A working version of FMM’s design detection game is available.
Download and install the applicable version of “Processing”.
https://processing.org/download/?processing
Get the fifthmonarchyman progam code from here, and paste it into the Processing script area.
Sample data here
http://pastebin.com/raw/MjV8RmvW
You need two files in the same folder as the Processing executable.
real.txt and fake.txt
The testing and such starts here
test strings
Meanwhile, your method is doomed by your own standard:
Why are you pursuing a method that depends on an obviously false assumption?
I don’t think learning (syntax) is different than learning (semantics). It’s simply a lower rung on the Y axis. The more I know the higher up I go.
I don’t have to know that 8675309 is Jenny’s number to know that it is a string of 7 numbers with no repeats. I can also learn that 8675309 is is prime, and is part of a twin prime pair. I also can learn that the sum of the digits is 38. Etc etc
All of these things I can know with out any prior knowledge of the string simply by studying it and perhaps comparing it with models that are close.
Of course there are some things about Jenny’s number that I will never know but that does not mean I can know nothing.
Jenny’s number is unique nonrandom and noncomputable but in the context of my it is not complex enough for me to infer design.
That does not mean that can’t serve as a handy illustration of what the game is about.
peace
Its the new homeopathy!
No offense, but I sometimes feel like you are being deliberately obtuse. When I say there is no general Algorithm that will produce Jenny’s number I mean just what I say.
You can’t get to Jenny’s number by adding one unless you know when to stop. This is patently obvious.
I can’t be held accountable for your choosing to take my meaning in the most uncharitable woodenly literal way possible.
Claiming that I moved the goal posts when I merely pointed out what anyone who is following this discussion should already know is just silly talk
peace
No the assumption is not false.
The only way you can conclude that it is a false assumption is by assuming “falsely” that true randomness exists.
Do you have any evidence whatsoever to support this assumption?
peace
You are making the same sort of leap that keiths is making.
You are assuming that just because we can’t say that a particular string is noncomputable means that it is computable.
It is possible that all strings are noncomputable and everything is designed.
My Game does not rule that possibility out
What my game does is allow you to say that some strings are definitely noncomputable. That is all
Some S are D
does not logically imply that
Some S are not D
peace
Would you agree this would be falsified by two processes creating indistinguishable objects?
No
There are infinite ways to produce any finite string
peace
faded_Glory:
fifth:
No, he’s describing a case in which you will say that a string is non-computable despite the fact that you just computed it!
In fG’s scenario, your game falsely flags computable strings as non-computable.
What good is your game if it produces scads of false positives, with no way of telling when it is doing so?
Slow down and think, fifth.
fifth,
No offense, but often I feel you are being obtuse despite trying not to be.
That isn’t what you said. You wrote:
I showed you a general algorithm that produces them, which is exactly what you claimed does not exist.
Think before clicking on ‘Post Comment’, fifth.
Are you still clueless as to what nononcomputable means in this context.
hint again:
Just because I can produce a sting with an algroythym does not mean that that string is computable
When I say something is noncomputable I mean that there is no generalized algroythym to produce it.
Before you jump on this and say something like any finite string can be produced by the general algroythym of adding one over and over. Please go back and read where I addressed this and this time with an eye to understanding.
Really, is that your advice to me?
Given we have seen you repeatedly jumping off on wild tangents without so much as a moments reflection about what is being said.
I would just say “physician heal thyself”
peace
Geeze
You did no such thing, you showed a general algroythym that produces an infinite amorphous cloud of numbers.
That sort of thing will never get you to Jenny’s number unless you know exactly when to holler stop
peace
fifth:
Rich:
fifth:
Each of which should leave a trace, according to you. But the strings are indistinguishable, which means that your method is hopeless, by your own standard:
Why?
You seem to think that in order for a string to have a trace of the process that produced it it needs to be distinguishable from other strings.
I see no reason why this conclusion should follow from the premise.
look at the following strings
12345
12345
They are indistinguishable but they both show evidence of being produced by something with a familiarity with Arabic numerals.
this is not rocket science
peace
fifth:
No, I showed a general algorithm that produces an ordered list of numbers, including both Mary’s and Jenny’s. It’s exactly what you claimed does not exist:
fifth:
Of course it will. It will get you to 5, to 764, to 8674309 (Mary’s number), and to 8675309 (Jenny’s number). It will get you to every finite integer, as you know perfectly well. The stopping requirement is something you added after you realized your mistake.
Are you now asking for a “general” algorithm that stops at every number, including Mary’s and Jenny’s?
Think, fifth.
keiths:
fifth:
Come on, fifth.
If the traces aren’t distinguishable, then you are saying that your method depends on a distinction that you cannot make:
A perfect foot-shot.
If you’d admit your mistakes, you could at least salvage some dignity and have another go at the problem. As it is, you’re stuck in your own web of contradictory statements and errors. Every move you’re making leaves you more tangled than before.
This sort of thing is exactly what I have come to expect from you. It’s why I long ago gave up on any kind of fruitful discussion with you
I have been talking about “halting” and “choosing” for literally months and you act as if I just came up with it this morning,
I only just yesterday explained this to all to you again in reference to Pi
Again it’s like you don’t even know what is being discussed here. After all this time
I have no idea if this total lack of understanding is genuine or if you simply are pretending for some strange reason.
Either way it’s just not worth the trouble to continue to correct you here.
You will just need to wait till OMagain finishes his work and we can begin some hypothesis testing.
I predict you will continue to crow on with no idea what you are talking about
peace
fifthmvonarchyman,
It sounds like you’re not only moving the goalposts, you’re changing the sport. My proposal is about the Financial Turing Test paper you referenced in this comment and previous others:
If you are changing the criteria from that described in the Financial Turing Test paper to that of your ill-defined game, please say so explicitly.
If not, please address my questions here and here in that context.
fifthmonarchyman,
You’ve already recognized that any finite string is computable. You’re contradicting yourself.
fifthmonarchyman,
When no one but you can understand what you’re talking about, you might want to consider where the root cause of the confusion lies.
The problem is that I have no confidence you have ever even played the game in question. Your proposal doesn’t seem to be about the actual game. So I’m not sure you even get what is going on here
You don’t mention the role of feedback or the experience that the game engenders. You don’t talk about what it means to learn the patterns in the stings etc etc etc,
your mention of a 75% success rate is totally out of context and ignores the near universal ability of humans with feedback to distinguish between the original strings and randomized copies
My proposal was an attempt to get you to focus on the actual game, Instead of irrelevant minutiae that misses the entire point.
peace
Lots of people seem to understand just fine. It’s only a couple folks that seem clueless.
It’s mostly the same folks who don’t seem to understand that lossless information integration is about data compression.
peace
And there is that total lack of understanding again,
Other folks on your side get it even if they wish I used another term, Why this profound mental block in your case?
peace
Who on “our side” gets it?
fifthmonarchyman,
I have demonstrated that I understand the Financial Turing Test better than you in this comment.
I have quoted the actual paper to support my summary and the 73% success rate. You have spewed nonsense without any reference to the paper.
I have focused exclusively on the contents of the paper. Either point out where the paper supports your vague allusions or admit that you are not talking about the same process.
fifthmvonarchyman,
Name one other person in this thread who understands how to play your game.
It’s high time you provided some detailed examples.
Patrick:
fifth:
petrushka:
For that matter, who on fifth’s side — including fifth himself — gets it? If fifth “got it”, he would present it. Instead, he’s still desperately modifying his position in an attempt to escape his earlier mistakes.
I for one would be delighted if fifth could produce someone who both “got” his method and actually believed that it works. Is there such a person, fifth? If so, then perhaps he or she could do a better job of defending your “method” than you have.
Even Salvador is trying to save fifth from embarrassing himself.
fifth,
Previously you were talking about halting within the otherwise infinite decimal expansion of a constant like pi. Now you are talking about halting within an infinite list of finite strings. It’s completely different.
And you did make it up today. You had to make it up after you fooolishly wrote:
Your statement is clearly wrong, but you only realized that after I pointed it out and provided just such a general algorithm. Hence the typical FMM bluffing and “I didn’t mean what I wrote, and it’s your fault for taking me at my word” protests.
Worse still, your objection would be incoherent even if you had been making it for months. I already pointed out the flaw:
If the lack of such a nonsensical general algorithm is an indicator that the numbers in question are non-computable, as you claim…
…then every integer is non-computable.
If you want to hang your hat on that ridiculous claim, then be my guest.
Just yesterday I said
quote:
It is possible that all strings are noncomputable and everything is designed.
My Game does not rule that possibility out
end quote:
Did you miss that comment or ignore it or what?
Integrating functions are noncomputable
This was mathematically proven in the “is consciousness computable” paper. That we are still discussing this demonstrates that you have either not read the paper or have failed to understand it.
I have been touting that paper and it’s implications since I first arrived on this website. The only thing that is new is that you are perhaps finally understanding what this discussion is about
It’s a simple concept there is no reason for you not to be able to get it.
Peace
Nobody undrstands my entire argument as I have not made it yet
However on this site Neil understands the general idea as does FG and OMagain and Walto I believe.
Most of the rest of the folks don’t care or are simply waiting for the game and OP and are not engaged at all in the details.
peace
That is the point the paper is simply a summary of a game, You need to play the game to understand the paper. The authors of the paper have said as much and made the game available to anyone.
Your hack is supposed to duplicate the abilities of humans in playing the game
Why don’t you play it and see what you are up against?
peace
That’s interesting. As far as I know I’m the only person besides you that has installed the software and run some data through your game. And I haven’t a clue what it is you claim to see.
All (I think) I understand is the technical details of your game. I do not understand what you claim it actually demonstrates, nor do I understand why you think a string that you just computed would not satisfy the criteria for being ‘non-computable’ if such would be the outcome of your game.
In other words, I do not understand why my thought experiment of inputting a computed model M1 of an original O1 into a second round of the game, where I suspect it would now be flagged as a ‘non-computable’ string O2 when compared to a new model M2, is not fatal to your entire experiment. I do not understand what you mean when you say “When (I say) something is noncomputable I mean that there is no generalized algroythym to produce it.”
Why is your own GA that computes Model M1 not ‘a generalized algorithm’? Please be clear, precise and exact in your definition of ‘non-computable’, and explain carefully how your game can differentiate between ‘computable’ and ‘non-computable’ strings according to your definition.
It was not mathematically proven, it seems more like proof by redefintion
Maguire is lead author of the paper
“But there is a catch, argues Phil Maguire at the National University of Ireland in Maynooth. He points to a computational device called the XOR logic gate, which involves two inputs, A and B. The output of the gate is “0” if A and B are the same and “1” if A and B are different. In this scenario, it is impossible to predict the output based on A or B alone – you need both.
Crucially, this type of integration requires loss of information, says Maguire: “You have put in two bits, and you get one out. If the brain integrated information in this fashion, it would have to be continuously haemorrhaging information.”
Maguire and his colleagues say the brain is unlikely to do this, because repeated retrieval of memories would eventually destroy them. Instead, they define integration in terms of how difficult information is to edit.
Consider an album of digital photographs. The pictures are compiled but not integrated, so deleting or modifying individual images is easy. But when we create memories, we integrate those snapshots of information into our bank of earlier memories. This makes it extremely difficult to selectively edit out one scene from the “album” in our brain.
Based on this definition,Maguire and his team have shown mathematically that computers can’t handle any process that integrates information completely. If you accept that consciousness is based on total integration, then computers can’t be conscious.
“It means that you would not be able to achieve the same results in finite time, using finite memory, using a physical machine,” says Maguire. “It doesn’t necessarily mean that there is some magic going on in the brain that involves some forces that can’t be explained physically. It is just so complex that it’s beyond our abilities to reverse it and decompose
Neuroscientist Anil Seth at the University of Sussex, UK, applauds the team for exploring consciousness mathematically. But he is not convinced that brains do not lose information.“Brains are open systems with a continual turnover of physical and informational components,” he says. “Not many neuroscientists would claim that conscious contents require lossless memory.”
Maguire acknowledges that their proof would not hold up if information integration in the brain is reversible. “Maybe, if you had a very clever algorithm, you could still break down peoples’ memories and edit them.”
Journal reference: arxiv.org/abs/1405.0126v1
https://www.newscientist.com/article/dn25560-sentient-robots-not-possible-if-you-do-the-maths#.U30qmfldV8E
The rumors of my understanding are greatly exaggerated.
All this is predicated on brains being logic gates, which I don’t think is even in the running as a theory.
This is, more or less, the problem.
The word “computable” is properly applied to formal problems. We lack a suitable formalization for what FFM is trying. So he is using “computable” and “not computable” in a confusing and misleading way. He is probably talking about whether his problem (whatever it is) can be adequately formalized, and using “computable” to mean that it can.
I think much of the confusion could be cleared up by having FMM state whether he can classify data sets and sort them into two piles.
If the answer to this is yes, then it would be helpful to know what it means to be in one pile or another.
I have heard at least three ways of characterizing one of the piles:
1. One of the piles produces a characteristic “W” when graphed by FMM’s program.
2. One of the piles represents “real” data, whatever that means.
3. The data contains traces of the means by which it was generated.
All of this seems to boil down the the claim that pile one is the pile that FMM makes when playing the game.
The only certain characteristic of this pile is that FMM has sorted stuff into it.
Apparently there is no independent way to break out of this circular definition. And that, apparently, is the point of the game.
I am curious, when FMM says he is 100 percent right, or 73 percent right, what he means.
He apparently means that if multiple people play the game, there will be agreement among the players as to which way things get sorted.
If there is some visible attribute to the graphs, it is likely that people will sort them the same way.
But what would that mean, if there is no independent way to sort the data?
I don’t pretend to understand the ‘game’. But there is a specification (an existing program that does X) which I can copy and recast into a similar form. I don’t need to understand it to do that. Source code will be published and pull requests can be made. Perhaps something up this weekend.
I appreciate your work.
I think the common feature of the strings in pile 1 is that they can all be visually distinguished (using his game) from lookalike’s (the ‘models’ or ‘fakes’) that resemble them closely but not fully.
Fmm claims that this feature proves something or the other, in his words that they are ‘non-computable’. Which is not saying that they can’t be generated by some algorithm (because he says some of them can, cf. my example of a cascade of strings), but that there is no ‘generalized algorithm’ that can produce them.
I have no idea what that means.
fG
Well, I suppose it is reasonable to assume that different data will graph differently.
The question is, what is the characteristic of “real” data that is lost when slightly modified?
I believe fmm’s point to be that the ‘originals’ contain some or other property that cannot be algorithmically reproduced without specifying the actual elements of the string one by one. Next, he goes off in different directions: one is his infamous example of ‘Jenny’s number’ and the other is his game where observers are asked to tell originals apart from closely resembling models.
Jenny’s number, of course, has nothing to do with any particular property of a string, and everything with the fact that the observer has a priori knowledge of the meaning of some strings (i.e. he knows some telephone numbers by heart). I do not know why fmm considers it remarkable that someone can recognise a number that they know by heart, but hey, maybe fmm suffers from poor memory? Some people do, after all.
Now, where was I? O yes, the game. The game on the other hand, relies on the presence of patterns in certain strings, and the fact that humans in general are rather good at pattern recognition (sometimes so good that they recognise patterns where there aren’t any, c.f. Rohrschach tests , clouds in the sky and much more). This is nothing new, and again I do not know why fmm thinks his game will demonstrate anything that hasn’t been known already for quite a long time.
I had a look at one of his examples, and clearly the ‘model’ has lower spatial correlation than the original. This is because the GA he used to produce it was stopped at an R square of 0.8. Because of this, the model is less predictable than the original. People can spot this. When this happens, fmm claims that it demonstrates something about the original – it is ‘non-computable’, in his words (a poorly defined and even worse understood term in this discussion), and furthermore his claim is that this ‘non-computability’ demonstrates ‘integrated information’ and hence ‘design’ in the origin of the string. Personally I don’t follow the argumentation here.
One could of course run the GA to much higher correlations, and at some point model and original will be so close that an observer might miss the subtle differences during the game (cf. Dawkins’ WEASEL program which will eventually converge on the target phrase). In that case, the outcome of the game will be different than before: because the original and a computed ‘model’ cannot now be told apart it is considered ‘inconclusive’ whether the original is ‘non-computable’/ ‘designed’.
Fmm has not explained why 0.8 is the ‘magic number’, or even why there has to be any ‘sameness’ threshold at all on which the conclusions about the original string so crucially depend.
Note that in fmm’s setup there is no possible outcome that the original is ‘not-designed’. In other words, everything that can ever be tested will either be designed, or possibly so. No wonder that fmm claims his game is a reliable test of design: the outcome of non-design isn’t even possible, not even in theory!
One more thing: when I object that ‘Jenny’s number’ relates to semantics (the mutually agreed content of the string) and the string game to syntax (the architecture of the string), fmm starts talking about some or other ‘Y axis’ that both these aspects lie on. I do not understand what variable is supposed to plot along this Y-axis, nor what would be the variable on the X-axis (assuming it exists – but why would there be an ‘Y’ without an ‘X’?). I do believe that syntax and semantics are very different things indeed, and struggle to understand what variable could possible span a continuum across them.
fG
It certainly looks like bullshit.
I would like to see FMM compete with GAs on a 10,000 node travelling salesman problem.
Writing a shitty GA does not inhibit well written GAs from outperforming humans on real problems.
One other minor quibble. If some class of string produces a tell-tale pattern, then there is no point in comparing it to a close string.
This is why we need to make the game shareable so that we can compare results. It’s my hypothesis that people will sort the piles in similar ways (and that computers will not),
I think that people are hardwired with the ability to see pattern and with feedback to reconigize non random noncomputable strings.
I think that is basically what we do when we infer design in everyday life. We reconigize something in a artifact that that tells us that it’s source thinks like us.
I need to test that hypothesis.
peace
I think your comments here as well as those of several others demonstrate a certain level of understanding of what I’m trying to accomplish among most of you.
I don’t think anyone here understands my entire argument because I have not made it explicit yet.
For that we need to wait till the Game is done
peace
bingo. When I say that a function is noncomputable in this context all we mean is that there is no finite algorithmic process to accomplish it.
peace
indeed that is the question.
The answer is the pattern.
ie
The information we have losslessly integrated when we “learn” the pattern of the string.
peace
can you tell the difference between the following phrases?
Methinks it is like a weasel
and
Methinks it is like a easel
Do you think that the second phrase has the same meaning as the first?
If you can see the difference and reconigize the change in meaning you understand the concept of lossless integration information.
And the point of the game.
,8 is not a magic number I choose it because in every day life that is well above the threshold I use to determine if a particular regression is good.
In the paper a model was considered close if it differed by one bit. Of course that was assuming a digital process
I think a big part of the work that needs to be done in this enterprise is to clarify what “close” means.
For example I would like to see if we get different results at .8 than we do at .95
I don’t know but I have a suspicion we would not.
Testing this stuff out is what science is all about
peace