... why would i want to hang out with people that aren't as smart as me?
This is a bad translation of what actually goes in a geek's head. Geeks like intellectual stimuli as they easily get bored.
Intellectual stimuli comes often from another smart person, but can also be obtained from somebody who is funny, witty, well travelled, or artistically inclined.
Most people do not interact this way, which means that most geeks find the average joe dull. Why would I want to hang out with a dull person?
The problem is not so much to understand the content of a page. That can be done in many instances. It is not that hard to understand if a page is talking about a river "bank" or a money "bank". Usually there are enough quotes and links within the page to allow for this automated differentiation.
The real problem is at the other side, when the user fires Google and enters the standard 2-4 query terms "bank australia". There is a lot less information there for a computer to decide that the user is looking for a bank in Australia.
Metadata on the web pages is pretty much useless for understanding what the user wanted.
Surely this kind of issue is what Tim Berners-Lee and the W3C is trying to address with the Semantic Web.
Indeed, but how close are they from achieving anything of significance? Ai has been working on a Universal Onthohology for ages and gotten nowhere.
The fact that Berners-Lee agree that it would be a "cool thing to have" does not make it any more likely to happen (by the way, TB-L first proposed the semantic web almost five years ago).
I attended a presentation by a French team three years ago, in which they had actually used feedback scalpels in the operating room.
The doctor usually stops cutting well before any feedback is felt, but sometimes it goes near the feedback and overrides it, because his visual inspection is better than the PET scan used to determine feedback levels. The feedback is simply a way to tell a doctor "stop, are you sure about what you are doing?".
They also had samples of robots sawing bones, opening crania and doing, get this, prostate inspections. By the end of the presentation all the males in the room had placed our hands instinctively covering that sensitive area.
I'm seriously considering going back to telephone modem. I'm using cable modem here, and the service seems to go down every other day and be no faster than 100Kbps. Before that I had DSL and that worked like a charm, but there's none to be had around my new house.
I've known a fair number of Ph.D.'s in CS, and they had a hard time getting jobs after graduation .
This used to be a really big problem until the early nineties. By 1998, many large corporations were openly hiring PhDs for advanced development work.
Most of my friends with PhDs had no problem securing a well paid job in industry (and almost all could get a job in academia after 1999, although sometimes only in minor universities).
Almost no mention of non-geek writers to be found. Where are William Styron, Garcia Marquez, Toni Morrison, Michael Ondatje, Kenzaburo Oe, among many others?
Nicholas Negroponte is the same guy who predicted that there would be $1 trillion in e-commerce by 2000, and that micropayments will "change consumer behavior enormously.
Negroponte is yet another snake-oil salesman kept alive by the popular 'science' press.
They are always chasing sexy projects with results always being "around the corner".
Given the untold number of millions spend in the media lab, what do they have to show for it??
No, you're being too general. Banner ads are graphical.
The medium is not the message. I can replace the banner ad at the top of slashdot with a text string "Cheap IBM computers at www.dotbomb.com"
and it is still a banner ad, even if non-graphical. (How do you suppose they look like in lynx, by the way?)
Banner ads are less specifically related to the search in question.
Incorrect. Banner ads in search engines have been highly targeted since the get go.
You can either buy a certain number of generic impressions (which are usually very cheap) or you can buy a combination of keywords (if they search for BMW show the banner ad for www.hot_bimmers.biz). Replacing text for graphics makes them less obtrussive but doesn't change their fundamental banner-ad nature.
(3) charge for "sponsored links" separate from the "unbiased search results"
Nope. This is just a variation of banner ads. The fact that they are presumably related to the query makes them better and more targeted banner ads, but banner ads they are.
Search engine economics is not nearly as glum as you paint them to be.
There is a long list of search engine carcasses to support my gloomy picture (Infoseek, Lycos, Excite, OpenText, Altavista). Until a search engine finds a new revenue model (either new source of income or dramatically lower costs through a technological breakthrough) they will all be condemned to fail.
?? I don't quite understand this question. But if it helps the cost of approx 1 cent is the final cost, including bandwith, sys-admin support, R&D, not just the CPU.
2. Where did you get your figures on the computational cost of a search, and its relation to the size of the web ?
First hand experience on the development side of search engines.
3. Where did you get your figures on the size of the web ?
Widely available in the academic literature. (Try searching for information retrieval).
Can you supply a reference for the figure of 1 cent per impression for banner ads ?
Give me a break. This is also widely available. Take ten seconds to do a search in google for page impression rates and you'll get as answer the standard rate of $10 per 1000 impressions for tons of sites out there.
and do you know how much google gets for its text ads ?
Google is a private company so not much information is available. I think in the past they have claimed, IIRC, to charge a bit more than standard the industry rate.
If you do the math, the economics of search engines just do not work out. Over the last few years the amount in dollars of CPU time required per search has remained more or less constant (yes, CPUs are faster and cheaper, but the web is growing equally as fast).
In practice, each query to a search engine costs about 1 cent. This means the search engine has to recover 1 cent from each user per each query. What is the ongoing rate for banner ad? well, glad you ask: 1 cent for every impression. So assuming you were able to place add impressions in every single search page (which is quite unlikely) you are just breaking even, which brings us to alternative source of revenues.
All of these alternative source of revenues so far boil down to two types:
(1) charge for doing searchers
(2) charge for the listings
There are two ways to charge for doing searches: one is subscription service for users, the other is to license the search technology for third parties. A surprising discovery of the information revolution is that the value of an invidual item is incredibly low, as the editors of Salon magazine, brill's content or Slate can attest to. Therefore users are not likely to jump in and pay for searchers.
If you license the search engine to a company the same effect comes into play: most companies do not own valuable enough information to justify the cost of a search engine.
So (1) is not working how about (2)?
Charge for listings has been tried in many different ways: skewed rankings, faster and more frequent crawling, directory insertion. Skewed rankings is a non-starter as it drives users away (even so, every so often the search-engine-near-bankruptcy-du-jour goes that way).
Charging for frequent crawling works but not many companies sign for it.
So (2) didn't work either, which leaves most search engines struggling to keep afloat. Now here comes the interesting part: as the web continues to grow, the original search engine architecture starts to show its defficiencies.
Rearchitecting an entire search engine live is a major endeavour, with software and hardware costs well into the tens of millions of dollars, but we just said that the search engine company was barely keeping afloat! So they are unable to rev-up into the new generation.
The only group of people who can secure tens of millions of dollars is a startup backed by a bunch of hot shots from academia/industry lab. In comes the upstart out goes the old, monolithic giant. You can tell that story many times over just by changing the names:
I also experienced many problems with my CPx. Loud whirring from the hard drive, periodic freezes, keys not working and on and on. As well the hard drive almost never shuts down, even though RAM is plentiful.
X11 was designed from the start to allow implementations on both thin clients and workstations, a goal that it achieved admirably.
False. That is what they like to claim but the facts do not support their inflated statements.
As I said, X operates under the illusion that is supporting thin-clients, illusion that you have have bought into wholesale, I might add.
The facts are that there are no thin x-clients, and will never be. Having purchased many an X-terminal, I know first hand historical prices of bare-bones X terminals. X requires enough juice on the client side that you might as well have a PC/workstation, and that is exactly what you historically have paid for.
But once you have this PC-like device sitting in front of you (likely labelled NCD) can it run any program locally? No, because according to Athena it is a "thin client".
The sooner we recognize the abject failure X is, the faster somebody else will come up with a decent alternative. So far OS X is in the lead.
I keep seeing people dissing X as a horribly inefficient system that is long overdue for replacement, but the justification always seems to be a myth.
Here's one. The whole X window system was designed from the get go with the idea of having a dumb terminal at the user end. As the design progressed, it became apparent that the X-terminal would require some higher intelligence. In the end X-terminals have a fully functional CPU that is being wasted by 90% of the function calls which were designed before the change of CPU demands.
"Stallman's philosophy is not open source, it's not the spirit of sharing, it's not generous. It has other purposes, it's designed to create a wall between commercial development and free development." (9/7/2000)
I'm a fan of open source (having actually chipped in personally on a few projects), but I agree with Dave Winer on that statement.
It states the same sentiment encapsulated in my.sig below. Think about it before blindly drinking the RMS brand of the OS kool-aid:
Re:Be flexible but go with your strengths
on
Coder or Architect?
·
· Score: 3, Interesting
You should check out Norm Matloff's Debunking the Myth of a Software Labor Shortage [ucdavis.edu].
Actually, you shouldn't. The rapid increase of salaries since the report first came out (1998) pretty much invalidates any claims made by Norm.
They can't stand it because they can't stand the fact that it's such a good test.
Bzzt wrong. They can't stand it because it clouds what AI is about.
We have no need to replicate humans, they are available in ample supplies and we have more fun ways to make more. We need programs that are intelligent in some specific domains, and in fact so good that nobody would ever confuse them for humans, and if somebody did, the programmer would take offense.
The Turing test on the other hand requires the "human-simulator" to hide it's amazing math abilities and its perfect typing. That is not a test of intelligence in any reasonable way. This is just teaching a program artificial stupidity as opposed to artificial intellingence.
From a purely symbolic point of view, the day a computer passes the Turing test will be important, but my guess is that by then most intelligent decisions would have long gone to computers and we would be limited to maintenance tasks and odds and ends (not unlike a production line in a factory where most direct manual labor has gone to machines, and humans maintain them and do the odd task that is difficult for a machine).
Every so often what is a reasonable, modestly interesting idea gets high high-jacked by the snake oil peddlers in whose hands it becomes the solution to the world problems.
The popular press quickly grasps on to it, as these magic bullets increase the circulation of OMNI and Scientific American. Eventually the politicians hear about them and allot untold amounts of money to these efforts.
After 5-10 years nothing much comes out of this, and the snake oil peddlers move on to another area.
Among the thusly overinflated areas we have:
- AI
- neural networks
- expert systems
- nanotechnology
- chaos theory
- e-commerce
- parallel computing
- distributed computing
- complexity (a la Santa Fe Institute) theory
- logic programming
the latest two additions are
- the semantic web
- autonomic systems
/.ers are well advised to apply a healthy dose
of skepticism to any such magic bullet claim.
Slashdot Mirror
... why would i want to hang out with people that aren't as smart as me?
This is a bad translation of what actually goes in a geek's head. Geeks like intellectual stimuli as they easily get bored.
Intellectual stimuli comes often from another smart person, but can also be obtained from somebody who is funny, witty, well travelled, or artistically inclined.
Most people do not interact this way, which means that most geeks find the average joe dull. Why would I want to hang out with a dull person?
The problem is not so much to understand the content of a page. That can be done in many instances. It is not that hard to understand if a page is talking about a river "bank" or a money "bank". Usually there are enough quotes and links within the page to allow for this automated differentiation.
The real problem is at the other side, when the user fires Google and enters the standard 2-4 query terms "bank australia". There is a lot less information there for a computer to decide that the user is looking for a bank in Australia.
Metadata on the web pages is pretty much useless for understanding what the user wanted.
Surely this kind of issue is what Tim Berners-Lee and the W3C is trying to address with the Semantic Web.
Indeed, but how close are they from achieving anything of significance? Ai has been working on a Universal Onthohology for ages and gotten nowhere.
The fact that Berners-Lee agree that it would be a "cool thing to have" does not make it any more likely to happen (by the way, TB-L first proposed the semantic web almost five years ago).
I attended a presentation by a French team three years ago, in which they had actually used feedback scalpels in the operating room.
The doctor usually stops cutting well before any feedback is felt, but sometimes it goes near the feedback and overrides it, because his visual inspection is better than the PET scan used to determine feedback levels. The feedback is simply a way to tell a doctor "stop, are you sure about what you are doing?".
They also had samples of robots sawing bones, opening crania and doing, get this, prostate inspections. By the end of the presentation all the males in the room had placed our hands instinctively covering that sensitive area.
are tying themselves up by squabbling over trivial matters like credit and ego.
There are other reasons beyond ego. Your entire research fuinding might double if proper attribution is made.
I'm seriously considering going back to telephone modem. I'm using cable modem here, and the service seems to go down every other day and be no faster than 100Kbps. Before that I had DSL and that worked like a charm, but there's none to be had around my new house.
I've known a fair number of Ph.D.'s in CS, and they had a hard time getting jobs after graduation .
This used to be a really big problem until the early nineties. By 1998, many large corporations were openly hiring PhDs for advanced development work.
Most of my friends with PhDs had no problem securing a well paid job in industry (and almost all could get a job in academia after 1999, although sometimes only in minor universities).
Almost no mention of non-geek writers to be found. Where are William Styron, Garcia Marquez, Toni Morrison, Michael Ondatje, Kenzaburo Oe, among many others?
If it was a comet that exploded over Tunguska, I think way, way more people would have seen the comet trail heading towards Earth before it exploded.
Only if it had already gone past the sun. If it was on its way there there would be no cauda.
t took over ten years before there was any kind of command history ...--you couldn't even properly edit the command line until doskey came along.
As opposed to Unix in which it only took, what, fifteen years before those things made it to a shell?
Nicholas Negroponte is the same guy who predicted that there would be $1 trillion in e-commerce by 2000, and that micropayments will "change consumer behavior enormously.
Negroponte is yet another snake-oil salesman kept alive by the popular 'science' press.
They are always chasing sexy projects with results always being "around the corner".
Given the untold number of millions spend in the media lab, what do they have to show for it??
No, you're being too general. Banner ads are graphical.
The medium is not the message. I can replace the banner ad at the top of slashdot with a text string "Cheap IBM computers at www.dotbomb.com"
and it is still a banner ad, even if non-graphical. (How do you suppose they look like in lynx, by the way?)
Banner ads are less specifically related to the search in question.
Incorrect. Banner ads in search engines have been highly targeted since the get go.
You can either buy a certain number of generic impressions (which are usually very cheap) or you can buy a combination of keywords (if they search for BMW show the banner ad for www.hot_bimmers.biz). Replacing text for graphics makes them less obtrussive but doesn't change their fundamental banner-ad nature.
Eventually, we WILL rest upon on an as-yet-unrealized distributed search engine
I agree.
If it is not worth private investment it has to become a public good. Either is nationalized as others have suggested, or maintained by public fiat.
(3) charge for "sponsored links" separate from the "unbiased search results"
Nope. This is just a variation of banner ads. The fact that they are presumably related to the query makes them better and more targeted banner ads, but banner ads they are.
Search engine economics is not nearly as glum as you paint them to be.
There is a long list of search engine carcasses to support my gloomy picture (Infoseek, Lycos, Excite, OpenText, Altavista). Until a search engine finds a new revenue model (either new source of income or dramatically lower costs through a technological breakthrough) they will all be condemned to fail.
How did you assign costs to CPU time ?
?? I don't quite understand this question. But if it helps the cost of approx 1 cent is the final cost, including bandwith, sys-admin support, R&D, not just the CPU.
2. Where did you get your figures on the computational cost of a search, and its relation to the size of the web ?
First hand experience on the development side of search engines.
3. Where did you get your figures on the size of the web ?
Widely available in the academic literature. (Try searching for information retrieval).
Can you supply a reference for the figure of 1 cent per impression for banner ads ?
Give me a break. This is also widely available. Take ten seconds to do a search in google for page impression rates and you'll get as answer the standard rate of $10 per 1000 impressions for tons of sites out there.
and do you know how much google gets for its text ads ?
Google is a private company so not much information is available. I think in the past they have claimed, IIRC, to charge a bit more than standard the industry rate.
If you do the math, the economics of search engines just do not work out. Over the last few years the amount in dollars of CPU time required per search has remained more or less constant (yes, CPUs are faster and cheaper, but the web is growing equally as fast).
? ? ..... stay tuned.
In practice, each query to a search engine costs about 1 cent. This means the search engine has to recover 1 cent from each user per each query. What is the ongoing rate for banner ad? well, glad you ask: 1 cent for every impression. So assuming you were able to place add impressions in every single search page (which is quite unlikely) you are just breaking even, which brings us to alternative source of revenues.
All of these alternative source of revenues so far boil down to two types:
(1) charge for doing searchers
(2) charge for the listings
There are two ways to charge for doing searches: one is subscription service for users, the other is to license the search technology for third parties. A surprising discovery of the information revolution is that the value of an invidual item is incredibly low, as the editors of Salon magazine, brill's content or Slate can attest to. Therefore users are not likely to jump in and pay for searchers.
If you license the search engine to a company the same effect comes into play: most companies do not own valuable enough information to justify the cost of a search engine.
So (1) is not working how about (2)?
Charge for listings has been tried in many different ways: skewed rankings, faster and more frequent crawling, directory insertion. Skewed rankings is a non-starter as it drives users away (even so, every so often the search-engine-near-bankruptcy-du-jour goes that way).
Charging for frequent crawling works but not many companies sign for it.
So (2) didn't work either, which leaves most search engines struggling to keep afloat. Now here comes the interesting part: as the web continues to grow, the original search engine architecture starts to show its defficiencies.
Rearchitecting an entire search engine live is a major endeavour, with software and hardware costs well into the tens of millions of dollars, but we just said that the search engine company was barely keeping afloat! So they are unable to rev-up into the new generation.
The only group of people who can secure tens of millions of dollars is a startup backed by a bunch of hot shots from academia/industry lab. In comes the upstart out goes the old, monolithic giant. You can tell that story many times over just by changing the names:
Lycos--OpenText--AltaVista--Hotbot--Google--???
What, you never seen a misteak before?
Back in the early days we used to keep a mispelling index to compare relative search engine sizes:
For example search for interenete
Google: 76 results
Altavista: 9
Lycos: 49
HotBot: 35
I also experienced many problems with my CPx. Loud whirring from the hard drive, periodic freezes, keys not working and on and on. As well the hard drive almost never shuts down, even though RAM is plentiful.
X11 was designed from the start to allow implementations on both thin clients and workstations, a goal that it achieved admirably.
False. That is what they like to claim but the facts do not support their inflated statements.
As I said, X operates under the illusion that is supporting thin-clients, illusion that you have have bought into wholesale, I might add.
The facts are that there are no thin x-clients, and will never be. Having purchased many an X-terminal, I know first hand historical prices of bare-bones X terminals. X requires enough juice on the client side that you might as well have a PC/workstation, and that is exactly what you historically have paid for.
But once you have this PC-like device sitting in front of you (likely labelled NCD) can it run any program locally? No, because according to Athena it is a "thin client".
The sooner we recognize the abject failure X is, the faster somebody else will come up with a decent alternative. So far OS X is in the lead.
I keep seeing people dissing X as a horribly inefficient system that is long overdue for replacement, but the justification always seems to be a myth.
Here's one. The whole X window system was designed from the get go with the idea of having a dumb terminal at the user end. As the design progressed, it became apparent that the X-terminal would require some higher intelligence. In the end X-terminals have a fully functional CPU that is being wasted by 90% of the function calls which were designed before the change of CPU demands.
"Stallman's philosophy is not open source, it's not the spirit of sharing, it's not generous. It has other purposes, it's designed to create a wall between commercial development and free development." (9/7/2000)
.sig below. Think about it before blindly drinking the RMS brand of the OS kool-aid:
I'm a fan of open source (having actually chipped in personally on a few projects), but I agree with Dave Winer on that statement.
It states the same sentiment encapsulated in my
You should check out Norm Matloff's Debunking the Myth of a Software Labor Shortage [ucdavis.edu].
Actually, you shouldn't. The rapid increase of salaries since the report first came out (1998) pretty much invalidates any claims made by Norm.
They can't stand it because they can't stand the fact that it's such a good test.
Bzzt wrong. They can't stand it because it clouds what AI is about.
We have no need to replicate humans, they are available in ample supplies and we have more fun ways to make more. We need programs that are intelligent in some specific domains, and in fact so good that nobody would ever confuse them for humans, and if somebody did, the programmer would take offense.
The Turing test on the other hand requires the "human-simulator" to hide it's amazing math abilities and its perfect typing. That is not a test of intelligence in any reasonable way. This is just teaching a program artificial stupidity as opposed to artificial intellingence.
From a purely symbolic point of view, the day a computer passes the Turing test will be important, but my guess is that by then most intelligent decisions would have long gone to computers and we would be limited to maintenance tasks and odds and ends (not unlike a production line in a factory where most direct manual labor has gone to machines, and humans maintain them and do the odd task that is difficult for a machine).
Every so often what is a reasonable, modestly interesting idea gets high high-jacked by the snake oil peddlers in whose hands it becomes the solution to the world problems.
The popular press quickly grasps on to it, as these magic bullets increase the circulation of OMNI and Scientific American. Eventually the politicians hear about them and allot untold amounts of money to these efforts.
After 5-10 years nothing much comes out of this, and the snake oil peddlers move on to another area.
Among the thusly overinflated areas we have:
- AI
- neural networks
- expert systems
- nanotechnology
- chaos theory
- e-commerce
- parallel computing
- distributed computing
- complexity (a la Santa Fe Institute) theory
- logic programming
the latest two additions are
- the semantic web
- autonomic systems
/.ers are well advised to apply a healthy dose
of skepticism to any such magic bullet claim.