Chinese history is replete with scientific and intellectual firsts like Zheng He's explorations -that were crushed by a stifling political power structure that protected incumbent interests and crushed change. After his explorations, Zheng He fell out of favor and the fleet was dismantled, there was no stomach to exploit contact with the outside world.
Jared Diamonds 'Guns, Germs and Steel' has a rather gripping discussion of the whole thing.
I worked in the search engine industry for some years. Term 'search algorithm' is a facile gloss over what is a complex business, usually meaning how a set of indexed objects are selected and ranked for presentation to the user. Link analysis, clustering and 'communities of interest' have been heavily investigated for years by both commercial and academic groups. There are various flavours ranging from a global popularity measurement, like Larry Page's "Page Rank" (part of Googles ranking algorithm) to damped term-frequency schemes in which pages linked to and linking from a page contribute search terms (keywords) on a sliding scale based on link type and 'distance' from the page. It is very hard to tell from the press release the particular twist these guys are applying.
The problem with most of these schemes is that they don't scale worth a damn or fail in the real world. What works on a small number of sites fails completely when you want to enable users to search the 'entire web'. Web search engines run very tight code on very many machines and have to cope with the fact that there is a lot of crap out there. Assumptions about link popularity follow the old rule that an optimist is likely to be dissappointed but a pessimist can always be pleasantly surprised.
Search engines have two ways of discovering pages, spidering and direct submission by site owners. The is quite an industry spamming them both ways. Spam here meaning trying to trick the engine into returning your site as relevant to a query when it isn't. Everyone wants to be top of the result list, even for searchers who are looking for something else - the goal is to drive traffic by any means.
All the engines are secretive about their spidering and ranking approaches because they are in an ongoing arms race with the spamming industry. Andre Broder at Altavista describes himself as being in the business of "adversarial information retrieval". 95% of submissions to Altavista are rejected as spam. At one time Infoseek had a site try to submit 2 million pages for indexing and was forced to limit all sites to one page.
Spammers try to increase their rankings by cross linking huge numbers of URL's, this is one reason why there are so many porn pages that consist of nothing but banner adds and links to other pages of the same. So link clustering analysis is used to eliminate many of these from being indexed.
Its pretty amazing that Google, et al produce results as good as they do!
People won't buy expensive sets unless there is great programming that they make a difference for; and set prices won't come down until purchase volume goes up.
So why no programming? The networks can afford to upgrade their broadcast infrastructure; its expensive but not a show stopper. But, production costs for HD are estimated to be 3-10 times what they are for regular programming. Sets are much more expensive. Current sets look as cheesy as they do when you visit the studio. In HD you can see that what is supposed to be wood is MacTac on cardboard. "Television makeup" looks like it was applied with a trowel. Go to a store and watch Jay Leno in HD to see what I'm talking about.
Its the classic chicken and egg dilemma. The networks won't inflate their production costs until their are enough sets to make it worth their while and consumers won't buy unless the sets give them some value. Most people in North America have bought HD sets for playing DVD's rather than for though they are overkill for the 720*480 DVD format.
Currently regular NSTC broadcast TV looks weird on most HD sets. Resolution enhancement ("line doubling") can improve the appearance of NSTC broadcasts, but when the sets I've seen stretch the signal to the wider format, people take on a squat appearance. You'll actually violate the warranty on many HD projection sets if you watch NSTC broadcasts in their native aspect ratio because the bars down the side will burn in.
Interestingly, wide screen HD capable TV's are selling well in Europe, because current PAL broadcasts look great on them. PAL has higher resolution and a slightly wider picture than NSTC, but uses a 50 Hz interlaced refresh rate and versus 60 Hz here, and the European HD standard has a slightly narrower picture. Manufacturers are selling HD TV's as "100 Hz" sets, using resolution enhancement to stretch the picture and display it at 100Hz progressive scan. The distortion is very hard to see and the pictures look great. Interestingly these sets seem to be smaller and weigh much less than the CRT based HD sets I've seen in Canada and the US.
Brin's Book - The Transparent Society
on
David Brin on Privacy
·
· Score: 2, Interesting
I bought Brin's book (ISBN: 0738201448) when it came out in '99. I was struck with his sense that surveilance in the larger sense was technologically inevitable- not only cameras, but every expenditure, even RF tags on your money . He argued that it was impossible to supress this capability; that doing so would simply give those in power the ability to take unfair advantage; so we should make everything completely transparent. If we all have the legal right to spy on each other, the little guy can't be sanctioned for finding out what the bigs guys are up to - kind of a pessimists take on "information wants to be free".
Maybe I'm failing to adapt to change, but the prospect of what he proposes makes me really uncomfortable and could lead to a level of social conformity that most of us would find stiffing. Also, I don't know that I have that little faith in our (western civilization's - I'm Canadian) ability to govern our behaviour and that of our institutions.
The book is worth a read - I may just haul it out and take another look.
While there are aspects of Creative Commons agenda that push the envelope of achievability, a well organized set of free license agreements, is a nice concept.
An artist might, for example, agree to give away a work as long as no one is making money on it but include a provision requiring payments on a sliding scale if it's sold. As participation in the Commons project increases, a variety of specific intellectual-property license options will evolve in response to user needs, which in turn would create templates for others with similar requirements.
Within a few months, artists, writers and others will soon be able to go online, select the options that suit them best and receive a custom-made license they can append to their works without having to pay a dime to a lawyer, let alone the thousands of dollars it typically costs to purchase similar legal services.
A consistent set of licenses that cover the objectives of GPL, LGPL, Berkely, Artistic, etc. and other points on the spectrum to fully commercial would be a great benefit to us all.
You can read the paper at http://www.nature.com/cgi-taf/dynapage.taf?file=/n ature/journal/vaop/ncurrent/index.html (free registration required).
The work is significant because it describes how a specific set of relatively small number of changes in the genes of many-limbed arthopods produce the major changes in body design the result in the siz legged insects. Anyone familiar with genetics (and software) can theorize small changes in code can produce large changes in results; these folks have done the hard slogging to investigate a specific instance of it actually occurring.
BTW, I find the creationism/evolution flames are getting a little out of control. I'd be interested in hearing an opinion of the actual work from someone with some expertise.
All customers are not created equal. I would probably fall in that 10% of users who eat 70% of Rogers capacity, if I was one their customers. I got my Bell Sympatico high speed connection in the first week of their ADSL offering (five years ago?): 2Mbps in, 1 Mbps out, static IP address for $68.95/month. My effective throughput varies, but is typically in the range of 1.1-1.6 Mbps in and 700-900 Kbps out, with the occasional drop out, which is WAY faster than anyone I know whose using Rogers.
You can't buy this anymore - Sympatico reduced the speed of the service and cut their price to compete with Rogers cable modem service.
I've never been tempted to move to Rogers lower (effective) speed cable modem service or to Bells 'One Meg Modem' high speed offering, even though these are lower cost. Also, I like Sympatico's enlightened attitude to connection sharing. Far from discouraging it, their web site will sell you a 4-port D-Link DSL Broadband Router for $130 (US$82)
There's a simple editor on the humanmarkup.org site called HumanML Write. The output is kind of bizarrely interesting, e.g. When I see things like I wonder what the debates about enumeration values will be like.
Ai's Chief Scientist is Jason Hutchens,winner of the 1996 Loebner pseudo-Turing test competition. Their website contains some of Hutchen's old papers, you can get a better selection
at his personal home page http://www.amristar.com.au/~hutch/. Interestingly, Hutchen's previous gig
was working on AI for LionHead studio's "Black
and White" game.
Slashdot Mirror
Troll doesn't quite express it.
The rights management/Descarte bit that culminates in
"'Yeah?' Liam said. 'Who's God, then?" "Crypto,' Murray said."
is pretty amusing.
Chinese history is replete with scientific and intellectual firsts like Zheng He's explorations -that were crushed by a stifling political power structure that protected incumbent interests and crushed change. After his explorations, Zheng He fell out of favor and the fleet was dismantled, there was no stomach to exploit contact with the outside world.
Jared Diamonds 'Guns, Germs and Steel' has a rather gripping discussion of the whole thing.
I worked in the search engine industry for some years. Term 'search algorithm' is a facile gloss over what is a complex business, usually meaning how a set of indexed objects are selected and ranked for presentation to the user. Link analysis, clustering and 'communities of interest' have been heavily investigated for years by both commercial and academic groups. There are various flavours ranging from a global popularity measurement, like Larry Page's "Page Rank" (part of Googles ranking algorithm) to damped term-frequency schemes in which pages linked to and linking from a page contribute search terms (keywords) on a sliding scale based on link type and 'distance' from the page. It is very hard to tell from the press release the particular twist these guys are applying.
The problem with most of these schemes is that they don't scale worth a damn or fail in the real world. What works on a small number of sites fails completely when you want to enable users to search the 'entire web'. Web search engines run very tight code on very many machines and have to cope with the fact that there is a lot of crap out there. Assumptions about link popularity follow the old rule that an optimist is likely to be dissappointed but a pessimist can always be pleasantly surprised.
Search engines have two ways of discovering pages, spidering and direct submission by site owners. The is quite an industry spamming them both ways. Spam here meaning trying to trick the engine into returning your site as relevant to a query when it isn't. Everyone wants to be top of the result list, even for searchers who are looking for something else - the goal is to drive traffic by any means.
All the engines are secretive about their spidering and ranking approaches because they are in an ongoing arms race with the spamming industry. Andre Broder at Altavista describes himself as being in the business of "adversarial information retrieval". 95% of submissions to Altavista are rejected as spam. At one time Infoseek had a site try to submit 2 million pages for indexing and was forced to limit all sites to one page.
Spammers try to increase their rankings by cross linking huge numbers of URL's, this is one reason why there are so many porn pages that consist of nothing but banner adds and links to other pages of the same. So link clustering analysis is used to eliminate many of these from being indexed.
Its pretty amazing that Google, et al produce results as good as they do!
People won't buy expensive sets unless there is great programming that they make a difference for; and set prices won't come down until purchase volume goes up.
So why no programming? The networks can afford to upgrade their broadcast infrastructure; its expensive but not a show stopper. But, production costs for HD are estimated to be 3-10 times what they are for regular programming. Sets are much more expensive. Current sets look as cheesy as they do when you visit the studio. In HD you can see that what is supposed to be wood is MacTac on cardboard. "Television makeup" looks like it was applied with a trowel. Go to a store and watch Jay Leno in HD to see what I'm talking about.
Its the classic chicken and egg dilemma. The networks won't inflate their production costs until their are enough sets to make it worth their while and consumers won't buy unless the sets give them some value. Most people in North America have bought HD sets for playing DVD's rather than for though they are overkill for the 720*480 DVD format.
Currently regular NSTC broadcast TV looks weird on most HD sets. Resolution enhancement ("line doubling") can improve the appearance of NSTC broadcasts, but when the sets I've seen stretch the signal to the wider format, people take on a squat appearance. You'll actually violate the warranty on many HD projection sets if you watch NSTC broadcasts in their native aspect ratio because the bars down the side will burn in.
Interestingly, wide screen HD capable TV's are selling well in Europe, because current PAL broadcasts look great on them. PAL has higher resolution and a slightly wider picture than NSTC, but uses a 50 Hz interlaced refresh rate and versus 60 Hz here, and the European HD standard has a slightly narrower picture. Manufacturers are selling HD TV's as "100 Hz" sets, using resolution enhancement to stretch the picture and display it at 100Hz progressive scan. The distortion is very hard to see and the pictures look great. Interestingly these sets seem to be smaller and weigh much less than the CRT based HD sets I've seen in Canada and the US.
I bought Brin's book (ISBN: 0738201448) when it came out in '99. I was struck with his sense that surveilance in the larger sense was technologically inevitable- not only cameras, but every expenditure, even RF tags on your money . He argued that it was impossible to supress this capability; that doing so would simply give those in power the ability to take unfair advantage; so we should make everything completely transparent. If we all have the legal right to spy on each other, the little guy can't be sanctioned for finding out what the bigs guys are up to - kind of a pessimists take on "information wants to be free".
Maybe I'm failing to adapt to change, but the prospect of what he proposes makes me really uncomfortable and could lead to a level of social conformity that most of us would find stiffing. Also, I don't know that I have that little faith in our (western civilization's - I'm Canadian) ability to govern our behaviour and that of our institutions.
The book is worth a read - I may just haul it out and take another look.
A consistent set of licenses that cover the objectives of GPL, LGPL, Berkely, Artistic, etc. and other points on the spectrum to fully commercial would be a great benefit to us all.
You can read the paper at http://www.nature.com/cgi-taf/dynapage.taf?file=/n ature/journal/vaop/ncurrent/index.html (free registration required).
The work is significant because it describes how a specific set of relatively small number of changes in the genes of many-limbed arthopods produce the major changes in body design the result in the siz legged insects. Anyone familiar with genetics (and software) can theorize small changes in code can produce large changes in results; these folks have done the hard slogging to investigate a specific instance of it actually occurring.
BTW, I find the creationism/evolution flames are getting a little out of control. I'd be interested in hearing an opinion of the actual work from someone with some expertise.
Thanks for the heads up. I didn't notice this on their web-site. Do you have a URL?
All customers are not created equal. I would probably fall in that 10% of users who eat 70% of Rogers capacity, if I was one their customers. I got my Bell Sympatico high speed connection in the first week of their ADSL offering (five years ago?): 2Mbps in, 1 Mbps out, static IP address for $68.95/month. My effective throughput varies, but is typically in the range of 1.1-1.6 Mbps in and 700-900 Kbps out, with the occasional drop out, which is WAY faster than anyone I know whose using Rogers.
You can't buy this anymore - Sympatico reduced the speed of the service and cut their price to compete with Rogers cable modem service.
I've never been tempted to move to Rogers lower (effective) speed cable modem service or to Bells 'One Meg Modem' high speed offering, even though these are lower cost. Also, I like Sympatico's enlightened attitude to connection sharing. Far from discouraging it, their web site will sell you a 4-port D-Link DSL Broadband Router for $130 (US$82)
There's a simple editor on the humanmarkup.org site called HumanML Write. The output is kind of bizarrely interesting, e.g. When I see things like I wonder what the debates about enumeration values will be like.
Ai's Chief Scientist is Jason Hutchens,winner of the 1996 Loebner pseudo-Turing test competition. Their website contains some of Hutchen's old papers, you can get a better selection
at his personal home page http://www.amristar.com.au/~hutch/. Interestingly, Hutchen's previous gig
was working on AI for LionHead studio's "Black
and White" game.