Happy Fifth Birthday GAC and Mindpixel!

mindpixel writes "GAC is five today! Wow, that was fast! To celibrate, I am releasing 80,000 mindpixels with their corresponding probability of truth for research use."

Comments

celibrate?

[enrico_suave]

is that like a party without any nookie?

e.

I love /.

[mindpixel]

For five years now, more than 50,000 people have been working to make a map of common sense. The project is known as Mindpixel. It was launched on July 6, 2000. On August 24, 2000 Chris McKinstry (me) and Mindpixel were profiled by Robert X. Cringely. In September of 2000 Both Time and Wired magazines carried news of the merger of Mindpixel with the MIT Media Labs Open Mind Common Sense Project.

Now, what can you do with this data? Well, once it is in the google index - tomorrow, I suspect. Then the 3.5mb page of 80k validated pieces of knowledge will be able to do for consensus internal knowledge what wikipedia does for consensus external knowledge. I hope that eventually, google will trust Mindpixel as it does Wikipedia. Then commercial applications of semantic spectrum based technology can proceed, and the 50,000 owners of the

Re:I love /.

[TigerNut]

Google may trust Mindpixel, but I sure don't have to. Not when you're claiming that blue and yellow make green (about 0.95 probability), without qualifying that it only applies to pigments.

not a lot of comments yet

[zerkon]

everyone elses browser must be choking to death on the 80,000 lines of text like mine is...

Some help

[mopslik]

You're not alone. The GAC link leads to a minimal MindPixel front page, which reads "Digital Mind Modeling Project" and prompts me to log in. The blog link informs me that MindPixel is "a map of common sense". The 80,000 link initially crashed Firefox on my Win2K machine here at work, but on a retry, gave me a page which begins "Is ice cream cold? Is earth a planet? Is green a color?"

Fortunately, Wikipedia gave some insight. But yeah, the article summary was a bit too vague for my liking.

Many of these are inaccurate...

[Muad'Dave]

...given the vagaries of English. For example:

Is rape a good thing?

Most people would say 0%, but rape is also a type of seed-bearing plant, so rape is a good thing for getting rapeseed (canola) oil. For this assertion to be useful, there must be a way to distinguish from the plant and the crime.

In fact, 3 of the first 5 are ambiguous or subject to interpretation:

1. 1.00 is icecream cold?
2. 1.00 is earth a planet?
3. 1.00 Is it hot during the summer?

Is ice cream cold cold relative to liquid nitrogen? no.
Earth is also a collection of organic and non-organic substances that plants grow in.
Hot, relative to what? At the north pole, it's never 'hot'.

Re:Many of these are inaccurate...

[wishus]

Inaccurate, and weird.

0.97 Is Jerry Garcia dead?
0.90 Did Jerry Garcia die in 1995?
0.85 Was Jerry Garcia a member of "the Grateful Dead" before he died?
0.76 Did Jerry Garcia play guitar for the Grateful Dead?
0.32 Did Jerry Garcia have 9.5 fingers?

I don't understand how it's 32% probable that Jerry Garcia had 9.5 fingers. Does that mean that, of all the Jerry Garcias in the world, 32% of them have lost half a finger? Or that Jerry Garcia of the Grateful dead had 9.5 fingers for 32% of his life?

Re:Yeesh

[TopShelf]

You have to wonder about the quality of the data involved here, given lines like:

0.04 Can a fantasy beast can utter juniper bushes?
0.04 can you speak russian?
0.04 Will answer number 7 actually give you cheese?
0.04 Does an hour consist of 30 minutes?

Re:No they don't.

[exp(pi*sqrt(163))]

He's one of the most successful crackpots out there with stories in Wired and Time. (And /. of course.)

You don't know what this all is?

[myukew]

Fortunatly Slashdot had an inteview with the project founder

Re:What you are looking at...

[QMO]

I think that the idea is VERY interesting.

There are obvious things, that GAC is certain (or nearly certain) about.
There are relatively obsure things that GAC is unsure about.
Other things that GAC is likely to be wrong about.

It is a very interesting way to get a sample of common knowledge.

The hard part seems (to me) to be to figure out how to use it.

What does mindpixel think of slashdot?

[penguin121]

... after a quick investigation:

1.00 What is slashdot? (Huh?)
0.83 Is Slashdot actually a website?
0.77 Does slashdot postings cause extra traffic for its mentioned websites?
0.76 Is Slashdot a web site? (this one seems to vary a bit)
0.39 Is slashdot.org good?
0.35 Is the website at slashdot.org full of trolls and mindless linux bigots?
0.30 Was mindpixel slashdotted?
0.13 Is Slashdot the greatest site ever?
0.05 Has the average person (e.g. your Mother) ever heard of Slashdot?
and finally
0.00 is slashdot good journalism? (How sad)

Of course this is just a sampling of all the related mindpixel questions, but we can conclude that slashdot is a popular topic for mindpixel, but nobody mistakes it for good journalism...

Ahh, wonderful

[joto]

We now can have a 5-year anniversary of bashing a completely useless project brought to you by some internet cook who thinks he has "solved" AI by writing a program that even a 5-year old would understand is useless.

If you want a real database of "common-sense" knowledge, you should check out CYC instead. It might be harder to do it that way, but it sure pays off if you actually want to use it for something beyond spamming usenet groups and slashdot.

mindpixel youre WRONG

[mnmn]

I signed into mindpixel, and submitted my first mp. As I was typing the second one, it dawned on me that the project was not free. I looked around and couldnt find a spot to download mindpixels. Later I read that people who submit 'enough' mindpixels will be given shares in a subsequent company.

This is not only wrong, its surprising that you are posting it on slashdot of all the places. Youre planning to take public knowledge from the public, and what do you give back in return? I can come up with some algorithm, and try to parse mindpixels, but you own all the mindpixels in the first public frenzy, after which people will stop submitting mindpixels to every such database online.

'Mindpixels' should be free, and I'll wait till I see a free (GPL or otherwise) site where I can both submit and download all the 'mindpixels'. You can develop some algorithim or neural network and thats all yours. But leave the public knowledge so generously given to you in the name of science, to the public.