What's to stop these ghosts being maliciously "trained" to give the wrong answer... I remember a teacher at college (Mr Tittershill), who was routinely used in a joke on freshers (report to Mr Boobershill at the senior common room, NOW!...)
Is it only me who first thinks of "how to game the system" when presented with a new technology ? Perhaps I should have been a hacker:-)) [note to US authorities - this is a joke, and I have no intention of committing any crimes (cyber- or otherwise) when visiting the USA]
Simon
Barking mad
on
See Spot Surf
·
· Score: 5, Insightful
The problem with a 'free society' is that some people have far far too much time on their hands:-) All the more kudos to the site-creators for finding an oft-overlooked gap in the market for a site that most people (the sane ones) never suspected would exist...
Oh yeah, and (from the article):
Rosemary Pepper is a writer and founder of an online feminist magazine who wrote the copy for the Dogster site. She is the "I Like It Ruff" columnist.
/me definitely doesn't mention anything about spicy feminists liking it "ruff", 'cos that would be politically incorrect [grin]
Surely this is a classic example of the Manager pattern. You have a bunch of objects [Avatar] (all alike, at least programmatically:-) who want to perform operations on other objects. If the system has a [GuildManager] class, then access to this distributed network of avatars can be forced through the choke-point of 'can this avatar join this guild'.
The trade-off in terms of scalability is in frequency versus computation. If the operation is commonplace (such as moving around), then a distributed system has a problem. If the operation is not commonplace (such as joining a guild!) then it's painless to use the 'choke' of a manager class to resolve any issues.
Even in the commonplace situation, I would have thought it useful to use overseer-objects whose job it is to remove the extra (unnecessary) information from the problem before trying to solve it... There's no need to care about the avatar in sector (-1000,-1000) if we're currently in sector (0,0)...
It's a cliche, but the rule is 'divide and conquer'. Screaming and leaping is a satisfactory, but usually fatal approach to problem solving, unless you're Kzin.
Steven had posited in the 70's that the black holes leak (Hawking radiation), but the paradox is that they radiate a 'black-body' spectrum (entirely thermal radiation) in inverse proportion to their mass (so as they get smaller, the radiation increases). The problem here is that all the information went in, but it's very difficult to infer information from a black-body radiated spectrum (!). Steven therefore thinks that information is lost forever.
The article though is a bit hand-wavy over why the information is preserved in this new theory... (I guess Nth dimensional maths doesn't appeal to the reporter:-). I don't think the fact that the string-theory radius matches the black-hole radius is sufficient to prove the case, though it's an interesting pointer, a curious coincidence if indeed it is such...
Effectively this is a conjecture - if the strings continue to exist, then they'd have the same size as the black hole appears to have. The throwaway statement " That means a black hole can be traced back to its original conditions, and information survives." seems a bit of a stretch though:-)
The 1.1 license is not GPL-compatible. To avoid new issues with application programs that may be licensed under the GPL, the 1.1 licence is not being applied to client side libraries.
So, it seems that the main reason for a fork is no longer an issue ? No-one is going to be writing a new X-Server (well, I guess some of the embedded folks might, but that's about all I can think of), and they state that there's no issues with any client programs that you link with... No problem for most of us then, unless it's for political reasons.
My position is that if you write/own the code you get to say how it's used. I don't think there's *any* argument against that, and I can see why they want to promote themselves in this world where perception is all. The issue is that all decisions have consequences - which may be why client-programs are not part of the deal:-) I seriously doubt that XFree86 *want* a code fork, and I think that freedesktop.org will give them a serious run for their money if the fork goes ahead.
I wonder if the forking argument itself (please say that correctly:-) has gathered sufficient momentum to cause the predicted split though - that would be a pity if so. For all that KDE and Gnome are competing desktops, and they have both co-evolved to their benefit over time, I think two competing windowing-system standards might have a harder time co-evolving... If they didn't, you'd have to wonder why there were 2 in the first place!
I've always considered coding to be an artistic pursuit - the perfect form, in coding's case, is the elegant form IMHO - the creation of the simplest tool to do the job well (and fulfill the requirements spec, of course:-). In music, the art is the whole expression: the rise and fall in volume, the tempo changes, the different instruments, the silences, the mood-creation. Music is the pursuit of immersion. Coding is the pursuit of elegance. At least for me.
On the other hand, I can't really see "Spreadsheet in D minor" becoming too popular... entering incrementing data by performing a crescendo on the keyboard will take a while to catch on:-)
So whereas there are similarities, I think there are differences too, and I think the two input mechanisms reflect that. There is the other point that not all of us are maestro's with a musical instrument... the user-interface of the ivories might be slightly less user-friendly than the traditional QWERTY (or AZERTY, or whatever is your poison:-)
It's a shame that my first thoughts on this are: as soon as it becomes popular, it'll be used commercially, and start to lose its appeal. It's easy to see the commercialism of the web in the same light as the agents viewed humanity, in the Matrix - a plague.
Consider what you use the internet for, and how it's changed:
email was once a useful tool, now it's a spamfest. Still useful to me, but going downhill rapidly.
Webpages used to be information sources - can you believe there was an argument once over whether markup tags should be for pixel-perfect layout or for meta-information like TeX ? How naive is that ? As for the intrusive adverts that take over your screen, the less said the better. I will never buy anything from anyone who does this - I will seek out a more expensive competitor if necessary...
The more-successful protocols - those that actually deliver information are those left commercially-free. FTP is pretty basic, but you get what you want and nothing else. Usenet news has flamewars galore, but the limitations on what can be posted in non-binary groups actually seem to work well.
When I first started using the web, I set up a website for my image-processing postgrad group. We emailed CERN to let them know there was another website on the net:-) Imagine that today [grin]. The point is that I've seen the degeneration of the net into what it's become, and it's a sad story. Let's just hope that with this medium (the content being provided by lots of people rather than a concentrated few) we can buck the trend...
... so I guess he gets charged under the Patriot act. I don't condone what he did - he deserves a damn good kicking (metaphorically speaking, of course) for taking time from the emergency services, but a TERRORIST ? WTF ?
I can't believe there's not a more-appropriate crime to charge the guy with. Is there some sort of requirement to charge him with the most-serious charge you can, in the USA ? Perhaps that would explain it ?
There's little-to-no HDTV over here. The only place I've seen it in fact is in post-production studios, where they'll use it as a master-format.... Pity:-(
Which is the point, of course. People do feel better if they think they've "done something" about any problem - even one as trivial as opening the currently-closed door. Passively waiting for something to happen doesn't come easily to many of us...
What I find odd are those who hit the 'summon elevator' button more than once - A lift algorithm isn't going to take into account the number of times you press, and I doubt the buttons are pressure-sensitive:-) In fact, it's more likely that the time-of-last-press is one of the inputs to the algorithm, with earlier presses having a priority. So the more you press, the less chance you have of getting a busy lift:-))
Well, this is massively off-topic. I'd reply in email, but you're being cowardly...
The reason it's "forgetting" you is probably due to you having a dynamically-assigned IP address, and it changing from time-to-time. If you come in with the same IP address, I guarantee it will remember you - if you keep returning with different IP addresses, I don't see how it can "remember" you...
I actually quite like the Newton, though I think it was a bit ahead of its time and tried to do too much. It's a cool hack to get it playing Nintendo games, but would you really walk around with a bag to hold the 'pad, to play games on ?
It's useful when allied with a briefcase. I can't see it really as a games platform (on the other hand, my phone plays Doom quite well, (Nokia 9000:-) so whatever floats your boat - the phone's easier to carry though:-)
So Ballmer is saying "It's more expensive". I'm pretty sure it was supposed to be more expensive - MS were the cheaper (initial cost) of the two solutions for Munich, in fact the article more or less says this.
So what exactly is this article, apart from a chance for MS to spin the loss of some major business into more fear, uncertainty and doubt ?
Webmonkey was pretty good - I remember referring to it now and then... I know we can use the wayback machine / google (at least for a short while) etc. but if they're closing the doors, are they doing anything about the content ? Seems a shame to throw it away... Nothing mentioned on the site:-(
Er, you take my comments out of context then proceed to go on a rant about something that I didn't say (!)
I wrote:
It seems harsh if they were exposed with the company knowing about it, that they were denied some compensation for that
Note the emboldened bit that you conveniently removed and replaced with ellipsis, to completely change the sense of what I said. The point was that IBM allegedly knew the effects of the chemicals on the workers, but let them work in those conditions anyway.
Later I said if they were found against, you have to assume that they couldn't prove it and I always think it's easy to make a snap judgement based on your feelings for the parties involved though - In other words, the court had the full facts, and it's dangerous for outsiders to make emotional decisions without being in possession of those facts.
... if they truly were exposed with the company knowing about it, that they were denied some compensation for that. It seems the judge did them no favours by allowing them to take their case outside the California employees system, because the burden of proof is much higher in a "normal" court (as well as the damages...)
I wonder, if they'd stayed put, would they have won something rather than nothing... I always think it's easy to make a snap judgement based on your feelings for the parties involved though - if they were found against, you have to assume that they couldn't prove it... Harsh, though.
Entropy in statistical mechanics and entropy in information theory are not the same thing
I'm aware of that, but the mathematical derivation works fine. I was trying to demonstrate the point without resorting to some pretty unfriendly maths and the information-theory version comes to the same equations via a far less obvious route... I did say it was a 'hand-waving' argument...
1) You're invoking Shannon, whose work provides a law limiting the amount of information that can be sent down a channel with a given capacity. I never mentioned capacity - I said two things were the inverse of each other, and they essentially are. Shannon's work obviously still applies, if limits are placed on capacity.
2) Information is indeed the inverse of entropy. Consider the standard example: Two glass enclosures, one (Y) is a perfect vacuum, the other *X) has 1 million Hydrogen (H2) molecules in it and one Helium (He) atom.
At the beginning of the thought-experiment, the probability that the He atom is in container X [p(X)] is 1.0, and the probability of it being in container Y [p(Y)] is 0.0
Now introduce a path between the two containers. Due to statistical mechanics, (and more coarsely, thermodynamics), the gas in X will migrate to Y over time. As it does so, the probability p(X) drops, and p(Y) increases, until an equilibrium is reached where p(X) = p(Y) = 0.5. There is a linear relationship between p(X) and p(Y) throughout this time.
The point of equilibrium is the point of maximum entropy and the point of minimum information. Conversely, the starting conditions are the point of minimum entropy, and maximum information.
You therefore have a starting state where entropy is a minimum and information a maximum. You have a final state where entropy is a maximum and information a minimum. You have a linear progression between the two points. This is an inverse relationship between information and entropy.
QED.
Perhaps I should point out that my PhD was in information processing using Neural Networks. The information contained within the feature-sets used was characterised by loss-of-entropy within those feature sets as information increased. Ignoring the hand-waving argument above, I could probably still do the maths to prove the case as well... Trust me that the argument above is a *lot* easier to follow...
Simon.
Re:Inexpensive for testing purposes,
on
Mini-ITX Clustering
·
· Score: 2, Interesting
I don't think they equate the same. I said CPU power, not measured performance. I remember sitting on a UK working group panel debating the Block-Synchronous Parallel computing strategy for highly-parallel systems. I was only there because it was good for my CV:-) But I did learn a reasonable amount, all those years ago...
That said, the only time a cluster of servers will do better than a fast single node is when the task divides well over the cluster. Great for clustered webservers, even distributed databases (in fact most server processes), but pretty damn useless if you're trying to do interactive work, or calculate something which *doesn't* divide well. Anything with time-dependent processing (ie: you need the results of the last step to calculate the current one) will run as slow as your fastest node, minus some for overhead...
This doesn't dispute your point of course, but I think the sense of how you said it over-stated the case for the usefulness of the system.
Simon
Inexpensive for testing purposes,
on
Mini-ITX Clustering
·
· Score: 3, Insightful
... but that's about all it'll be useful for. A Nehemiah CPU is really weedy by todays standards, even the 1GHz one is about the same as a 600MH P3. So, he's got 12 of them, which is probably less CPU power than an average dual P4 motherboard...
Still, you can get some stats on how the clustering works, what's the best algorithm for dispersing problems, and these boards are cheap, but that's about the only advantage I can see...
This is a fabulous marketing manouvre. It's completely ludicrous of course, but it makes the connection between not-upgrading and being-vulnerable in the pointy-haired heads.
There *must* however be laws against making statements *that* outrageous...
This is essentially a problem in pattern recognition, and it's a damn hard problem to solve because of the disparity between the high-volume and low-volume words.
Information is essentially the inverse of entropy. Entropy can be calculated, and you can use Bayes probability theory to get a hold on the information content of a given word within a set of words.
What is difficult to do, and what search engines are trying to do, is measure the mutual information inherent between the set of pages that the word appears in, and the word itself, then apply that to all the words in the searched-for phrase; this is commonly called 'context'. This is plainly impossible to do for every given phrase, for every word combination, for every page indexed. The best you can do is use a statistical approach (and Bayes is your friend again) to come up with "good" matches.
The problem with the statistical approach is the class unbiasing, since once you have wildly different statistical populations, your choice of context gets harder and harder - the "easy" standard models don't cope very well. You don't have the computational resources to do a good analysis, so you're essentially stuck between a rock and a hard place.
This is why the google idea of strengthening the importance of a word depending on linked pages was such a good one - it "did" the hard work by relying on the entire planet to do it for them, by creating links. Of course, what one man can do, another can undo, and Google has got progressively worse over time. It's still by-far the best though, and my search engine of choice. When you look at the queries from search-sites, I get 100x as many from Google as Yahoo (next nearest)....
People think searching is easy, and it is. What's really really hard is searching *well*.
Slashdot Mirror
What's to stop these ghosts being maliciously "trained" to give the wrong answer... I remember a teacher at college (Mr Tittershill), who was routinely used in a joke on freshers (report to Mr Boobershill at the senior common room, NOW! ...)
:-)) [note to US authorities - this is a joke, and I have no intention of committing any crimes (cyber- or otherwise) when visiting the USA]
Is it only me who first thinks of "how to game the system" when presented with a new technology ? Perhaps I should have been a hacker
Simon
Oh yeah, and (from the article):
Simon
Surely this is a classic example of the Manager pattern. You have a bunch of objects [Avatar] (all alike, at least programmatically
The trade-off in terms of scalability is in frequency versus computation. If the operation is commonplace (such as moving around), then a distributed system has a problem. If the operation is not commonplace (such as joining a guild!) then it's painless to use the 'choke' of a manager class to resolve any issues.
Even in the commonplace situation, I would have thought it useful to use overseer-objects whose job it is to remove the extra (unnecessary) information from the problem before trying to solve it... There's no need to care about the avatar in sector (-1000,-1000) if we're currently in sector (0,0)...
It's a cliche, but the rule is 'divide and conquer'. Screaming and leaping is a satisfactory, but usually fatal approach to problem solving, unless you're Kzin.
Simon
Steven had posited in the 70's that the black holes leak (Hawking radiation), but the paradox is that they radiate a 'black-body' spectrum (entirely thermal radiation) in inverse proportion to their mass (so as they get smaller, the radiation increases). The problem here is that all the information went in, but it's very difficult to infer information from a black-body radiated spectrum (!). Steven therefore thinks that information is lost forever.
:-). I don't think the fact that the string-theory radius matches the black-hole radius is sufficient to prove the case, though it's an interesting pointer, a curious coincidence if indeed it is such ...
:-)
The article though is a bit hand-wavy over why the information is preserved in this new theory... (I guess Nth dimensional maths doesn't appeal to the reporter
Effectively this is a conjecture - if the strings continue to exist, then they'd have the same size as the black hole appears to have. The throwaway statement " That means a black hole can be traced back to its original conditions, and information survives." seems a bit of a stretch though
Simon
So, it seems that the main reason for a fork is no longer an issue ? No-one is going to be writing a new X-Server (well, I guess some of the embedded folks might, but that's about all I can think of), and they state that there's no issues with any client programs that you link with
My position is that if you write/own the code you get to say how it's used. I don't think there's *any* argument against that, and I can see why they want to promote themselves in this world where perception is all. The issue is that all decisions have consequences - which may be why client-programs are not part of the deal
I wonder if the forking argument itself (please say that correctly
Simon.
I've always considered coding to be an artistic pursuit - the perfect form, in coding's case, is the elegant form IMHO - the creation of the simplest tool to do the job well (and fulfill the requirements spec, of course :-). In music, the art is the whole expression: the rise and fall in volume, the tempo changes, the different instruments, the silences, the mood-creation. Music is the pursuit of immersion. Coding is the pursuit of elegance. At least for me.
:-)
:-)
On the other hand, I can't really see "Spreadsheet in D minor" becoming too popular... entering incrementing data by performing a crescendo on the keyboard will take a while to catch on
So whereas there are similarities, I think there are differences too, and I think the two input mechanisms reflect that. There is the other point that not all of us are maestro's with a musical instrument... the user-interface of the ivories might be slightly less user-friendly than the traditional QWERTY (or AZERTY, or whatever is your poison
Simon
Consider what you use the internet for, and how it's changed:
The more-successful protocols - those that actually deliver information are those left commercially-free. FTP is pretty basic, but you get what you want and nothing else. Usenet news has flamewars galore, but the limitations on what can be posted in non-binary groups actually seem to work well.
When I first started using the web, I set up a website for my image-processing postgrad group. We emailed CERN to let them know there was another website on the net
Simon
... so I guess he gets charged under the Patriot act. I don't condone what he did - he deserves a damn good kicking (metaphorically speaking, of course) for taking time from the emergency services, but a TERRORIST ? WTF ?
I can't believe there's not a more-appropriate crime to charge the guy with. Is there some sort of requirement to charge him with the most-serious charge you can, in the USA ? Perhaps that would explain it ?
Simon.
There's little-to-no HDTV over here. The only place I've seen it in fact is in post-production studios, where they'll use it as a master-format.... Pity :-(
Simon
Which is the point, of course. People do feel better if they think they've "done something" about any problem - even one as trivial as opening the currently-closed door. Passively waiting for something to happen doesn't come easily to many of us...
:-) In fact, it's more likely that the time-of-last-press is one of the inputs to the algorithm, with earlier presses having a priority. So the more you press, the less chance you have of getting a busy lift :-))
What I find odd are those who hit the 'summon elevator' button more than once - A lift algorithm isn't going to take into account the number of times you press, and I doubt the buttons are pressure-sensitive
Simon
Posted with no karma bonus...
Well, this is massively off-topic. I'd reply in email, but you're being cowardly...
The reason it's "forgetting" you is probably due to you having a dynamically-assigned IP address, and it changing from time-to-time. If you come in with the same IP address, I guarantee it will remember you - if you keep returning with different IP addresses, I don't see how it can "remember" you...
Simon.
I actually quite like the Newton, though I think it was a bit ahead of its time and tried to do too much. It's a cool hack to get it playing Nintendo games, but would you really walk around with a bag to hold the 'pad, to play games on ?
:-) so whatever floats your boat - the phone's easier to carry though :-)
It's useful when allied with a briefcase. I can't see it really as a games platform (on the other hand, my phone plays Doom quite well, (Nokia 9000
Simon
So Ballmer is saying "It's more expensive". I'm pretty sure it was supposed to be more expensive - MS were the cheaper (initial cost) of the two solutions for Munich, in fact the article more or less says this.
So what exactly is this article, apart from a chance for MS to spin the loss of some major business into more fear, uncertainty and doubt ?
Simon.
Webmonkey was pretty good - I remember referring to it now and then... I know we can use the wayback machine / google (at least for a short while) etc. but if they're closing the doors, are they doing anything about the content ? Seems a shame to throw it away... Nothing mentioned on the site :-(
Simon
Er, you take my comments out of context then proceed to go on a rant about something that I didn't say (!)
I wrote:
It seems harsh if they were exposed with the company knowing about it, that they were denied some compensation for that
Note the emboldened bit that you conveniently removed and replaced with ellipsis, to completely change the sense of what I said. The point was that IBM allegedly knew the effects of the chemicals on the workers, but let them work in those conditions anyway.
Later I said if they were found against, you have to assume that they couldn't prove it and I always think it's easy to make a snap judgement based on your feelings for the parties involved though - In other words, the court had the full facts, and it's dangerous for outsiders to make emotional decisions without being in possession of those facts.
Jeez.
Simon
... if they truly were exposed with the company knowing about it, that they were denied some compensation for that. It seems the judge did them no favours by allowing them to take their case outside the California employees system, because the burden of proof is much higher in a "normal" court (as well as the damages...)
I wonder, if they'd stayed put, would they have won something rather than nothing... I always think it's easy to make a snap judgement based on your feelings for the parties involved though - if they were found against, you have to assume that they couldn't prove it... Harsh, though.
Simon
The spamhaus website has been listing the USA for a loooong time now as the #1 spam source. It's got the names of the top spammers there too...
Simon
I'm aware of that, but the mathematical derivation works fine. I was trying to demonstrate the point without resorting to some pretty unfriendly maths and the information-theory version comes to the same equations via a far less obvious route... I did say it was a 'hand-waving' argument...
Simon
Two things here:
1) You're invoking Shannon, whose work provides a law limiting the amount of information that can be sent down a channel with a given capacity. I never mentioned capacity - I said two things were the inverse of each other, and they essentially are. Shannon's work obviously still applies, if limits are placed on capacity.
2) Information is indeed the inverse of entropy. Consider the standard example: Two glass enclosures, one (Y) is a perfect vacuum, the other *X) has 1 million Hydrogen (H2) molecules in it and one Helium (He) atom.
At the beginning of the thought-experiment, the probability that the He atom is in container X [p(X)] is 1.0, and the probability of it being in container Y [p(Y)] is 0.0
Now introduce a path between the two containers. Due to statistical mechanics, (and more coarsely, thermodynamics), the gas in X will migrate to Y over time. As it does so, the probability p(X) drops, and p(Y) increases, until an equilibrium is reached where p(X) = p(Y) = 0.5. There is a linear relationship between p(X) and p(Y) throughout this time.
The point of equilibrium is the point of maximum entropy and the point of minimum information. Conversely, the starting conditions are the point of minimum entropy, and maximum information.
You therefore have a starting state where entropy is a minimum and information a maximum. You have a final state where entropy is a maximum and information a minimum. You have a linear progression between the two points. This is an inverse relationship between information and entropy.
QED.
Perhaps I should point out that my PhD was in information processing using Neural Networks. The information contained within the feature-sets used was characterised by loss-of-entropy within those feature sets as information increased. Ignoring the hand-waving argument above, I could probably still do the maths to prove the case as well... Trust me that the argument above is a *lot* easier to follow...
Simon.
I don't think they equate the same. I said CPU power, not measured performance. I remember sitting on a UK working group panel debating the Block-Synchronous Parallel computing strategy for highly-parallel systems. I was only there because it was good for my CV :-) But I did learn a reasonable amount, all those years ago...
That said, the only time a cluster of servers will do better than a fast single node is when the task divides well over the cluster. Great for clustered webservers, even distributed databases (in fact most server processes), but pretty damn useless if you're trying to do interactive work, or calculate something which *doesn't* divide well. Anything with time-dependent processing (ie: you need the results of the last step to calculate the current one) will run as slow as your fastest node, minus some for overhead...
This doesn't dispute your point of course, but I think the sense of how you said it over-stated the case for the usefulness of the system.
Simon
... but that's about all it'll be useful for. A Nehemiah CPU is really weedy by todays standards, even the 1GHz one is about the same as a 600MH P3. So, he's got 12 of them, which is probably less CPU power than an average dual P4 motherboard...
Still, you can get some stats on how the clustering works, what's the best algorithm for dispersing problems, and these boards are cheap, but that's about the only advantage I can see...
Simon
This is a fabulous marketing manouvre. It's completely ludicrous of course, but it makes the connection between not-upgrading and being-vulnerable in the pointy-haired heads.
There *must* however be laws against making statements *that* outrageous...
Simon.
Let's just hope this doesn't fortell a delay in the PS3 then :-(
Simon
This is essentially a problem in pattern recognition, and it's a damn hard problem to solve because of the disparity between the high-volume and low-volume words.
Information is essentially the inverse of entropy. Entropy can be calculated, and you can use Bayes probability theory to get a hold on the information content of a given word within a set of words.
What is difficult to do, and what search engines are trying to do, is measure the mutual information inherent between the set of pages that the word appears in, and the word itself, then apply that to all the words in the searched-for phrase; this is commonly called 'context'. This is plainly impossible to do for every given phrase, for every word combination, for every page indexed. The best you can do is use a statistical approach (and Bayes is your friend again) to come up with "good" matches.
The problem with the statistical approach is the class unbiasing, since once you have wildly different statistical populations, your choice of context gets harder and harder - the "easy" standard models don't cope very well. You don't have the computational resources to do a good analysis, so you're essentially stuck between a rock and a hard place.
This is why the google idea of strengthening the importance of a word depending on linked pages was such a good one - it "did" the hard work by relying on the entire planet to do it for them, by creating links. Of course, what one man can do, another can undo, and Google has got progressively worse over time. It's still by-far the best though, and my search engine of choice. When you look at the queries from search-sites, I get 100x as many from Google as Yahoo (next nearest)....
People think searching is easy, and it is. What's really really hard is searching *well*.
Simon
Yay
Simon.