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
All I’m saying is that I chose correctly probably 30 times in a row with out a single mistake after an initial learning curve, And as you well know I’m not the sharpest wedge of cheese here
😉
peace
Tell us again what you claim to be able to do.
I could have sworn you were claiming to be able to distinguish real data from random datasets. or from scrambled datasets.
It’s really simple
I claiming that nonrandom noncomputable strings can be distinguished from random strings or models that are close.
That does not mean I can magically tell you which string is real and which one is not only that there will be a difference between real and fake strings that will become apparent with comparison and feedback.
in the case of the first two strings you provided the difference is a distinctive W pattern that one string has that the other does not.
This does not mean that the “W” pattern is the signature of design it only means that the two strings are distinguishably different
peace
And I am telling you that I am providing nonrandom numerically coded data representing real objects. Not algorithmically generated.
And either randomly generated datasets of the same length having the same minimums and maximums. Or scrambled versions of the original data.
Now, I spent some time asking you for the rules, and you assured me that any object could be numerically represented. What I’m submitting is coded instructions for constructing objects.
Cool (you or me) should be able to see a difference between the real data and your fakes using the game.
that is the hypothesis anyway,
There is no expectation of a particular pattern that is a dead giveaway for real.
Every real string will have it’s own pattern. What will remain consistent is that with feed back we should see a difference between the real and fake ones
peace
Well, for fuck’s sake, you don’t need a program to see that the randomly generated data is different from the real data.
It’s like….
Well, different numbers and all.
Of course I can see a difference. The question is whether you can tell which one is the original and which one is either random or scrambled.
Anyone else out there following this? Am I out in left field?
You just claimed that you could not see a difference and got rather upset when I dared doubt you
peace
fifthmonarchyman,
When you originally described this game you did claim to be able to distinguish designed strings from random strings. You failed repeatedly on the binary strings I provided, but that was your claim.
Are you now saying that you cannot detect design? If so, what is the use of your game?
Instead of running off half cocked at 100 miles an hour why don’t you just relax and take a look at the original strings I submitted.
Have you taken the time to do that?
what did you see?
peace
petrushka,
My understanding of fifthmonarchyman’s claim is the same as yours.
Seeing a difference and choosing a designed string are two very different things.
Am I the only one who knows there is a difference between being able to see a difference between a random and a designed string and being able to choose which string is the designed one?
peace
This is kind of embarrassing.
Yes, I can see that one graph goes up in places where the other goes down.
The important question is whether you can say that one dataset represents an object (or at least meets your standard of non algorithmic, etc) and another dataset is gibberish.
fifthmonarchyman,
What was the point of testing the strings I provided if not to determine whether or not they were designed? It’s what you claimed to be able to do.
Can you see that one graph has a characteristic W pattern that the other does not?
peace
I have never doubted that you can see patterns in the strings you have examined. I don’t care.
You have made a rather sweeping claim about the general ability to see distinguishing patterns. If you look back, you might detect just a bit of adversarial discussion about this, and a lot of doubt on my part and on the part of others.
So now we are at the point where we can submit data for you to examine. Quite frankly, I see no distinguishing patterns that can tell me that a string is random or if it’s actual published data.
My hypothesis is that if one of a set of strings is nonrandom and noncomputable you will be able to see a difference between it random strings and close models.
Is that clear?
peace
Fifth, I wouldn’t be asking you if such patterns jumped out at me. I don’t even remember which is which (although I have the original files labeled, so I can go back and compare what I posted with the original).
I suppose it’s clear, but it doesn’t make any sense. If the numbers are different, the graphs will be different.
The question is whether you can reliably label datasets as real data or gibberish.
fifthmonarchyman,
So? It’s easy to see differences between any non-identical strings. That’s what non-identical means. That’s profoundly uninteresting. Being able to determine if a string is designed, as you claimed to be able to do, would be interesting.
I have had the game for over a year now and I have looked at lots of strings one of which was OMagian’s I correctly identified it as designed.
OMagain was in the process of making the game sharable not so I could examine even more strings but so that you all could examine some of your own.
This again is not about some superpower that I have it is about a general pattern recognition ability that humans have that computers don’t.
peace
Put your money where your mouth is. Compare two random but non-identical strings in the game and see if you can see a difference
hint you won’t be able to
peace
Horowitz doesn’t have super powers, but he can play the piano, and I can’t. I came here to listen to you play.
Humor me.
Show me that after a year of practice, you can reliably label datasets as either meaningful (and non-computable) or gibberish.
Well, damn, Fifth, I published an image showing a difference in the game.
Difference is not what we are interested in.
We are interested in whether you can reliably label the one that is not random or scrambled.
here is how went about it
I made a wild guess and checked to see if I was right.
for example first I looked at the number of times a number repeated
I kept trying different patterns till I found one that consistently worked. It did not take very long at all.
The feedback that the game gives is critical. you instantly know if you are on to something or if you are all wet.
like I said it’s like a puzzle
peace
I have no doubt that with some kinds of data and some kinds of transformations, you probably can. I have chosen real data that defines real objects. Let’s see you demonstrate your program with something other than financial graphs.
That is of no interest in real data or random data. It might be for some kinds of measurements, but it cannot be a general rule.
Now that is more like it a friendly challenge.
Fire away, give it a go with me if you like
Just don’t expect me to magically ascertain which of the strings is the designed one.
I will just be able to see a difference or not
peace
random numbers
I can’t seem to upload an image
http://www.adesblog.com/images/large_random_number_generator.jpg
the big take away with this game is that there are no general rules. If there were computers could do it
I’m afraid I have — from the beginning — misunderstood you.
As the game/whatever stands, I have no interest.
Where does the intergrated information theory come in?
I honestly thought we were going to witness a tool for detecting integrated information.
fifthmonarchyman,
If they’re not identical then there are differences. So what?
I repeat: What was the point of you playing your game on the strings I provided if not to determine which were designed and which were not? Your claim was that you could do so.
Changing that claim to being able to tell that different strings are different is uninteresting in the extreme.
I submit that what happens when I learn the pattern in a string is lossless information integration,
This is the reason that Computers are unable to duplicate my accomplishment
moving a step forward I would speculate that “design” is simply the inverse of the information integration process.
peace
What is your accomplishment again?
i can see a difference between nonrandom noncomputable strings and random strings or strings from models that are close
peace
This reminds me of people who can hear the difference between hundred dollar a foot speaker wire and lamp cord, until it becomes a double blind test.
if the two strings are both random you can’t see differences
if one of the strings is not random you can see differences
Computers can’t generally and reliably see differences in either case.
If you give me a string my hypothesis is that I can tell you if it is designed.
I do this by seeing if I can distinguish between it and randomized strings and strings from models that are close
You claimed to be able to whip up a quick little two week hack that was supposed to be able to do just this. if it’s so uninteresting why did you bother to make that claim
peace
I have no problem with a double blind test that is why there are so many different assays and why Omagain was making the game shareable.
Until it’s shareable and results are transparent and public there is no way to completely eliminate cheating.
however
I envision a competition with one group trying to prove that real strings are nothing special and another group trying to show that they are. neither side knowing if their particular results support or undermine their case
IMO that would be fun and scientific and a big improvement from the same tired culture war discussions that you all like so much.
It would also be a way move this discussion forward a click or two,
peace
Care to elaborate?
I personally thought Patrick should have seen it coming from a mile away. You a fisherman,Fifth?
I’m a bit confused about what it means to be a real string and what it means to be special.. My operational definition of special is that given a pile of real strings and a pile of gibberish strings, you can sort them into two piles and label them.
How is this wrong?
fifthmonarchyman,
No, I said I think it would be possible to use machine learning to do as well or better than humans on the Finance Game in the paper you referenced. Given the way the goalposts are moving in this thread, I’m glad I haven’t had time to try it yet. Before I do, please address the remaining questions I asked to clarify your criteria. I’d rather not repeat this fiasco.
newton,
What can I say? I’m an eternal optimist. One of these days an intelligent design creationist might come up with something interesting to say about reality. Today is obviously not that day.
I think a big part of our difficulty here is philosophical.
For me there are no such thing as “gibberish” strings there is only special and “undetermined at present”.
If you give me a pile of strings I can pick out those which jump out as “special” but what I can’t do is label a string as gibberish.
Or conversely If you give me one string at a time I can give you my best appraisal as to whether it is designed or I can tell you it does not meet the threshold of the test.
Think of throat cultures again. If you had a pile of 10 throat cultures you could say which throats were positive for a particular bacteria but you could not say which throats were bacteria free.
a negative on the test does not mean bacteria free it only means that the bacteria do not meet the threshold of this particular test.
does that help?
peace
there is no goalpost moving. It’s not even kick off yet.
The game is not even complete AFAIK.
We certainly have not had any agreement as to the rules of the game.
peace
What???
I’m confused I thought you just said that
quote:
being able to tell that different strings are different is uninteresting in the extreme.
end quote:
are you now once again saying that you might find such an endeavor to be worth the trouble ?
Patrick this is not difficult. You need to be able to duplicate what I just did with petrushka’s strings and do it as generally and reliably as I do.
I was given two strings that I was completely unfamiliar with and after a short trial and error period was able to distinguish between them with near 100 percent accuracy by looking for a particular pattern that is expressed in one of the strings.
What you are trying to duplicate is the human pattern recognition ability, That is what the paper is trumpeting. That is the challenge.
Your “questions” are designed to allow you to get by with something far less than you originally claimed you could do.
peace
fifthmonarchyman,
You claimed to be able to detect designed strings, to the point where you tried (and failed) with some I provided. You claimed to have used your game to detect design. Now you’re saying that all you can tell is when strings are different (hardly an impressive accomplishment) and you’re saying that the rules of this game you’ve supposedly been playing aren’t defined.
That’s a lot of goalpost moving and wasting of other people’s time.
fifthmonarchyman,
First, I haven’t seen you do anything with petrushka’s strings.
Second, my discussion of using machine learning is in the context of the Finance Game paper, not your amorphous, undefined-except-when-it-suits-you game.
Please answer the questions so that I can be reasonably confident that you won’t raise spurious objections once I have time to do the work.