Vanilla 1.1.9 is a product of Lussumo. More Information: Documentation, Community Support.

    • CommentAuthorjoshs
    • CommentTimeMay 6th 2010
     
    Posted By: timetrumpet
    True, but not at issue.
    It is the only issue. And here comes the straw man:


    This particular point of disagreement rests on whether or not a developer who fails to anticipate that an errant result may be produced in some circumstances is an idiot. Your answers so far suggest that you hold this opinion. I strongly disagree.
    You are being deliberately obtuse or an idiot. The premise on which an algorithm is based provides a means to check the results. It is by definition inviolate. You can't even slay your own straw man.

    Excess pointless drivel snipped.
    • CommentAuthortinker
    • CommentTimeMay 6th 2010
     
    "Excess pointless drivel snipped." Good of you joshs - but I thought pointless drivel was your forte?
    • CommentAuthorjoshs
    • CommentTimeMay 6th 2010
     
    Posted By: tinker"Excess pointless drivel snipped." Good of you joshs - but I thought pointless drivel was your forte?
    Then you were mistaking me for yourself.
    • CommentAuthortinker
    • CommentTimeMay 6th 2010
     
    hehe.
    •  
      CommentAuthorNewsEditor
    • CommentTimeMay 6th 2010
     
    Posted By: tinker"Excess pointless drivel snipped." Good of you joshs - but I thought pointless drivel was your forte?


    Benign insults and humour that would make a 6 year old laugh are his forte Tinker.
    • CommentAuthorjoshs
    • CommentTimeMay 6th 2010
     
    Posted By: NewsEditor
    Posted By: tinker"Excess pointless drivel snipped." Good of you joshs - but I thought pointless drivel was your forte?


    Benign insults and humour that would make a 6 year old laugh are his forte Tinker.
    Perhaps. OTOH supporting scammers is your forte Craig.
    •  
      CommentAuthorNewsEditor
    • CommentTimeMay 6th 2010
     
    Posted By: joshs
    Posted By: NewsEditor
    Posted By: tinker"Excess pointless drivel snipped." Good of you joshs - but I thought pointless drivel was your forte?


    Benign insults and humour that would make a 6 year old laugh are his forte Tinker.
    Perhaps. OTOH supporting scammers is your forte Craig.


    What scams? Bringing people information on technologies that have been denied to them? If that's the scam - then guilty as charged.
    • CommentAuthorBigOilRep
    • CommentTimeMay 6th 2010
     
    Oh yes, the scammer supporter. I'm still waiting to hear from your lawyers for that alleged libel Craig - or was that just another empty e-threat from a child in his bedroom?
    •  
      CommentAuthorNewsEditor
    • CommentTimeMay 6th 2010
     
    Posted By: BigOilRepOh yes, the scammer supporter. I'm still waiting to hear from your lawyers for that alleged libel Craig - or was that just another empty e-threat from a child in his bedroom?


    Your maw's got bawz.
    • CommentAuthorjoshs
    • CommentTimeMay 6th 2010 edited
     
    Posted By: NewsEditorPerhaps. OTOH supporting scammers is your forte Craig.

    What scams? Bringing people information on technologies that have been denied to them? If that's the scam - then guilty as charged.
    LOL! What technologies have the public been denied that you have exposed?

    Archer Quinn had no technology. Mylow had no technology. Steorn have no technology. They all have had scam stories and nothing more.
  1.  
    Posted By: joshs
    Posted By: timetrumpet
    True, but not at issue.
    It is the only issue. And here comes the straw man:


    This particular point of disagreement rests on whether or not a developer who fails to anticipate that an errant result may be produced in some circumstances is an idiot. Your answers so far suggest that you hold this opinion. I strongly disagree.
    You are being deliberately obtuse or an idiot. The premise on which an algorithm is based provides a means to check the results. It is by definition inviolate. You can't even slay your own straw man.

    Excess pointless drivel snipped.


    You snip the portion that is difficult - nay impossible to refute. Convenient for you.

    Try being less worried about being "wrong". It's a great burden you should free yourself from.

    The point under discussion in this exchange is whether it is reasonable to cast someone as an idiot or negligent for failing to predict a circumstance where the results deviate from their expectation.

    Your usual dull face-saving attempted sophistry makes no difference to the fact that it is unreasonable to predict all sources of error introduced by the likes of approximation and iteration. You can make great strides in quantifying the uncertainty in your results; you can't eliminate the unexpected; you hope you can investigate and mitigate against it when it's identified.

    i don't believe that you don't realise this. I do believe that you'll perform any ungainly public linguistic or logical dance you believe is necessary to avoid facing the fact that you called it incorrectly.
  2.  
    Posted By: SPECTRE
    Posted By: timetrumpet@SPECTRE

    No. You're failing to apprehend the discussion. You should really stop before you embarrass yourself further.

    They will do what they're programmed to do (barring catastrophic hardware failure, cosmic ray events or other imponderables). This is not always what you intend them to do. There are a number of reasons of varying subtlety and complexity that may contribute to the degree of separation between the two. You might want to go and read up on a few.


    Because you're actually just plain wrong, which is no surprise, I'll simply answer by saying shit or get off the pot. Put up your evidence of your absurd claim that doesn't make any sense to any other person on this board except tinkerbell. Otherwise, your boring pontifications are just that; boring, rambling, and wrong. As Josh pointed out, your entire argument is an epic fail.


    I'll come back to you (and others) presently, but just to clarify; I'm not claiming to represent @tinkers position WRT the pitfalls of simulation or the limits you can expect to put on the performance of complex systems in the abstract. And I'm not claiming they make sense to him - you are.

    I don't think they make sense to *you*, but that's because you don't have the knowledge you need to support your position and you choose to remain ignorant. I allow for the fact that I think you're discussing the subject in a second language but it's evident that there are great swathes of subject matter that are a closed book to you.

    Recognise that and do something about it.
    •  
      CommentAuthorpcstru
    • CommentTimeMay 6th 2010
     
    Posted By: timetrumpet
    The point under discussion in this exchange is whether it is reasonable to cast someone as an idiot or negligent for failing to predict a circumstance where the results deviate from their expectation.


    Originally the point under discussion was :


    Posted By: pcstru
    Posted By: timetrumpet
    Simulations exist. No simulation is perfect. It's perfectly possible that a simulation shows a gain. It's even more probable that any gain is a result of error somewhere along the line.


    It's not 'probable' - it's an absolute certainty. A simulation correctly programmed with the known laws governing motion, charge and magnetism cannot show a gain.



    Nothing to do with the competence of the programmer but all to do with the fact that IF the simulation is a correct representation of the maths, then the simulation cannot show a gain.

    But now you seem to be trying to wriggle away from that as fast as you can.
    • CommentAuthorBigOilRep
    • CommentTimeMay 6th 2010
     
    Posted By: timetrumpet

    Recognise that and do something about it.

    And you are a self-righteous arrogant prick fond of sophistry. Recognise that and do something about it.
    • CommentAuthorjoshs
    • CommentTimeMay 6th 2010
     
    Posted By: timetrumpetYou snip the portion that is difficult - nay impossible to refute. Convenient for you.
    Try being less worried about being "wrong". It's a great burden you should free yourself from.

    The point under discussion in this exchange is whether it is reasonable to cast someone as an idiot or negligent for failing to predict a circumstance where the results deviate from their expectation.

    Your usual dull face-saving attempted sophistry makes no difference to the fact that it is unreasonable to predict all sources of error introduced by the likes of approximation and iteration. You can make great strides in quantifying the uncertainty in your results; you can't eliminate the unexpected; you hope you can investigate and mitigate against it when it's identified.

    i don't believe that you don't realise this. I do believe that you'll perform any ungainly public linguistic or logical dance you believe is necessary to avoid facing the fact that you called it incorrectly.
    Kettle and pot. You lost the argument and have been trying to redirect it to a straw man you think you can win. The entire discussion began and ended with whether a computer program written on a presumption of CoE could evaluate for CoE. It cannot. If A is presumed, then A cannot be evaluated. If a result indicates that a presumption has been violated then the result is by definition in error. This in fact happens all the time, and it is used to check for things like unnatural results due to numerical, algorithmic, or source data error. You can spin all you want. You're getting nowhere.
  3.  
    Posted By: pcstru
    Posted By: timetrumpet@SPECTRE

    No. You're failing to apprehend the discussion. You should really stop before you embarrass yourself further.

    They will do what they're programmed to do (barring catastrophic hardware failure, cosmic ray events or other imponderables). This is not always what you intend them to do. There are a number of reasons of varying subtlety and complexity that may contribute to the degree of separation between the two. You might want to go and read up on a few.


    So a simulation of a system based on known physics, which incorporates CoE as a basic assumption (as proved mathematically by E. Noether), could produce a result which shows additional energy and you say that there is a possibility (maybe just a very slim one) that the result is correct?

    Mathematically that is much the same as 2+2=5. How about a slightly more complex example - a^2+b^2=c^2. Mathematically we have a proof that this is true. Yet from what you say, a simulation which produces a result 5.0067 from a=3 and b=4, might just be right - albeit a slim possibility? If not, how is that different from any of the equations for motion, magnetism or charge?



    I'll separate the two strands of the discussion but just the simple "joshs" case first. I'll return to the other at some point as it's harder to discuss.


    So, is the simulation accurate? That is, when it iterates through it's various cycles at it's chosen resolutions in space, time, whatever, does the result tally with what I'd expect from my design or symbolic laws and equations, and with what I'd expect to see in the real world?

    The "simple" errors I may see here stand a chance of being quantified and fully understood; compare the symbolic manipulation of fractions with their decimal representation; you'll be seeing approximation and rounding errors, and over millions of iterations with millions of variables, suddenly your subtle fogging effect shows solarization, or your 3D tank has become fatally embedded in your wall.

    These are relatively easy to mitigate against when they're discovered, but at a cost; finer resolution in space and time, greater floating point accuracy at a cost of greater processing and storage requirements. You're always trading off these constraints depending on the required accuracy and problem domain. You may have to revise them in light of actual results, you can't catch them all ahead of time.

    So that leads you to simulations that may give unexpected results - gains in energy for instance - but these can be accounted for in retrospect by systematic errors and uncertainty. Nobody needs to be an idiot when they design the simulation; nobody needs to be a liar if they report it appears to show a gain.


    The second main source of error is error "by design". That is, divergence from the predicted and observed behaviour due to an incomplete or oversimplified model - no account taken for relativistic effects, or solar pressure, or whatever relatively subtle effect exists in the real world as opposed to your virtual abstraction of it.

    Again, you may need to actually see your spacecraft drift off course in an unexpected manner before you realise that the omission that you made for the sake of optimisation (or just hadn't considered as significant in the design) needs to be accounted for.

    So second version; no idiots, no liars, but (usually) a clearly identifiable source of error.


    But these are in the same class as the relatively simple examples you've quoted in your response. They may be "real" output from the simulation, but recognised as incorrect on examination or in retrospect.


    The second much subtler issue - the "might just be right" issue - is a different discussion. It's related to the second point above - have I included every "law" - but needs to be laid out separately. And it's qualitatively different from the mathematical analogy you provide.
  4.  
    Posted By: pcstru
    Posted By: timetrumpet
    The point under discussion in this exchange is whether it is reasonable to cast someone as an idiot or negligent for failing to predict a circumstance where the results deviate from their expectation.


    Originally the point under discussion was :


    Posted By: pcstru
    Posted By: timetrumpet
    Simulations exist. No simulation is perfect. It's perfectly possible that a simulation shows a gain. It's even more probable that any gain is a result of error somewhere along the line.


    It's not 'probable' - it's an absolute certainty. A simulation correctly programmed with the known laws governing motion, charge and magnetism cannot show a gain.



    Nothing to do with the competence of the programmer but all to do with the fact that IF the simulation is a correct representation of the maths, then the simulation cannot show a gain.

    But now you seem to be trying to wriggle away from that as fast as you can.


    This lot came up during my last (time consuming) reply. I'm not wriggling away, I'm heading towards it as fast as I can, but not as fast as I'd like.
    •  
      CommentAuthortimetrumpet
    • CommentTimeMay 6th 2010 edited
     
    @joshs

    You're the one harping on about "competency":

    "That's a circular expression of claiming that resource constraints force error that violate fundamental principle. Even in your circling the drain argument you admit the program is in error if it violates its fundamental assumptions. In such a case if the program designer doesn't understand how to find the boundaries where the program generates incorrect results, then he is not competent to design the program. So you are still left in the losing position of trying to defend the disproven idea that a competent program premised on a rule it must universally hold can be written to find an exception to that universal rule. If the program is competently written, it does not violate its underlying rules. Saying that it can is a direct logical self-contradiction. It is a fail."


    "Competency" is a poor choice of term - but it *is* your term after all.


    ETA: And this from an earlier reply:
    "Anyone who constructs a simulation who does not understand how to test the simulation for validity is incompetent."

    That certainly seemed to be the focus of *your* attention at one point, a lead I've followed. You're now content to accept you overstated the case apparently. Fine. Did it hurt much?



    "Saying that it can is a direct logical self-contradiction."


    How true. I'll return to that. Logical self-contradiction is the nub of the matter. It may be a fail, but it isn't a fail for me, it's a fact of reality, virtual or otherwise.
  5.  
    The End - Der Untergang - Downfall - Next level:

    http://www.youtube.com/watch?v=0JpmvtPXQR4
    •  
      CommentAuthorpcstru
    • CommentTimeMay 6th 2010 edited
     
    Posted By: timetrumpet
    I'll separate the two strands of the discussion but just the simple "joshs" case first. I'll return to the other at some point as it's harder to discuss.

    So, is the simulation accurate? That is, when it iterates through it's various cycles at it's chosen resolutions in space, time, whatever, does the result tally with what I'd expect from my design or symbolic laws and equations, and with what I'd expect to see in the real world?

    The "simple" errors I may see here stand a chance of being quantified and fully understood; compare the symbolic manipulation of fractions with their decimal representation; you'll be seeing approximation and rounding errors, and over millions of iterations with millions of variables, suddenly your subtle fogging effect shows solarization, or your 3D tank has become fatally embedded in your wall.

    These are relatively easy to mitigate against when they're discovered, but at a cost; finer resolution in space and time, greater floating point accuracy at a cost of greater processing and storage requirements. You're always trading off these constraints depending on the required accuracy and problem domain. You may have to revise them in light of actual results, you can't catch them all ahead of time.

    So that leads you to simulations that may give unexpected results - gains in energy for instance - but these can be accounted for in retrospect by systematic errors and uncertainty. Nobody needs to be an idiot when they design the simulation; nobody needs to be a liar if they report it appears to show a gain.

    The second main source of error is error "by design". That is, divergence from the predicted and observed behaviour due to an incomplete or oversimplified model - no account taken for relativistic effects, or solar pressure, or whatever relatively subtle effect exists in the real world as opposed to your virtual abstraction of it.

    Again, you may need to actually see yourspacecraftdrift off course in an unexpected manner before you realise that the omission that you made for the sake of optimisation (or just hadn't considered as significant in the design) needs to be accounted for.

    So second version; no idiots, no liars, but (usually) a clearly identifiable source of error.

    But these are in the same class as the relatively simple examples you've quoted in your response. They may be "real" output from the simulation, but recognised as incorrect on examination or in retrospect.

    9 paragraphs just to state that simulations can have errors which can be accounted for. Here's my statement again : "It's not 'probable' - it's an absolute certainty. A simulation correctly programmed with the known laws governing motion, charge and magnetism cannot show a gain."

    Note carefully the word "correctly programmed". But you choose to spend 9 paragraphs detailing incorrect implementations.

    The second much subtler issue - the "might just be right" issue - is a different discussion. It's related to the second point above - have I included every "law" - but needs to be laid out separately. And it's qualitatively different from the mathematical analogy you provide.


    If it is qualitatively different, then please address the substance of that difference. 10 paragraphs and you haven't addressed the simple questions, yet you seem to think others are missing the point!