I don't know whether you ever spend time creating or editing articles on Wikipedia on subjects you have more than average knowledge of.
I sometimes do.
It is my finding that Wikipedia articles are a very good example of evolution in action:
* they tend to mutate quite often (by being edited) rarely coming into being in a single step with material created from scratch; * many mutations do not survive the fitness bar, and are either eradicated completely or morph into something that is unrecognizably different; * many articles do indeed have a lot of 'dead weight' that contributes little or nothing to the article's content * some articles do change drastically, being split or merged into completely different ones or seeing their topic change to a completely different understanding of the same title
These things are what makes Wikipedia an evolving system, according to my understanding of what that word means.
But other evolving systems have certain characteristics that Wikipedia doesn't share. In particular, evolution is more organized in Wikipedia than in some other evolving systems. In Wikipedia, all articles have a linear version history. There is no asexual reproduction, i.e. branching of multiple copies article which then evolve separately; and there is no sexual reproduction, i.e. merging of articles as the main way to create new articles. We do know document maintenance systems in which these things are in fact standard and commonly used (namely, distributed version control systems) but Wikipedia is not one of them.
Another difference you claim is that Wikipedia article do not "experience a fitness algorithm (that doesn't rely on intelligence)". But I don't think evolutionary systems actually determine fitness with an algorithm, other than those created in computer models. And as to whether intelligence is used in the process, I simply don't think that criterion is part of what makes a process evolutionary. What makes it evolutionary is that the results do not come about in a single step of creation, but instead, develop gradually from small insignificant beginnings, and forever continue to develop in unforeseen directions.
As to your opening sentence: additionally, females (if old enough) also want to see females scantily dressed, while males don't seem to have the same desire to see males scantily dressed.
Yes, he could use ssh for all that, but in most cases, it is much more convenient to work locally than remotely.
This contradicts my experience as a long-time Unix user. The ease of working remotely has always been one of the main attractions of the Unix platform for me. The option of *not* having a graphical subsystem on our production servers. On machines of any importance, I have always been working remotely. Most of the machines I would work on would be *installed* remotely.
I haven't found it any harder or easier to work with my Linux- or Solaris-based stuff since my desktop changed from Solaris to Windows XP. I also work on Windows remotely by the way (remote desktop).
I can only agree: as a Unix sysadmin (180 Sun workstations) who would occasionally be forced to use Windows, I used to positively hate Exceed: hard to use and manage, clumsy in all possible ways. I was truly relieved when X support was added to Cygwin and never use anything else these days.
That being said, Solaris was on the way out when I quit my sysadmin job some 8 years ago. PCs with Linux were not only much cheaper, but also better (more complete, easier to manage) as desktops. I really wonder how you'd justify a Solaris desktop for the general user these days. Since the arrival of Windows 2000 (rock solid, in my experience) it replaced Solaris/Linux as my desktop; Cygwin/X with ssh/VNC to a Linux box, or a VMWare/Virtual PC with Linux, provides all the Linux functionality I need these days. (Despite Cygwin's poor performance. I've never seen it crash btw.)
So to the original question I can only answer: as soon as I stopped doing system administration, I never had any much need for Solaris or Linux (except for server side stuff, for which I still prefer Linux over Windows). I still use xterms and the Unix commandline tools a lot, but I'd never use Linux as a desktop environment now. Just feel no need for it.
Re:I'm groovy and haven't found an alternative yet
on
Alternative to Groove?
·
· Score: 1
Thanks for the explanation. I think I now understand what Groove does. But how does it improve on a global filesystem like AFS (www.openafs.org)?
I think your comment about "different rules" is misleading. The rules of a field are, by definition, the basic rules that + and * satisfy (see http://en.wikipedia.org/wiki/Field_(mathematics) for the details).
The problem Anderson tackled is that in standard mathematics, division by 0 is undefined. In everything you do with division, both in mathematics and in software, you have to work around that problem. You have to introduce special cases (or be prepared to handle an exception) whenever you need to divide. This is tedious, ugly, error-prone.
So get rid of the special cases by making division by 0 a well-defined result. The result cannot be any of the existing numbers, since that would blow up the whole mechanism of multiplication and division. (In mathematical terms: they wouldn't obey the laws of a field anymore.)
IEEE floating point arithmetic does this: it adds a value NaN to the numbers and defines division by zero to have the result NaN. Wait, says Anderson, now we still have to litter our code (or mathematical formulas) with special cases whenever we *use* a number. Take the idea one step further and also make all operations on numbers well-defined for NaN. Then we'll never run into a exception, and we can inspect whether we still have a normal number whenever we feel like doing so.
So his nullity is NaN, except that it *is* a number, in the sense that all operations on numbers are well defined on it.
Your PHP example is a prime example of stupid language design, called for by (non-)programmers who only see the stupid way of mixing two languages (in this case, PHP and HTML): namely, mixing them arbitrarily at the source code level, instead of the analysed syntax tree level. This is a prime example of what better language design *can* fix.
For example, the use of <?php ?> inserts might be restricted so severely that they could never break the resulting output document's validity *by design*. So better language design *would* fix your misery. I don't know about you, but I have an application here of 74 PHP scripts full of tag soup and I'd love to have a syntax checker tell me whether all 74 of them are guaranteed to spit out correct HTML at runtime no matter what inputs I call them with. Yes, indeed, that would mean I have to "think less", if by "thinking" you mean running View Source results from browsers into a HTML validator until the frequency of syntax errors approaches zero. But smartness in a language often isn't popular; most PHP (non-)programmers will reject it as "too hard" and "too restrictive" and "unnecessary". They don't what to think about it!
Does your work (or that of any of the others in the room) ever involve tasks where you have to intensely think about a problem, spending real mental effort to understand it and juggling the abstractions around in your mind? (I.e. do the PhDs ever perform any work that is remotely related to what they did to acquire their PhDs?)
If the answer is yes, can you perform such tasks in that environment, or do you do your actual thinking at home?
The followup question is, of course: what if, while you are trying to do that type of thinking, you find you have to look up details and explanations once every other minute, scrolling around through let's say a hundred thousand lines of structured text, or through specialized websites? Can you still do that at home?
You have a very good point there. I heard a long radio interview last year with the inventor of this system, and he made this point quite strongly: traffic lights and signs actually distract people. Every sign you put up, every light you blink, takes a little of the driver's attention away from the actual traffic. Signs are symbolic, so they require the drivers to actively look for them, see them, interpret them, and apply them to the situation at hand: valuable milliseconds of attention that, if you encounter one sign every second, add up to a constant distraction.
Most signs symbolically assert states of affairs that the driver may need to be aware of: sharp turn ahead, pedestrians crossing, dangerous intersection, parking space, city centre to the left. It's much safer to convey this information naturally, in the design of the roads and the space around them, so drivers pick up the information automatically. Traffic engineers have become a lot better at doing this over the years, so many of the signs we see today are mostly reminders of what is already clear from the situation. Monderman takes it one step further and says: we shouldn't just redesign our traffic systems to turn most of the signs into superfluous reminders, we should also take the step of not putting these signs up in the first place.
I was there, as a sysadmin for +/- 100 Solaris desktop systems, when IE for Solaris came out.
For our users (academia), Internet access and specifically web browsing was of prime importance. For us sysadmins, reliability and stability of hardware and software, and scaleability of installation and maintenance were the primary concerns. In the hardware area, PCs had finally caught up with Sun's own hardware, but on the OS front, Windows was still fundamentally unsuitable for permanent Internet connectivity. Internet connectivity and web browsers for Windows were there, but it was a fundamentally unsupportable platform. So although we already had many Windows PCs, we were still much happier giving everybody a Sun with Netscape running on it; even though the GUI and most of the applications were starting to look arcane, primitive, weird, hard to use, compared to Windows apps. Netscape was the one exception. Everybody used it, and most people were happy to use Suns because of it. Linux was already making inroads, but we weren't sure it would be stable enough and would support the ever changing PC hardware well enough. Besides, we had a support contract from Sun.
So IE for Solaris came out, and since it was free, I tried it out. It was vastly superior to Netscape. Much faster, made pages look a lot better. However, it caused a serious problem with X (as far as I remember, X would lock up). So I called Sun support (not something I needed to do very often) and explained the problem. The reply: *Microsoft*? We don't support Microsoft applications.
That was the turning point for me. We have Windows and Linux desktops now, like everybody else.
Whatever moral you like to draw from this story, IE for Solaris clearly made sense at the time.
Yes, of course the intent behind making Virtual PC and Virtual Server free (and yes, they are *already* free) is to make people run more Windows operating systems.
But the point that you assholes don't seem to get is that Virtual PC and Virtual Server do not, in any way, limit you to Windows operating systems. The host system must be Windows, yes, but on my Virtual PC I have Solaris and SuSE Linux running happily alongside Windows Server, Windows XP and Windows 2000 (and concurrently, too) without any problem. All the software does is give you extra options, if you already have a Windows OS on whjich to host it. This ability is incredibly useful, I use it almost daily.
Jeez, you get very useful, very user-friendly software thrown at you for free and all you can do is complain. If you don't want to use the software, for whatever reason, then just don't! There are plenty of alternatives available. But many of the posters here don't even have a rough idea what this software is capable of. I'm sorry but that's just pathetic.
Google's core business (their web search and ad revenues from it) does not appear to threaten Microsoft's business in any way. I don't see the *need* for Google to produce revenue from other applications. Just to make the Slashdot crowd happy?
Conversely, Microsoft is only a threat to Google in as far as its web search facility takes away business from Google's. Which it hardly seems to do. And it doesn't need to, either, Microsoft's business doesn't rely on being successful in web advertising or anything of the kind.
So they aren't really competitors; some of their products are, eg gmail vs Hotmail but none of this competition is a real challenge to either company, as far as I can see.
I think without knowing the work it's hard to tell who deserves more credit, the person with the basic ideas or the person who takes the trouble to actually check out all the details. The history of mathematics is full of proofs that turned out to fail, stumbling over what looked like a minor detail. Sometimes the problem is only spotted after many years. So it's like any complicated piece of machinery: attention to detail can be crucial.
News sites use African languages; blogging sites or community sites make sense in African languages; but for encyclopedias the barrier is higher, since what you write there is supposed to be objective, universal, noncontroversial, understandable by anybody. The language itself has to be universal enough. Swahili meets that criterion: almost all Tanzanians speak it, and this is the case in several other countries, so by writing in Swahili you do write for everybody. For most other African languages the situation is more complicated because they are not spoken nation-wide. When writing encyclopedia articles about Nigeria, I guess most Nigerians might prefer English.
(This can be checked, of course - just look at the representation of Africa in en.wikipedia.org and fr.wikipedia.org.)
Well, Google is definitely better, and I agree that it had to be before arbitrary non-techies would accept it. But Altavista was the first that I started to bring up in conversation on random subjects, the first that I suggested to non-technical Internet users.
There are two obvious ideas for indexing the Web: automatic full-text indexing (Google) or manually built directory hierarchies (Yahoo, dmoz.org).
The first manual directories were the pages maintained at CERN and NSCA - they were small. The first publically editable manual directory system I saw was Oliver McBryan's "Mother of all BBSs". He made it publically available together in 1994, together with his full-text search indexer, the WWWW. They were his tools for taming the Web.
It wasn't clear (at least to me) whether either of the approaches would scale. They competed for a while. McBryan didn't develop his systems further, but Yahoo copied the "Mother of all BBSs" concept and became very popular, while Webcrawler was the first popular Internet-wide full-text indexer.
Even at that time Webcrawler, which I believe covered about 10% of all web pages, was already superior to Yahoo, which couldn't match its completeness and lack of user bias. Later, Lycos and Altavista took over from Webcrawler and seriously aimed for 100% coverage. By then, I regarded Yahoo's pupularity as a matter of ignorance. Their directory tree is always helpful, of course, but if you can think of any search terms to use at all, even related ones, full-text search is so much better.
Novices don't understand that. To them, a manually created directory tree looks good: they don't realize how much is missing and how much is in there in places where they'll never think to look. Full-text search looks bad to them because all of the worthless results. They don't look past the bad results, they don't realize how many more good results they're getting, and they don't realize how easily they can experiment with the search terms to improve results, because they are not aware that keywords are just text strings found on a page.
So Yahoo's business was based on ignorance. But hey, that's where the users were...
Youre not being fair to Altavista. Altavista worked very well; like Google, it nearly always gave useful results. It proved to the world that Internet-wide full-text indexing was worthwhile. Google is just one of its successors, arguably the best, but still, only an incremental improvement.
I'm not sure it is a "fact" that a Christmas tree is a tree. It is a fact that we *call* it a tree when speaking English. The facts are more complicated.
The idea behind Cyc is that we can basically make so many detailed statements about Christmas and trees that we somehow approach what a tree really is. At the end of the day, of course, it's still language: phrases in some language (not English this time) used to talk about trees. And just as you're right that it does say something about Christmas trees that the English call it a Christmas tree, so Cyc descriptions do say something about the things they describe. But they don't say much unless you learn the language they are written in.
Slashdot Mirror
I don't know whether you ever spend time creating or editing articles on Wikipedia on subjects you have more than average knowledge of.
I sometimes do.
It is my finding that Wikipedia articles are a very good example of evolution in action:
* they tend to mutate quite often (by being edited) rarely coming into being in a single step with material created from scratch;
* many mutations do not survive the fitness bar, and are either eradicated completely or morph into something that is unrecognizably different;
* many articles do indeed have a lot of 'dead weight' that contributes little or nothing to the article's content
* some articles do change drastically, being split or merged into completely different ones or seeing their topic change to a completely different understanding of the same title
These things are what makes Wikipedia an evolving system, according to my understanding of what that word means.
But other evolving systems have certain characteristics that Wikipedia doesn't share.
In particular, evolution is more organized in Wikipedia than in some other evolving systems. In Wikipedia, all articles have a linear version history. There is no asexual reproduction, i.e. branching of multiple copies article which then evolve separately; and there is no sexual reproduction, i.e. merging of articles as the main way to create new articles. We do know document maintenance systems in which these things are in fact standard and commonly used (namely, distributed version control systems) but Wikipedia is not one of them.
Another difference you claim is that Wikipedia article do not "experience a fitness algorithm (that doesn't rely on intelligence)". But I don't think evolutionary systems actually determine fitness with an algorithm, other than those created in computer models. And as to whether intelligence is used in the process, I simply don't think that criterion is part of what makes a process evolutionary. What makes it evolutionary is that the results do not come about in a single step of creation, but instead, develop gradually from small insignificant beginnings, and forever continue to develop in unforeseen directions.
As to your opening sentence: additionally, females (if old enough) also want to see females scantily dressed, while males don't seem to have the same desire to see males scantily dressed.
Yes, he could use ssh for all that, but in most cases, it is much more convenient to work locally than remotely.
This contradicts my experience as a long-time Unix user. The ease of working remotely has always been one of the main attractions of the Unix platform for me. The option of *not* having a graphical subsystem on our production servers. On machines of any importance, I have always been working remotely. Most of the machines I would work on would be *installed* remotely.
I haven't found it any harder or easier to work with my Linux- or Solaris-based stuff since my desktop changed from Solaris to Windows XP. I also work on Windows remotely by the way (remote desktop).
I can only agree: as a Unix sysadmin (180 Sun workstations) who would occasionally be forced to use Windows, I used to positively hate Exceed: hard to use and manage, clumsy in all possible ways. I was truly relieved when X support was added to Cygwin and never use anything else these days.
That being said, Solaris was on the way out when I quit my sysadmin job some 8 years ago. PCs with Linux were not only much cheaper, but also better (more complete, easier to manage) as desktops. I really wonder how you'd justify a Solaris desktop for the general user these days. Since the arrival of Windows 2000 (rock solid, in my experience) it replaced Solaris/Linux as my desktop; Cygwin/X with ssh/VNC to a Linux box, or a VMWare/Virtual PC with Linux, provides all the Linux functionality I need these days. (Despite Cygwin's poor performance. I've never seen it crash btw.)
So to the original question I can only answer: as soon as I stopped doing system administration, I never had any much need for Solaris or Linux (except for server side stuff, for which I still prefer Linux over Windows). I still use xterms and the Unix commandline tools a lot, but I'd never use Linux as a desktop environment now. Just feel no need for it.
Thanks for the explanation. I think I now understand what Groove does.
But how does it improve on a global filesystem like AFS (www.openafs.org)?
The saying is not about shoddy newspapers, but about news in general: today's news is tomorrow's fish-wrap.
I've been trying to track its origin but didn't succeed.
Thanks for the pointer!
I think your comment about "different rules" is misleading.
The rules of a field are, by definition, the basic rules that + and * satisfy
(see http://en.wikipedia.org/wiki/Field_(mathematics) for the details).
The problem Anderson tackled is that in standard mathematics, division by 0 is undefined.
In everything you do with division, both in mathematics and in software,
you have to work around that problem. You have to introduce special cases
(or be prepared to handle an exception) whenever you need to divide.
This is tedious, ugly, error-prone.
So get rid of the special cases by making division by 0 a well-defined result.
The result cannot be any of the existing numbers, since that
would blow up the whole mechanism of multiplication and division.
(In mathematical terms: they wouldn't obey the laws of a field anymore.)
IEEE floating point arithmetic does this: it adds a value NaN to the numbers
and defines division by zero to have the result NaN. Wait, says Anderson,
now we still have to litter our code (or mathematical formulas) with
special cases whenever we *use* a number. Take the idea one step further
and also make all operations on numbers well-defined for NaN.
Then we'll never run into a exception, and we can inspect
whether we still have a normal number whenever we feel like doing so.
So his nullity is NaN, except that it *is* a number, in the sense that
all operations on numbers are well defined on it.
I must say I can't follow your argument.
Your PHP example is a prime example of stupid language design, called for by (non-)programmers who only see the stupid way of mixing two languages (in this case, PHP and HTML): namely, mixing them arbitrarily at the source code level, instead of the analysed syntax tree level. This is a prime example of what better language design *can* fix.
For example, the use of <?php ?> inserts might be restricted so severely that they could never break the resulting output document's validity *by design*. So better language design *would* fix your misery. I don't know about you, but I have an application here of 74 PHP scripts full of tag soup and I'd love to have a syntax checker tell me whether all 74 of them are guaranteed to spit out correct HTML at runtime no matter what inputs I call them with. Yes, indeed, that would mean I have to "think less", if by "thinking" you mean running View Source results from browsers into a HTML validator until the frequency of syntax errors approaches zero. But smartness in a language often isn't popular; most PHP (non-)programmers will reject it as "too hard" and "too restrictive" and "unnecessary". They don't what to think about it!
It has been fixed, of course, but you may have missed the solutions, which (apart from template-based programming) aren't called C++.
The one thing I never understood is the reversal of the OK and Cancel buttons.
I'd like to see the history of that.
Does your work (or that of any of the others in the room) ever involve tasks where you have to intensely think about a problem, spending real mental effort to understand it and juggling the abstractions around in your mind? (I.e. do the PhDs ever perform any work that is remotely related to what they did to acquire their PhDs?)
If the answer is yes, can you perform such tasks in that environment, or do you do your actual thinking at home?
The followup question is, of course: what if, while you are trying to do that type of thinking, you find you have to look up details and explanations once every other minute, scrolling around through let's say a hundred thousand lines of structured text, or through specialized websites? Can you still do that at home?
You have a very good point there. I heard a long radio interview last year with the inventor of this system, and he made this point quite strongly: traffic lights and signs actually distract people. Every sign you put up, every light you blink, takes a little of the driver's attention away from the actual traffic. Signs are symbolic, so they require the drivers to actively look for them, see them, interpret them, and apply them to the situation at hand: valuable milliseconds of attention that, if you encounter one sign every second, add up to a constant distraction.
Most signs symbolically assert states of affairs that the driver may need to be aware of: sharp turn ahead, pedestrians crossing, dangerous intersection, parking space, city centre to the left. It's much safer to convey this information naturally, in the design of the roads and the space around them, so drivers pick up the information automatically. Traffic engineers have become a lot better at doing this over the years, so many of the signs we see today are mostly reminders of what is already clear from the situation. Monderman takes it one step further and says: we shouldn't just redesign our traffic systems to turn most of the signs into superfluous reminders, we should also take the step of not putting these signs up in the first place.
At least that's how I remember the interview.
Besides, it wasn't a crappy answer at all.
I was there, as a sysadmin for +/- 100 Solaris desktop systems, when IE for Solaris came out.
For our users (academia), Internet access and specifically web browsing was of prime importance.
For us sysadmins, reliability and stability of hardware and software, and scaleability of
installation and maintenance were the primary concerns. In the hardware area, PCs had
finally caught up with Sun's own hardware, but on the OS front, Windows was still fundamentally
unsuitable for permanent Internet connectivity. Internet connectivity and web browsers for Windows
were there, but it was a fundamentally unsupportable platform. So although we already had many
Windows PCs, we were still much happier giving everybody a Sun with Netscape running on it;
even though the GUI and most of the applications were starting to look arcane, primitive, weird,
hard to use, compared to Windows apps. Netscape was the one exception. Everybody used it,
and most people were happy to use Suns because of it. Linux was already making inroads,
but we weren't sure it would be stable enough and would support the ever changing PC hardware
well enough. Besides, we had a support contract from Sun.
So IE for Solaris came out, and since it was free, I tried it out. It was vastly superior to Netscape.
Much faster, made pages look a lot better. However, it caused a serious problem with X
(as far as I remember, X would lock up). So I called Sun support (not something I needed to do
very often) and explained the problem. The reply: *Microsoft*? We don't support Microsoft applications.
That was the turning point for me. We have Windows and Linux desktops now, like everybody else.
Whatever moral you like to draw from this story, IE for Solaris clearly made sense at the time.
Well, only to those who have seen Enlightenment.
Yes, of course the intent behind making Virtual PC and Virtual Server free (and yes, they are *already* free) is to make people run more Windows operating systems.
But the point that you assholes don't seem to get is that Virtual PC and Virtual Server do not, in any way, limit you to Windows operating systems. The host system must be Windows, yes, but on my Virtual PC I have Solaris and SuSE Linux running happily alongside Windows Server, Windows XP and Windows 2000 (and concurrently, too) without any problem. All the software does is give you extra options, if you already have a Windows OS on whjich to host it. This ability is incredibly useful, I use it almost daily.
Jeez, you get very useful, very user-friendly software thrown at you for free and all you can do is complain.
If you don't want to use the software, for whatever reason, then just don't! There are plenty of alternatives available. But many of the posters here don't even have a rough idea what this software is capable of. I'm sorry but that's just pathetic.
Google's core business (their web search and ad revenues from it) does not appear to threaten Microsoft's business in any way.
I don't see the *need* for Google to produce revenue from other applications. Just to make the Slashdot crowd happy?
Conversely, Microsoft is only a threat to Google in as far as its web search facility takes away business from Google's.
Which it hardly seems to do. And it doesn't need to, either, Microsoft's business doesn't rely on being successful in
web advertising or anything of the kind.
So they aren't really competitors; some of their products are, eg gmail vs Hotmail
but none of this competition is a real challenge to either company, as far as I can see.
PHP started as a Perl script written by Rasmus Lerdorf for his personal home page.
i'm pretty sure that Perl is easier than Basic!
I think without knowing the work it's hard to tell who deserves more credit, the person with the basic ideas or the person who takes the trouble to actually check out all the details. The history of mathematics is full of proofs that turned out to fail, stumbling over what looked like a minor detail. Sometimes the problem is only spotted after many years. So it's like any complicated piece of machinery: attention to detail can be crucial.
(Yes, of course.)
News sites use African languages; blogging sites or community sites make sense in African languages; but for encyclopedias the barrier is higher, since what you write there is supposed to be objective, universal, noncontroversial, understandable by anybody. The language itself has to be universal enough. Swahili meets that criterion: almost all Tanzanians speak it, and this is the case in several other countries, so by writing in Swahili you do write for everybody. For most other African languages the situation is more complicated because they are not spoken nation-wide. When writing encyclopedia articles about Nigeria, I guess most Nigerians might prefer English.
(This can be checked, of course - just look at the representation of Africa in en.wikipedia.org and fr.wikipedia.org.)
Sharks?
To be honest, I worry more about traffic, cancer, heart disease, and antagonism along religious lines among the people in my own neighborhood.
Well, Google is definitely better, and I agree that it had to be before arbitrary non-techies
would accept it. But Altavista was the first that I started to bring up in conversation
on random subjects, the first that I suggested to non-technical Internet users.
There are two obvious ideas for indexing the Web: automatic full-text indexing (Google) or manually built directory hierarchies (Yahoo, dmoz.org).
...
The first manual directories were the pages maintained at CERN and NSCA - they were small. The first publically editable manual directory system I saw was Oliver McBryan's "Mother of all BBSs". He made it publically available together in 1994, together with his full-text search indexer, the WWWW. They were his tools for taming the Web.
It wasn't clear (at least to me) whether either of the approaches would scale.
They competed for a while. McBryan didn't develop his systems further, but Yahoo copied the "Mother of all BBSs" concept and became very popular, while Webcrawler was the first popular Internet-wide full-text indexer.
Even at that time Webcrawler, which I believe covered about 10% of all web pages, was already superior to Yahoo, which couldn't match its completeness and lack of user bias. Later, Lycos and Altavista took over from Webcrawler and seriously aimed for 100% coverage. By then, I regarded Yahoo's pupularity as a matter of ignorance. Their directory tree is always helpful, of course, but if you can think of any search terms to use at all, even related ones, full-text search is so much better.
Novices don't understand that. To them, a manually created directory tree looks good: they don't realize how much is missing and how much is in there in places where they'll never think to look. Full-text search looks bad to them because all of the worthless results. They don't look past the bad results, they don't realize how many more good results they're getting, and they don't realize how easily they can experiment with the search terms to improve results, because they are not aware that keywords are just text strings found on a page.
So Yahoo's business was based on ignorance. But hey, that's where the users were
Youre not being fair to Altavista. Altavista worked very well; like Google, it nearly always gave useful results.
It proved to the world that Internet-wide full-text indexing was worthwhile. Google is just one of its successors,
arguably the best, but still, only an incremental improvement.
I'm not sure it is a "fact" that a Christmas tree is a tree. It is a fact that we *call* it a tree when speaking English.
The facts are more complicated.
The idea behind Cyc is that we can basically make so many detailed statements about Christmas and trees that we somehow approach what a tree really is. At the end of the day, of course, it's still language: phrases in some language (not English this time) used to talk about trees. And just as you're right that it does say something about Christmas trees that the English call it a Christmas tree, so Cyc descriptions do say something about the things they describe. But they don't say much unless you learn the language they are written in.