I was active on comp.ai at the time Minsky made his offer [Google query], and I'm convinced the real reason academic AI hates the Loebner Prize is that it shows up how little they've managed to accomplish.
I agree that the entries are really bad-- one recent winner just said the same things no matter what the human asked. But one winner, unmentioned in Salon, was Thom Whalen, whose design was a genuine advance in the art. (Regrettably, Loebner changed the rules to exclude his approach in the future.)
What Whalen did was limit his domain to one topic, and compile a set of general answers to likely questions, which he matched by spotting keywords. So even if the answer wasn't a perfect match, it was general enough to be useful. This design should be better known and more widely used, and the Loebner contest would have been a good launchpad to bring it to people's attention if the academics weren't so prejudiced.
But the top academics get six-figure salaries for generating lots of jargon and no useful products, so a level playing-field is the last thing they want.
I've been blogging for 5+ years, and have evolved my routine into a system. Almost all of it is summarised in three rows of links at the top of my weblog-- top row for weekly visits, middle row for daily visits, and bottom row for continual updates.
The links are just abbreviations, so you have to explore to discover what they mean, but the advantage to this is that I can cite the abbreviation easily each time I link a story found via that source.
The idea of putting them in rows at the top is so that frequent visitors to my blog can jump to other sources if they don't find anything new/interesting at mine. (I call them 'jumpbars'.) Lately I've started adding little asterisks for sources that have recently done especially noteworthy updates.
My local startpage duplicates the jumpbars, and adds less-frequent sources like monthlies. When I started blogging I made a serious effort to learn the update schedules of every online periodical, and I created a generic startpage that summarised these. (It's badly out of date now.) The idea was to encourage people to copy this page and customise it to their interests. But knowing when zines usually update makes it easy to prioritize my surfing-schedule. (I wish all periodicals spelled this info out on their front page, eg The Onion comes out late Tuesday.)
I think NewsHub still isn't appreciated for its headline-aggregation pages. I'd use NewsLinx too except that most all the tech zines have decided to use obnoxiously junky html-design, so I stick with Slashdot and the Register for tech news.
My politics are lefty, and Sam Smith's Progressive Review gives a very deep daily summary with links, while Common Dreams reprints full articles from many major sources. A newcomer is Memory Hole that specializes in stories the mainstream media tries to suppress/ignore.
For space news, NasaWatch is tops. I've mostly given up on Drudge and Salon, and am having doubts about the BBC science page.
Other daily faves include the AstroPic of the Day, two poem-of-the-day sites, Zippy the Pinhead, and various blogs. A weekly that I think is underappreciated is Dean Baker's Economic Reporting Review that gives a very dry weekly critique of economics-propaganda in the NY Times and Washington Post. (They very systematically distort the facts with the obvious goal of redistributing the wealth upwards.)
Umm... o_O... well, that's what I call "security by obfuscation":-) (no offense meant).
Some offense taken. >:^(
You should really read the tutorial before spouting off what you think you know.
Simpler explanation for the people who want to understand it
Not simpler, and mostly unrelated to the tutorial I was trying to summarise. (I obviously wasn't trying to substitute for the tutorial, as you are, I was trying to suggest what was interesting about it, which you can't.)
What caused the anisotropies?
The whole point of the tutorial is explaining what regularities they're able to extract info from. You obviously don't know anything about that.
The AstroPic of the Day this morning includes links to Wayne Hu's accessible tutorials on the CMB: beginners (12pp) and intermediate (much longer, I think).
Short version, as best I've understood so far:
During the earliest expansion after the Big Bang, the attraction of gravity was counterbalanced by the pressure of photons, with slight fluctuations that echoed thru this superdense plasma as 'sound'.
The resonant frequency of the sound was limited by the speed of gravity (ie, speed of light) and the spatial 'horizon' it could reach over the course of the universe's short life. (Harmonics also arose at some point.)
When the plasma cooled enough for atoms to form, all the photons were released at once, in a pattern that retained the resonant-frequencies fluctuations, and that's what WMAP is measuring.
Re:Slashdot is aiming way too low
on
Infinite Games?
·
· Score: 1
A new bundle of papers is provided by the Mimesis / Liquid Narrative crew, and they're not linked from your website yet
As I said, my freshest AI-page is the timeline, which does include links to the PDFs.
I only skimmed one pdf, because the format is awkward, but the basic idea seems just like my confabulating arranger proposal.
Re:Slashdot is aiming way too low
on
Infinite Games?
·
· Score: 1
It's polite to wrap posts like that with a shameless plug tag.
As my subjectline indicated, the first paragraph was my message. But it would have been irresponsible of me to claim the BBC article was content-free without offering something for contrast, and (despite what another Slashdot followup suggests) my FAQ is the only site I know that treats these topics for a popular audience.
Also it helps to at least paste in a few lines from your site which are particularly relevant. Otherwise, your few visitors will immediately flee when they see a bunch of links to LISP, Mania, and Cyc, with no apparent attachment to video games.
It's an AI faq. If you read thru the brief summaries, you should be able to skip to the parts that interest you most.
"an insult to Slashdot's readership" You must read at +5 or something. That's a drastic overestimation.
Slashdot posters run the gamut, but the average is much more tech-literate than for the BBC site.
We're not looking for "perspective", background, or topics of academic interest. People are interested in results.
1) You don't speak for anyone but yourself. 2) The BBC article doesn't cite any credible results-- academic AI has a long history of hollow demos. 3) Perspective is exactly what Slashdot readers lack.
Those few who enjoy abstract theory already know where to find it
This is highly illogical. Most Slashdot readers enjoy some level of abstract theory, but almost none know where to find presentations about story-AI at an introductory level.
until then, every concrete step along the way will be interesting
You say that as if the BBC article offered one, but at best it offered a link to some PDFs that claim to be one.
Slashdot is aiming way too low
on
Infinite Games?
·
· Score: 5, Insightful
This article is a thumbsucker for technical illiterates, and an insult to Slashdot's readership. It gives zero insight into game design that hasn't been an industry cliche since the glory days of Infocom (20 years ago). Self-promoters make these boasts on a weekly basis, so Slashdot editors should know enough to refuse to link stories unless they include usable new content.
The problem that 'Liquid Narrative' is addressing goes back at least to George Polti's "36 Dramatic Situations" in the year 1900. My AI faq gives infinitely more perspective than this BBC pap, on the important questions. (It's getting a little stale, but I'm currently revising the timeline with lots of rich resources.)
Slashdot has really turned into Beavis-and-Butthead lately. Before taking down archive.org-- which struggles with demand at the best of times-- someone should have checked that:
1) there's basically nothing to download, and never will be
2) 99% of the 10k titles are utter junk anyway
Hey guys-- get off your millionaire butts and fix this broken resource.
Modulating intonations is part of the larger challenge of natural-language processing (NLP, a subdiscipline of AI). We simply don't have the sort of general theory of language-production that could systematically predict how the intonations should fall, any more than we have a theory of translation that can do substantially better than Babelfish.
Nor, to harp on my pet peeve, do we have a theory of semantics that can put XML to any important use on the average webpage. These all need a model of the human psyche, because all human language is flavored with metaphors from the realm of motives and plans, etc (the psychological realm). Psychological science isn't delivering the sorts of models that NLP-etc need, and probably won't for many decades yet. [My AI FAQ]
I think you'd lose more than you'd gain if you
tried to centralise this process-- it's
hard enough to keep a local webpage up-to-date.
I agree in theory that we need a Semantic Web
where content is easier to find, but I don't think
XML-etc can really help. [rant]
My current theory is that individuals need to
build the 'Necessary Web' which consists, like
an encyclopedia, of a page for each topic (or
many pages by different authors, on their own
websites). Four special traits make a page
qualify as 'Necessary':
-- an attempt to be FAQ-like, and briefly cover
all the important subtopics on a single page.
-- an attempt to sort thru and link all the best
web-resources on the topic. (By reducing the
linktext to one- or two-word [text buttons] you can fit hundreds of links into
a useful page.)
-- a timeline, to present the most possible data
in the neatest possible way. [theory]
-- The Open Web Content License to
encourage others to recycle-and-update your
content, requiring only that they clearly link
your page as one of the original sources.
Most recent example of this format: Linux/Unix (timeline w/100s of links)
I believe that once a critical mass of authors
adopt this format, taking on the most useful
topics, there will be a rapid shift from the
current search-frustrations to something very
much like the Semantic-Web ideal, without even
requiring any fancier technology than simple HTML.
"MS's promises have nothing whatsoever to do with 'understanding' the semantics of a letter to your girlfriend or whatever and expressing your sentiments as an XML tree. If you think it is, you have failed to understand the article!"
This is just the old XML bait-and-switch again. Bray writes of "all sorts of wonderful new things [that] can be invented". TimBL touts the Semantic Web as the immediate justification of XML.
"It is not an attempt to mark up semantics, it is an attempt to convert things like bold, italic, font size into XML representations."
This article is pure PR, with no new content. The XML-cult will keep waving their hands and promising great payoffs 'RSN' (real soon now) until people actually start trying to implement uniform semantic tags in their data and documents... at which point universal disillusionment will set in because the problem is way too hard even for trained AI-PhDs. [more]
The thread a couple of weeks ago about the death of META headers will apply 1000 times worse for semantic tags-- if the semantic web is going to work at all it needs to start from headers describing the webpage as a whole.
(Also, what's with XML-Journal's claim the article has three pages when it only has two?)
I wasn't paying attention when Ziff-Davis
spun off ExtremeTech, and the first review
I read there was for Lindows, last week.
But just based on that one and this one, I have to
wonder if they're too commercially
compromised to deserve a straight Slashdot
link, without some caveat...?
"A good basic guide from the founder of the web is at http://www.w3.org/DesignIssues/Axioms.html (Note that URI doesn't follow its own advice, heh)"
What's more, his 'good basic guide' is written by a PhD who's completely lost touch with the 99.999% of web-authors who don't have PhDs in compsci or physics or anything. The W3C has become a big irrelevant exercise in intellectual masturbation, and responsible web-authors ought to secede from its pseudo-AI lunacy.
My site is 100% static HTML, but my rules of thumb for URLs include:
- never more than 80 chars, so they can be emailed without wrapping
- no uppercase, ever (otherwise you'll forget where the caps were)
- never more than two directories deep (I sometimes break this due to bad planning)
- if a new page seems likely to grow into many pages, it should be created as foo/index.html instead of foo.html (Someone emailed me this brilliant tip, I forget who though.)
But the bottom line is to arrange directories and files (and their names) so that you can remember them without having to doublecheck.
It seems to me that the usual choice is to keep
treating it like highschool until you have your
degree-- do what you're told, and your income
will reward you.
As soon as you swerve from this narrow path of
conformity, your odds of getting a degree (and
that bigger income) start to plummet... but at
the same time your likelihood of finding meaning
in your life starts to escalate.
I could only take a half-year off without losing
my scholarships, but my willingness to toe the
line plummeted anyway, and I dropped out after
a year, then went back a couple more times to
other schools, but never got the 4-year degree.
I've rarely sought jobs where a degree would have
mattered (eg programming, especially) but I've
also had to get used to life being a constant
struggle, materially, in exchange for a fantastic
advantage in freedom of thought.
Slashdot Mirror
I agree that the entries are really bad-- one recent winner just said the same things no matter what the human asked. But one winner, unmentioned in Salon, was Thom Whalen, whose design was a genuine advance in the art. (Regrettably, Loebner changed the rules to exclude his approach in the future.)
What Whalen did was limit his domain to one topic, and compile a set of general answers to likely questions, which he matched by spotting keywords. So even if the answer wasn't a perfect match, it was general enough to be useful. This design should be better known and more widely used, and the Loebner contest would have been a good launchpad to bring it to people's attention if the academics weren't so prejudiced.
But the top academics get six-figure salaries for generating lots of jargon and no useful products, so a level playing-field is the last thing they want.
The links are just abbreviations, so you have to explore to discover what they mean, but the advantage to this is that I can cite the abbreviation easily each time I link a story found via that source.
The idea of putting them in rows at the top is so that frequent visitors to my blog can jump to other sources if they don't find anything new/interesting at mine. (I call them 'jumpbars'.) Lately I've started adding little asterisks for sources that have recently done especially noteworthy updates.
My local startpage duplicates the jumpbars, and adds less-frequent sources like monthlies. When I started blogging I made a serious effort to learn the update schedules of every online periodical, and I created a generic startpage that summarised these. (It's badly out of date now.) The idea was to encourage people to copy this page and customise it to their interests. But knowing when zines usually update makes it easy to prioritize my surfing-schedule. (I wish all periodicals spelled this info out on their front page, eg The Onion comes out late Tuesday.)
I think NewsHub still isn't appreciated for its headline-aggregation pages. I'd use NewsLinx too except that most all the tech zines have decided to use obnoxiously junky html-design, so I stick with Slashdot and the Register for tech news.
My politics are lefty, and Sam Smith's Progressive Review gives a very deep daily summary with links, while Common Dreams reprints full articles from many major sources. A newcomer is Memory Hole that specializes in stories the mainstream media tries to suppress/ignore.
For space news, NasaWatch is tops. I've mostly given up on Drudge and Salon, and am having doubts about the BBC science page.
Other daily faves include the AstroPic of the Day, two poem-of-the-day sites, Zippy the Pinhead, and various blogs. A weekly that I think is underappreciated is Dean Baker's Economic Reporting Review that gives a very dry weekly critique of economics-propaganda in the NY Times and Washington Post. (They very systematically distort the facts with the obvious goal of redistributing the wealth upwards.)
Remember his whitewash of Palladium?
Well, aren't we special! (You were supposed to apologise for acting like a clumsy ego-geek.)
What you don't know, but think you do, is Hu's main topic-- the anisotropies as reflecting the resonant frequencies of the 'primordial sound'.
Some offense taken. >:^(
You should really read the tutorial before spouting off what you think you know.
Simpler explanation for the people who want to understand it
Not simpler, and mostly unrelated to the tutorial I was trying to summarise. (I obviously wasn't trying to substitute for the tutorial, as you are, I was trying to suggest what was interesting about it, which you can't.)
What caused the anisotropies?
The whole point of the tutorial is explaining what regularities they're able to extract info from. You obviously don't know anything about that.
Short version, as best I've understood so far:
During the earliest expansion after the Big Bang, the attraction of gravity was counterbalanced by the pressure of photons, with slight fluctuations that echoed thru this superdense plasma as 'sound'.
The resonant frequency of the sound was limited by the speed of gravity (ie, speed of light) and the spatial 'horizon' it could reach over the course of the universe's short life. (Harmonics also arose at some point.)
When the plasma cooled enough for atoms to form, all the photons were released at once, in a pattern that retained the resonant-frequencies fluctuations, and that's what WMAP is measuring.
As I said, my freshest AI-page is the timeline, which does include links to the PDFs.
I only skimmed one pdf, because the format is awkward, but the basic idea seems just like my confabulating arranger proposal.
As my subjectline indicated, the first paragraph was my message. But it would have been irresponsible of me to claim the BBC article was content-free without offering something for contrast, and (despite what another Slashdot followup suggests) my FAQ is the only site I know that treats these topics for a popular audience.
Also it helps to at least paste in a few lines from your site which are particularly relevant. Otherwise, your few visitors will immediately flee when they see a bunch of links to LISP, Mania, and Cyc, with no apparent attachment to video games.
It's an AI faq. If you read thru the brief summaries, you should be able to skip to the parts that interest you most.
"an insult to Slashdot's readership" You must read at +5 or something. That's a drastic overestimation.
Slashdot posters run the gamut, but the average is much more tech-literate than for the BBC site.
We're not looking for "perspective", background, or topics of academic interest. People are interested in results.
1) You don't speak for anyone but yourself. 2) The BBC article doesn't cite any credible results-- academic AI has a long history of hollow demos. 3) Perspective is exactly what Slashdot readers lack.
Those few who enjoy abstract theory already know where to find it
This is highly illogical. Most Slashdot readers enjoy some level of abstract theory, but almost none know where to find presentations about story-AI at an introductory level.
until then, every concrete step along the way will be interesting
You say that as if the BBC article offered one, but at best it offered a link to some PDFs that claim to be one.
The problem that 'Liquid Narrative' is addressing goes back at least to George Polti's "36 Dramatic Situations" in the year 1900. My AI faq gives infinitely more perspective than this BBC pap, on the important questions. (It's getting a little stale, but I'm currently revising the timeline with lots of rich resources.)
1) there's basically nothing to download, and never will be
2) 99% of the 10k titles are utter junk anyway
Hey guys-- get off your millionaire butts and fix this broken resource.
Alan Kay, Doug Englebart, Will Wright, Chris Crawford, Doug Lenat, Jay Forrester, Ivan Sutherland
NASA should hire Chris Carter to plant fake clues, and build it up into such a wacky, all-inclusive conspiracy that it collapses from its own weight.
We Americans just call it the 'Quarter Pounder'.
While analysing the pacing of internal-monologs in Joyce's Ulysses, I calculated that 70 years of transcribed thoughts should fit in 37 gigabytes.
Nor, to harp on my pet peeve, do we have a theory of semantics that can put XML to any important use on the average webpage. These all need a model of the human psyche, because all human language is flavored with metaphors from the realm of motives and plans, etc (the psychological realm). Psychological science isn't delivering the sorts of models that NLP-etc need, and probably won't for many decades yet. [My AI FAQ]
I agree in theory that we need a Semantic Web where content is easier to find, but I don't think XML-etc can really help. [rant]
My current theory is that individuals need to build the 'Necessary Web' which consists, like an encyclopedia, of a page for each topic (or many pages by different authors, on their own websites). Four special traits make a page qualify as 'Necessary':
-- an attempt to be FAQ-like, and briefly cover all the important subtopics on a single page.
-- an attempt to sort thru and link all the best web-resources on the topic. (By reducing the linktext to one- or two-word [text buttons] you can fit hundreds of links into a useful page.)
-- a timeline, to present the most possible data in the neatest possible way. [theory]
-- The Open Web Content License to encourage others to recycle-and-update your content, requiring only that they clearly link your page as one of the original sources.
Most recent example of this format: Linux/Unix (timeline w/100s of links)
I believe that once a critical mass of authors adopt this format, taking on the most useful topics, there will be a rapid shift from the current search-frustrations to something very much like the Semantic-Web ideal, without even requiring any fancier technology than simple HTML.
The original meaning of the term 'weblog' was a log of (my) web-reading/recommendations. I don't really approve its generalisation to include diaries.
This is just the old XML bait-and-switch again. Bray writes of "all sorts of wonderful new things [that] can be invented". TimBL touts the Semantic Web as the immediate justification of XML.
"It is not an attempt to mark up semantics, it is an attempt to convert things like bold, italic, font size into XML representations."
No, you are simply wrong.
The thread a couple of weeks ago about the death of META headers will apply 1000 times worse for semantic tags-- if the semantic web is going to work at all it needs to start from headers describing the webpage as a whole.
(Also, what's with XML-Journal's claim the article has three pages when it only has two?)
But just based on that one and this one, I have to wonder if they're too commercially compromised to deserve a straight Slashdot link, without some caveat...?
What's more, his 'good basic guide' is written by a PhD who's completely lost touch with the 99.999% of web-authors who don't have PhDs in compsci or physics or anything. The W3C has become a big irrelevant exercise in intellectual masturbation, and responsible web-authors ought to secede from its pseudo-AI lunacy.
My site is 100% static HTML, but my rules of thumb for URLs include:
- never more than 80 chars, so they can be emailed
without wrapping
- no uppercase, ever (otherwise you'll forget where
the caps were)
- never more than two directories deep (I sometimes
break this due to bad planning)
- if a new page seems likely to grow into many
pages, it should be created as foo/index.html
instead of foo.html (Someone emailed me this
brilliant tip, I forget who though.)
But the bottom line is to arrange directories
and files (and their names) so that you can
remember them without having to doublecheck.
As soon as you swerve from this narrow path of conformity, your odds of getting a degree (and that bigger income) start to plummet... but at the same time your likelihood of finding meaning in your life starts to escalate.
I could only take a half-year off without losing my scholarships, but my willingness to toe the line plummeted anyway, and I dropped out after a year, then went back a couple more times to other schools, but never got the 4-year degree.
I've rarely sought jobs where a degree would have mattered (eg programming, especially) but I've also had to get used to life being a constant struggle, materially, in exchange for a fantastic advantage in freedom of thought.
Without even hovering or peeking you should have noticed that my User# is 25776 while yours is 155958.
The rule is very simple: minimise 'gotchas'. Any non-standard file-format deserves explicit tagging.
"Look at your browser's status bar"
This is just inconsiderate interface-design. Nobody really wants to have to hover-and-peek before every click.