceived of it as a relevent problem, and probably still discount it as irrelevent even now. That's the difference - and you are completely wrong. You cannot prove it'll ever happen, so you respect the razor's demand for removal. I cannot prove it will NOT happen, so I must demand that it will.
You're viewing from an engineer perspective... that which isn't explicitly allowed is absolutely forbidden. I'm viewing from a software perspective - that which isn't explicitly forbidden is absolutely demanded. The differences are subtle, but the impacts are severe... I'm allowed to handle a recursive abstraction. You must not (and god help us all if you did)... and in fact, you must disclaim that such a concept cannot exist.
Your model of a theory assumes that it is complete until shown otherwise, and the Razor has a high value in making the theory manageable. "It goes against scientific principles" - no, it goes against physics-specfic scientific *method*... the introduction of seemingly spurious parameters only complicates a model, making it unmanageable. So, the razor is good, and the razor is god in that arena.
Apply that method as a software theorist, however, and you're deader than a racecar with a
Physicists love to abstract a behavior... but they don't routinely abstract the abstractions, and study the abstractions of those abstractions for commonalities. And quite frankly, they should not... as you say, the "barking up the wrong tree" factor for them is potentially huge. Their work is based on the premise that knowlege (and possible behavior) is finite, ergo there are a finite number of predictions that must be made. That's fine, it makes physics viable as a research field - but that doesn't make it true, it only makes it convenient for the sake of the physicists. You perhaps should reconsider your definition of "scientific method", because it is not complete.
In other words, you've done nothing to demonstrate that there *aren't* a large number of dimensions - you said yourself, there's no evidence. The Razor demands their removal; but the software theorist views that as sheer arrogance - after all, we have plenty of GOOD evidence to predict their existence. There are at least two dimensions to space. Oh, hey... make that at least three. I'm seeing a pattern here, Errr.... ok, four. Induction, anyone? No reason to stop - not at 20, not at 30, not at a zillion. Especially not at 20 or 30 like the current batch of quark-heads use... dude, you've already got *30* freakin dimensions in your space, and you added them one at
Photon paradox: Double-slit experiment. Wave or particle, depending on which you choose to observe. The paradox exists due to the model by which you perceive it - if you use another model, it isn't a paradox at all, right? That's the idea I'm trying to get across, at least, that a paradox is only relative to the perception model and merely indicates that the model is incomplete / not a general case.
:)
For the causality thing - the easiest way to envision the game is to consider a simple iterative function, and a surface who's coords can be mapped as inputs. The output then gets used as the next input, etc - classic chaos. There's a problem that we need to avoid, however - if at any time a full set of coords is repeated, you'll end up with a loop and it won't ever break. THAT is how I envision the dreaded "breaking causality" you mentioned... if all inputs are duplicated, you've effectively violated Pauli's exclusion in a philosphical sense and "the universe" needs a ctrl-alt-del. Bad, bad, bad.
So, to get around that - given, say, 4 parameters - if any trio of coords have already been used together, it isn't a problem if the 4th *didn't* happen with them. Simplistically, "no two objects in the same space." I put a pen on my desk - remove it, then put my keyboard in that same spot. Relative to a 3-space model, I've violated that rule. Relative to a 4-space model, I haven't - the first trio is "already used", but the 4th wasn't (with that trio, at least.) Sadly, I can't describe it much better without being incredibly long-winded.
Now for my challenge about breaking causality - given an input space of (n) - and we only directly perceive 3 vectors of it - we know that we cannot (trivially) differentiate the remaining (n-3) vectors from the single sum of those vectors - we'll perceive them all as a single vector, called "time". (n) might indeed be 4, or it might also be a gazillion, we cannot easily tell - from our perspective, we'll chalk them all up as a single net-effect. (I've come up with a couple of great yet simple games to demonstrate this to lay people, if you're interested. It's an easy and fun way to generate a forehead slapping "Aha!" from a victim. And when you're done, they'll never view a sheet of paper or driving a car the same way again, lol.) So... given the case that (n) might be huge (induction sort of demands that (n) might be infinite, if you put any salt into that sort of thing), and our models are at best playing with 20 or 30 of them - we're artifically capping (n) at 20 or 30, when (n) is in fact greater. In that case, having a duplicate "state" is not an issue - it's only duplicated in respect to the (20 or 30) space we're observing from, not the true (n) space. So long as (n) contains one axis more than we're aware of (and again, there's at least 4 that we're aware of... so induction says... induction says the theorists are potentially screwed trying to model it, Argh.) then causality will not be broken, because there are inputs (states) that are still arbitrary.
Now consider all of that drivel with this game - and yes, this game is an abomination against everything... but it's my game, so we'll play it just to see what happens. We're going to play the "Positional Exclusion" game. We'll use a two-axis surface that is the universe, we'll put an object on it called P. And we'll have two rules - Our rule: we view the state of the game as a sequence of instantaneous observations. THE rule we're interested in: P can never exist in the same place (state) twice in a row.
So, we take our initial look, and P is somewhere. Look again, and P moves. How far? It moves the minimum distance demanded to not violate the rule - it moves the distance of its radius, let's say. Never more because it's a lazy P, and never less because it's forbidden. Look again, and it moves again... the distance of it's radius. A simple game, totally wrong and doesn't show much - in fact, we can't even explain "why" P moves. But watch what we can do with it.
Yeah, but a paradox is usually relative to a certain perspective. Hitting one indicates nothing. Photons are a paradox, and you're suggesting that they're "not allowed" in science.
:) It is simply a case where both results are true, but the question is incomplete.
Switch to a more macro perspective, and there is no paradox. Much like being local to a recursive function - from within it, there can be a paradox with causality. From outside, however, the causality of that function is obvious. In this case, what we perceive as "breaking causality" might in fact appear linear from a larger, more "macro" view... e.g. more than just niaeve 4-space, or worse 3-space.
Regardless of what n-space you choose, however, causality only demands ONE axis to maintain unique state. The remaining (n-1) can go "back in time" all day long, it really doesn't matter - and therein lies my point (argh! bad pun, sorry.) If our "perspective" is constrained to within the (n-1) space, then our perspective is incomplete - and causality *will* appear broken. Causality isn't, however, when perceived from the full (n) view.
And since I'm pretty damned sure we haven't actually solved for (n) yet, I'm pretty sure we can appear to break causality all day long relative to our current niaeve ruleset. And no, approximations of (n) that are convenient for a specific model do not make a solution for (n). When (n) is resolved, the resulting space would appear as being static / e.g. no such thing as "time", relative to *anything*. Time, after all, may well be the perception of the sum of the remaining n's that we can't directly observe.
Fear not the paradox
No, but oddly... Windows Longhorn is suddenly back on schedule again.
I'm still trying to figure out where they prove, or at least try ot demonstrate that such causality cannot be broken. Call me crazy...
Oh come now... the escrow attempt by the USPS was met with resounding support, honest. ;)
I think you can pretty much cut-n-paste that entire thread. Yuckie.
Actually, my reply would be (if I were to make one)
THERE, SEE? YOUR DAMNED GLOBAL WARMING HAS NOW SCREWED UP THE SUN!
I won't post that, though. Too many morons would probably think, "Hey... what if that's true..."
--- Begin Evil DoubleSig: ----
"If English was good enough for Jesus, it should be good enough for these people."
... some guys have invented "emulation mode" for atoms. I expect their project should have an Alpha release, soon... ;)
No, I have little confidence that're no exploits in any offered services. What I said is that a firewall has no relevence to that issue.
Learn to read english, you putz.
Dumb question, but is that related to the alg used in mrSID files? I'm grabbing 800~900K .sid files, and converted as tifs yields about 25meg per tile. Converting to .jpg, very lossy, yields about 1.5 megs per tile on average. Converting to jpeg with very low loss yields about 5 megs per tile.
.sids, and equally as curious.
I only ask because you mentioned sat imagery... I'm recently quite impressed by