I bet they were visiting to learn how to wire thalami.
I bet they were visiting to learn how to wire thalami.
Forget it. I cite the Muppet precident. If you don't want to see me squirt. Don't look.
Mark dude. You need to learn to use google.
This is how you ask google a binary common sense question:
If you parse this query correctly, the response to: is "chris McKinstry" a moron is 0.39.
add this query:
Probability of coherence is 0.34 for this one.
Sum the two and divide by two and you arrive at the human common ideal measure of my globally precieved hostility.
You have enough information now to answer any common sense question with a little google soap script. Have at it.
Note no probability is returned... so you are seeing your own mind only. Hope you are enjoying it. Doesn't look like a nice place to even visit. Not worth the femoral nanowire.
And what does no probability for this query say?
I dont get wtf you are talking about. And what does it have to do with google?
Aaron F Stanton
August 18th, 2005
Mark. Restrict your query like this:
34 results returned, so sum the Ps and divide by 34.
Aaron, you are pulling back too much:
One hit. P=0.86.
Mark, yeah, I scanned that thread, but it does'nt cause these google links to make more sense.
Sorry Mark. Divide by 33. That query needs an n-1 term.
The answer is P=0.44, BTW.
That's because he's a complete quack. Take a bunch of assertions from his GAC list or whatever that happen to have terms in them common to your search (even if those assertions are semantically COMPLETELY DIFFERENT) and then add up all of his "coherence" scores and average them.
It's utter nonsense, of course. It's a perpetual motion machine, only in this case you can clearly see Chrissy moving his balls manually.
Eric. Google the phrase: Does a giraffe eat large chunks of cement for breakfast?
Or click below:
Do it you stupid puppet, instead of just attacking it. Prove it doesn't work. Did I mention I think you're an ass? Or don't you know soap?
Shhhh you can't waste time on me. Concentrate on manipulating your balls.
A little code mophead and you can have a probabilistc human proposition evaluator that actually works.
You are a google key and a few lines of code away from something amazing and you're such a dick you're missing it.
"Prove it doesn't work. "
maybe you should start by proving that it does?
"whats a "human proposition evaluator"?"
*clearly* its an evaluator of human propositions.
what a stupid question.
Allow me to summarize for Chris:
"In order to prove what a clever fellow I am, do this google search that results in a page from **my own website**, and there are numbers there, suggesting that I am in fact very clever."
hmm, google doesnt work for me right now because the 12 pr0n torrents I'm downloading has saturated the DNS proxy in my NAT rounter.
I dont know why this thread made me think of having a wank..
Yeah... what's the point of running everything through google anyway? Just to get results from your own site? I just don't get it...
The point of using Google is that it has a copy of everything I have made public. I would rather use their infrastructure, and so would everyone else.
And the data was made by 50,000 people. It was a hell of a lot of work and took five years.
What is public is 80k propositions, each with an average of 20 human evaluations to create the semantic coherence measure. That is nearly 2 million human votes. It is the best global statistical model of human common sense you can access on the planet.
I can access nearly 2 million propositions. They are for sale or rent. I kick back 70% to the people who made the data and keep 30% for my own entertainment.
The point is to impress us all with how his site comes up first in a Google search. Nevermind that it includes a made up word first in the search, or anything.
fap fap fap
I'm impressed at this, I hadn't realised one could go to such a lot of effort to produce something that was semantically null.
No you bonehead, it is not my page rank I am pointing out. You're missing the point. Who has more computers than Google? Only one entity and its name is in the subject of this thread.
Perhaps he is missing the point because you never come out and say what it is? Tell us. What IS the point of your endevour?
What he thinks he's doing is discovering the zeitgeist.
No no no. How many hits on the corp.nsa.gov query?
Look at them.
Another site that has recently had the NSA all over their shit:
"That is nearly 2 million human votes. It is the best global statistical model of human common sense you can access on the planet."
0.80do you have guts?
Umm yeah.. that's all really insightful information in there. I just don't get the point.
I don't know if I've acquired immunity or something but for some reason I don't have an irresistable urge to follow links that aren't in themselves intrinsically interesting.
I fail to see as well why I need to inflate the logs of a site to stroke the ego of the person that runs it.
17 results.. mindpixel is on top. The rest are public IP logs. There would probably be alot more if not for the fact that you usually block off ip logs, or at the very least prevent them from being indexed.
What is the point? Give us an example of how this could be useful or something...
Chris, maybe I don't understand the breadth of your idea, but you're a stupid idiot.
Throwing a million sentences into google, and thinking that searching though them with keywords will somehow "digest" the query and come back with a valid answer?
August 18th, 2005
Project Gutenberg is also a multi-dimesional semantic motivation evalator.
What we can establish so far:
A) Chis claims his has a magical machine that can do wonderful stuff, but he wont:
___1: tell us how it works
___2: tell us what it is good for
B) When pressed Chris will respond with schizoid word sallad and citing impressive numbers of people.
C) Chris thinks the NSA is after him or his invention.
Am I getting this right?
He uses words that purport to explain the significance of what he does, most often those words are placed haphazardly within a sentence that usually begins with the word 'I'.
Rick is a far better written bot.
Because if we don't inflate the logs, we'll lose the opportunity to laugh at him. On the other hand, I'm actually kinda depressed now, because he's not actually a creative and original lunatic, he's just sad and obsessed. Kinda like an 80yr old granny who's convinced that Dallas was a documentary and still cries on the anniversary of JR being shot. Sure, it's funny, but really you have to wish that something could be done to help the poor dear.
Somewhere on the planet there is a reality show craving Christopher McKinstry to be on it. I just hope it isn't anywhere within my area though.
"Do it you stupid puppet, instead of just attacking it. Prove it doesn't work. Did I mention I think you're an ass? Or don't you know soap?"
that you claimed returns a value of 0.39 for the question "is Chrish McKinstry a moron" returns that value for a completely different bunch of questions, dipshit. All it tells us is that somewhere on the page of 0.39 results your name and the word moron appear. (The closest question on there is "is Al Gore a complete moron".)
If you can't even make any kind of "semantic coherence" assessments then what possible hope do you have of getting a machine to do it?
Now go away, you silly little man.
Wow, Mat, you visited his site? You're "all over his shit."
August 19th, 2005
That's essentially what I already said above.
August 19th, 2005
I've spent a bit of time poking round his site looking for any kind of actual explanation of any of his dubious claims, and it just seems to be a slightly dull version of the "rate my [foo]" family of sites; a bunch of 'tard questions, rated between 1 (completely true) to 0 (completely false) depending on the answers of a bunch of people.
Still, it answers some of those age old questions:
It's utter nonsense, and when pressed Chrissy spouts a bunch of Star Trek style tech-sounding nonsense.
August 19th, 2005
With questions like "Will GAC serpase its creators intelagence?" how can it possibly fail?
I *do* like this one, though -- "Will you answer "False" to this question?" Perhaps, if Chris ever gets it to work as planned, it'll stumble across this and explode in a Star Trek "Kirk destroying evil computer with paradoxical statement" style... :)
Yeah, feeding ?off into Mindpixel.
Would return a 100% probability.
"Mindpixel. The Hot or Not of symantic null statements."
August 19th, 2005
I think that the ESP game and the Peekaboom game are much more compelling and interesting methods of teaching computers about stuff, but hey, what do I know. I'm just a briefcase-size expert system anyway.
Can we somehow ban him from this place? I know I can choose to not look at his posts, but geez, he really is making ?off look bad. And that fact, really shows how fucking psycho this guy is. Man, I'm really concerned about his guy, he needs serious help. Just because the NSA visit's your site doesn't mean shit. You do realize there are Janitors and Secretaries that work inside the NSA building and can stumble upon webpages just the same as real researchers? I mean, I used to work at a thinktank for weapons research and I visited surfline.com all the time, what correlation does that have.. exactly, none. I hate drugs, but you Sir are why they exist. Please start taking some.
He's not that bad. A bit of a nutter, sure, but harmless enough...
So Jared, you think the NSA's cleaning staff is interested in my work? Sounds pretty unlikely.
And you hate drugs? Really? All of them? Insulin? Antibiotics? You must have a great deal of pharmacological experience that you're not telling about. Or, you have no idea what you are talking about. I vote for the latter.
Ok. So if you write the soap script I suggested, you can get google to respond in a human-like fashion to arbitrary propositions. Really. You can do a controlled experiment and prove it to yourself or prove me completely full of shit. Either way, you gotta write some code.
Give the script a proposition and give the same one to a pool of twenty people. If I am full of shit, you will not be able to detect any statistical relationship between the responses generated from my data and the responses of the human pool. If I am not full of shit, there will be a correlation between the human and machine generated responses. That has never been possible before. Even CYC can't respond to arbitrary propositions.
Now, of course you're working with only 5% of the data I collected, but the global statical relationship is there even in the small sample. Just look at the proposition distribution graph:
And if you are smart, you will see in the amazing symmetry of the data that every brain state is really two brain states: proposition and anti-proposition each 180 degrees out of phase with the other. But I don't expect anyone that smart here.
If you want to see something really remarkable, train a DTW-SOM on the data I made public and get a continuous response hypersurface that works much better.
Or just ignore me. Or even attack me if that's your style. Whatever. I'm easy.
"So if you write the soap script I suggested..."
Back up the truck; I have to write some code to prove your not a nut? Why don't you write the code and then prove your not a nut. Why do I need to do it. Shouldn't your work stand on it's own.
"You can get google to respond in a human-like fashion to arbitrary propositions."
If you parse this query correctly, the response to: is "chris McKinstry" a moron is 0.39."
mean anything at all? Seriously. How does it mean anything. Can you answer that in plain English? Why wouldn't it be 0.25 or 1.0?
And on the 'Off' chance someone here at Off is interested in the symmetry in my mindpixel data yuo can look at my symmetry graphs here now before they start showing up in textbooks:
My work stands on its own. But, if you want to prove something, you've got to do it yourself. It is such a little script, what's the big deal? I'd give you my code and my google key, but there are reasons I can't get into for me not doing that. And obviously, I don't want you morons eating my finite number of queries. Use your own.
And no, it does not parse to 0.39. You need to filter out the non GAC-80K pages and sum the probabilities from all the returned pages to get a first order approximation of a human response. It is less than 0.39.
And why is it what it is? Read some Riemann on extrinsic distance measures in arbitrary high-dimensional spaces.
"My work stands on its own."
That's what you believe. Doesn't seem any of the smart people around here agree with that.
"But, if you want to prove something, you've got to do it yourself."
Ummm.. so you're saying that you don't have to do through any scientific process? Einstein never said "E=mc^2 -- but you've got to prove it yourself!"
"You need to filter out the non GAC-80K pages and sum the probabilities from all the returned pages to get a first order approximation of a human response. It is less than 0.39."
Lets assume the value is 0.25 for arguments sake -- what does that mean? What's the significance of that number? Again, please, in some kind of plain English.
If it were .25 it would mean that if you took a pool of twenty random people with demographics similar to that of the 50K people who handmade my corpus, and give them the same proposition, you would get the same number in their composite response.
And of course I believe in the scientific process. It is just slow. You can hack some code today and see something really neat, or you can wait until it is patented and published.
Ok, ok. I sorta got that far.
Does it work on a proposition like "Chris McKinstry is a moron" even if that statement doesn't appear in the GAC-80K data? If so, how is that anything other than statistical noise?
The proposition isn't in the database. But pieces of it are. And with only 80K propositions, it is quite noisy, but there is a global pattern there. If you do my little code exercize, you can try it out on aything you can think of and you will see, thought it misses many things due to the smallness of the sample, it does surprising well due to the broadness of the sample.
The big problem is with symbolic pattern matching. It is much better to convert all the propositions to vectors and compare vectors. Spelling is a bitch otherwise. But to do that, you will need to copy everything I made public, which is easy because I put all 80K propositions in a single html file. And then convert the strings to vectors. Which isn't really hard. The easiest way is to use a 26 dimensional vector for each word where there is one bit for each letter. Smarter is to do so with the awareness of the actual statistics of English.
Anyway. You are a tiny amount of soap code from something amazing that was impossible before July 6, 2005.
"The proposition isn't in the database. But pieces of it are."
No a bunch of text is in the database; there are no pieces of that proposition in the database.
Our example, which is now:
"Chris McKinstry is a moron" = 0.25
The value 0.25 comes from this statement having some textual match to other statements (specifically those that contain the word "moron"). What's the intellectual value of that?
"it does surprising well due to the broadness of the sample."
But what it is doing suprisingly well? Is it guessing answers, is that the point?
I've had a slow day today, so I've been giving this a lot of thought. Pointless technobabble aside, the notion of a large bunch of true/false axioms providing enough data for a machine to surmise answers to questions that are not explicitly stated anywhere in the data is not an impossible one. It's non-trivial, the data would have to be totally consistent, and the mapping of the axioms to each other to allow a series of them to be combined to produce a new one would be mind-bendingly difficult. Still, it's not entirely science fiction. (I'm inferring from his repeated use of the word "vectors" that he somehow believes "information space" can somehow be used as "real space" and that individual bits of information can be expressed as vectors through the space; it's an intetersting idea, but seems unlikely to amount to anything more than sounding clever.)
However, the notion that a Google search turning up matches because of bits of common text proves anything is just idiotic, and Chris suggesting it does prove something damages any shred of credibility he may have had. It's irrelevant, misleading, obviously and demonstrably wrong, and just plain insane.
I have a small amount of experience in writing natural language parsers and inference engines -- nothing on the scale of Cyc or GAC, but within a small domain its not too hard to do. What Chris is attempting to do (or, in fact, what he claims to have done) is so unbelievably far in advance of anything currently available that I find it odd he chooses to hang about on here spouting nonsense to a bunch of random people who find him mildly amusing in a crackpot sort of way rather than going out and demonstrating his earth-shattering breakthrough to the world. (Well actually I *don't* find it odd; it matches perfactly the behaviour of so many peddlars of nonsense -- the dowsers, psychics, spirit photographers, and so forth.)