Privacy and Eric and Wendy Schmidt:
Okay Muppet, explain this...
Privacy and Eric and Wendy Schmidt:
You're an obsessive freak?
It probably had a lot to do with the way you were raised. Still, you should try to take some responsibility for your extremely dysfunctional thought processes.
Explain the google rank.
Gaming search engines is not new and certainly not brilliant.
Odd, that quuery returned one result a minute ago, and now it's gone...
I still get one hit with it.
Man, Google sucks. :)
No hits on 'muppet is a troll'.
August 9th, 2005
It brings up http://discuss.joelonsoftware.com/default.asp?joel.3.4600.11 for me.
That's the one it brought up for me initially, but it's now inexplicably returning nothing. It's possibly an artifact of the distributed nature of Google's index, but it's annoying whatever the reason.
THat's pretty good typing for a five year old...and considering his genetics, completely unconvincing. Nice try troll.
wow everyone, nice display of google-fu...
I think you got beaten Chris, want to try another round?
Hmmm... what do you call it when a troll trolls a troll??
"what do you call it when a troll trolls a troll?"
your tendency to attempt to dehumanize people by labelling them into boxes is kind of unnerving conner old boy....perhaps you should try interacting with actual people instead of imaginary monsters?
How in god's name is it possible to misspell a name you've only ever read?
Spelling aside Colm. He has a point. Engage the substance, not your stereotype.
What substance? He's effectively asked me why I haven't stopped beating my wife.
"What substance? "
wow, *more* rhetorical ignorance? you really should try to vary your approach a little conner.
I may very well beat my wife for all you know.
Well, since nobody else on the planet seems to be talking about "Eric Wendy Schmidt", very few items (15) refer to "Eric and Wendy Schmidt" and your site is the only one having all three words in the url ...
Just me (Sir to you)
August 10th, 2005
"your tendency to attempt to dehumanize people by labelling them into boxes is kind of unnerving conner old boy...."
And your tendency to see every case of generalisation or categorisation as "dehumanising" is fatuous.
JHC - to point out the blindly obvious you used part of Colm's surname as a first name. Kind of like calling Jesus ben Joseph just Joseph. If it helps just remember: full name required.
He likes calling people by their surname, just one of those simplistic devices to attempt to exert a non existent superiority.
And of course its objectifying.
Pot, kettle and black.
August 10th, 2005
"And your tendency to see every case of generalisation or categorisation as "dehumanising" is fatuous."
fatuous? man, that hurts.
just out of interest, in this thread, which generalisation did conner make that I called 'dehumanising" ?
"He likes calling people by their surname, just one of those simplistic devices to attempt to exert a non existent superiority."
heh. actually thats entirely not true. I called you lucy on occasion cause its a girls name and I find an immature amusement in using it to refer to a guy.
I have begun calling colm 'conner' because I did it once and then got the distinct impression that doing so annoyed him. I find an immature amusement in annoying conner.
"And of course its objectifying."
is it? god forbid that someone with the generous outlook and tolerant attitude of conner be the victim of any sort of labelling or objectifying himself.
"god forbid that someone with the generous outlook and tolerant attitude of conner be the victim of any sort of labelling or objectifying himself."
I think it's more a sort of dismayed amusement at the irony that someone who takes great pleasure in calling other people hypocrites, berates them for "immaturity", and chastises them for daring to label other people goes round doing exactly what he so despises in others.
"in this thread, which generalisation did conner make that I called 'dehumanising""
None. I was just assuming that you had a consistent view of such things; you found labelling people to be dehumanising in one thread, so my logical assumption was that you would find it dehumanising in this one too. Silly me.
>just out of interest, in this thread, which generalisation
>did conner make that I called 'dehumanising" ?
Heh. You're mirroring my argumentative style now.
heh, I might just start calling you by the collective name "mat conner"
certainly Ill respond to you both in one post.
"I think it's more a sort of dismayed amusement at the irony that <snip> so despises in others."
Im perfectly capable of being immature and labellign others. the difference between myself and people like conner (and yourself) is that I am aware when I do so.
heh. actually my point was that conner make no generalisations, instead he boxed people in with labels. they are two distinct behaviors.
your interpretation is good as well however :)
"Heh. You're mirroring my argumentative style now."
nope, you made no generalistations. you simply boxed someone in with labels. thats a *really* boring debate strategy conner.
"you made no generalistations. you simply boxed someone in with labels"
So "grouping people under one collective banner" is somehow OK, but "placing an individual in a group" isn't? An interesting proposition, and not one that I think I can understand. Is it because "the group" says nothing specific about an individual which somehow assuages your guilt at holding an opinion on the group as a whole? Isn't that *truly* dehumanising?
"So "grouping people under one collective banner" is somehow OK <snip bizarre rant> "
sorry, where did I say that again?
>nope, you made no generalistations. you simply boxed
>someone in with labels. thats a *really* boring debate
Which labels did I use that boxed anybody in?
"where did I say that again?"
When you suggested that generalising is OK.
"When you suggested that generalising is OK."
??? which I did in *which* post? sorry Im reading back but I cannot see it.
"Which labels did I use that boxed anybody in?"
you dont believe that labelling someone a troll is boxing them in?
"Im reading back but I cannot see it."
Well obviously you didn't *actually* say it, I'm just trying your tactic of making stuff up and then banging on about it as though you had said it. Did I not do it right?
>you dont believe that labelling someone a troll is boxing
I was making a humorous comment about muppet and Chris trolling each other. I don't believe that counts as boxing either of them in.
A humorous (ostensibly) statement that requires labelling two individuals as stereotypes. So yes, that's pigeon-holing people. What, do you think that explicit declarations are required? If they were, society would be in better shape.
August 10th, 2005
So are we to outlaw jokes now for fear of labelling people? Man, some people seem to have rods jammed firmly up their asses...
I just hope you start taking your own advice and stop labelling people.
I never claimed that labelling people was a negative action, Mat. You read that into my post, I didn't write it. I just said that what Colm was doing was indeed stereotyping. IE, he's full of shit. That was my only assertion. Anything else is your own projection.
August 10th, 2005
I wouldn't label you as such if you didn't attempt to conform to it yourself, muppet.
Ok, so I'm the only site with all three words...what about:
August 10th, 2005
And there are only 36 pages with those two words in the URL.
Well, it is a strange experience I am having. I am getting a hell of a lot of hits on the two queries I posted here. Very strange. I have never seen anything like it. Everyone on the planet is googling Eric and Wendy Schmidt.
August 10th, 2005
Yes, but how many have time machnes?
Ok Muppet, can you explain this query?
I think that can be safely explained by pointing out that *everyone* knows Kermit *is* a Muppet, and so that question has probably never been uttered before in the entire history of humanity. It's like me asking you to explain this:
What you're doing here is called Googlewhacking -- finding a particular phrase that only returns a single result. Dave Gorman devoted a stage show to it:
And a website devoted to the hobby:
I though googlewhacking was two dictionary words only...
And that is the point of mindpixel btw - to make explicit everything that is implicit.
Strictly speaking, yes, but this is close enough -- find an unusual or uncommon combination of words, and then Google it. The only difference is the Chrissy is reading something more significant into the search returning his site as the only result other than "it's the only place it appears"...
Oh, I have thousands of queries that return only my pages...
But, you guys should note that it is extremely interesting that the phrase "is kermit a muppet" is something everyone can agree on instantly, but because agreement is so strong, the phrase itself has essentially a zero probabilty in normal text. This is a BIG FAT HINT that mind is a high-dimensional space.
Er, no, it's more a hint that there are some questions so dumb that no-one asks them... :)
(And I'm not sure why I referred to you by name in such an odd way in my previous post. Perhaps I have a dead mindpixel... :D)
Sorry Mat. You haven't a clue. You're dead wrong.
Ok, so what *is* the significance of you having the only entry on Google for "is Kermit a Muppet"?
personally Im willing to bet that the significance is, as mat pointed out, that chrissy is the only person in the history of humanity to ask such a stupid question.
actually I agree with chrissy as well, that *is* very telling.
"to make explicit everything that is implicit."
Ok, you'd better address these gaping holes then:
...and so forth.
"is the geographic north pole north of the equator"
I *cant* believe no one has ever asked that before...
Your holes are as a result of two factors:
First, you are looking at symbolic data. If you were to train a DTW-SOM - a kind of neural network - on the public mindpixel data, the holes go away and you get a continuous space. The process involves turing each mindpixel into a vector and multiplying each vector by the meansured semantic coherence and then using dynamic time warping to normalize the vector lengths so they can be compared to each other.
Second, you are only looking at 5% of the data I have collected with your queries above.
While I'm still skeptical on all of Chris's claims, it's sure obvious he's got one heck of a random babble generator!
"the holes go away and you get a continuous space"
I'm intrigued. Does this mean that GAC *can* answer those questions?
"The process involves turing each mindpixel into a vector and multiplying each vector by the meansured semantic coherence and then using dynamic time warping to normalize the vector lengths so they can be compared to each other."
And by reconfiguring the main delflector dish to emit a polarised tachyon burst the subspace anomaly can be destabilised which will then collapse the transwarp field... :)
How do you turn a statement (or a question and its yes/no answer, which amounts to much the same thing) such as "Kermit is a Muppet" into a vector? Does that even *mean* anything? Dynamic time warping? Words (almost) fail me!
"Second, you are only looking at 5% of the data I have collected with your queries above."
I had a quick scan through that text file of 87,000 odd "facts", and you need to improve your vetting process somehow. "Is lead heavy" immediately caught my eye; it's a meaningless question without knowing how much lead you're talking about, what constitutes "heavy" (compared to a grain of sand a cat is heavy, but compared to an elephant it isn't), and under what conditions (on the moon the lead will weigh less), but somehow the answer turns out to be "Yes". You might as well ask "Is string long" for all the sense it makes.
"can you answer questions other than true or false?" is, according to the responses, answered "no", which is patently the wrong answer. "What time is it?" "3 o'clock."
As far as I can tell, all you've got so far is a very strange and unhelpful encyclopedia that answers a load of really bizarre questions in a non-useful way. Saying you "turn them into vectors" makes no sense that I can ascertain, and how do you measure "semantic coherence"?
Without understanding what the whole time warping thing is, and what you mean by vectors, I'm guessing (probably incorrectly) that you're attempting to construct a semantic net that will allow you to deduce a fact by following a bunch of axioms, so from the statements "All humans are warm blooded" and "I am a human" you could deduce the statement "I am warm blooded". However, without hand-building all the connections you could equally end up with the statement "I am all humans".
I recall several years ago reading about a similar project (which may actually have been your previous attempt -- the details are sketchy in my mind) that was essentially a failure as, although they had amassed a great number of axioms, it turned out to be a false assumption that merely by connecting a bunch of facts you could actually gain any "understanding" -- it requires context to understand, and understanding to get context, and knowledge alone couldn't bootstrap the process.
I believe the AI "problem" (both strong and weak) is (at least theoretically) solvable, but taking a brute force "lots of data" approach doesn't seem like it could ever work; you could probably generate lots of new axioms this way, and maybe even come up with some heretofore unknown facts, but it's never really going to amount to much more than that, and certainly never come anywhere *near* approaching your goal of teaching a computer "what it means to be human".
Still, there's very little concrete information available on your site as to how it works or what, really, it's supposed to do, so perhaps your meaningless-sounding jargon isn't as meaningless as it seems and you're on the verge of something special. I do, however, *seriously* doubt it.
"I *cant* believe no one has ever asked that before..."
Due to the limitations of Google it turned out to be quite hard to find questions that returned no results. Even "Is a circle circular?" gets hits, but they're all "[something]...is a circle. Circular...[something]" -- I hate that you can't filter out or search for punctuation in Google, but I guess skipping it makes indexing a whole lot simpler...
"but taking a brute force "lots of data" approach doesn't seem like it could ever work"
I think thats a rather pessimistic outlook. clearly once they have amassed a great heap of data, all they need to do is invent an artificial intelligence and they will have an artificial intelligence with access to a great heap of data.
*much* more useful than an artificial intelligence *without* access to a great heap of data :)
"DTW is an algorithm for comparing a sequence of observations with a reference or a template sequence "
" In DTW-SOM each unit in the map is associated with a reference sequence, and DTW algorithm is applied for searching the bmu. The flexibility of the DTW matching allows comparing sequences of different lengths, which makes it appropriate for speech processing. Hidden Markov models were applied in the state space map (Somervuo, 2000) to replace the Euclidean distance operators in the map units. Here a trajectory of bmus corresponding the input vector sequence is searched by Viterbi algorithm. Each input vector is then associated with certain bmu, which represents one of the HMM states. "
interesting, so not a neural net at all, more a way of breaking semantic data down into a searchable format.
<g> nothing about a magical ability to fill in gaps however, perhaps thats the *original* research that Chris has done in this area.
True, true. :)
I think at the very best you might get some kind of weak AI out of it, but as that's little more than an encyclopedia with automatic cross referencing it's not going to "understand" anything. Still, a weak AI with a massive scope would be quite an achievement; limited-domain weak AIs are ten a penny these days, but trying to extend them further has so far resulted in a series of disappointments...
For a strong AI I think a totally different tack is required -- some sort of modelling of the low-level processes in a brain with some "fast track" evolution seems to be the best approach, and there has been some success with very small scale experiments. The current state of the art is reckoned to be on the same intellectual level as a slug, though, so although some people on ?off might start being overtaken soon humanity as a whole is relatively safe. :D
What he's going to do it put a mindpixel terminal out in a thunderstorm and wait for lightning to strike it. That is, of course, how to invent AI.
Has anybody here read CS Lewis' _That_Hideous_Strength_? That book had some freaky ideas in it. They took the freshly severed head of a serial killer and hook it up to life sustaining medical equiptment. Then they cut the top of the head off and pumped some kind of steriods or stimulants into the brain so that it grew large out of the top of the head. Somehow this also caused it some kind of uber-sentience capable of communicating with what the scientists believed where simply higher forms of life. In reality though, they were demon spirits. Since it was a head, the interface was simple, you talked to it, and it talked back with some kind of freaky mechanical vocal chords. Some twisted stuff.
Ive often wondered how it would be best to approach creating an AI, I can kind of imagine an 'info stream' type setup with an arbitary number of bots watching it and pulling whatever looks interesting to *them* out of it.
At the other end there would be a 'reaction stream' up which the response would flow, and bots would also watch that checking and rechecking the response against their specific set of interests.
after that you would need some way to prioritise the response bots so that they knew which could override which, and some feedback loop so that every bot got a chance to look and finally approve the response.
then you just add bots for each stimuli you wanted to grok, and each stimuli/response combo you wanted to handle.
I haven't yet decided whether the stimuli bots and the response bots should be one and the same thing.
would be *really* fun to do....one day Ill write myself my own Rick Tang...