I truly think that AI will be a revolution in mathematics rather than in computing, some bastard child of calculus, statistics and decision trees.
This is quite a reasonable description. It clearly will be something new rather than an adaption of something known. But just because we've tried all of the known things for the past fifty years. The interface in machine learning between logic and stats is quite interesting at the moment, with some pretty cool work coming out of it. So the "breakthrough" (probably won't be such a singular thing, if you'll forgive the terrible pun:) is going to be somewhere in the area of that bastard child.
Thanks for the reasoned response. I had written you off as yet another singularitist with blinkers, but you've proven me quite wrong on that one. Personally (when I'm not playing devil's advocate) I come down on the strong-AI side of the debate, it's just the timing part that pisses me off. Kurweil uses this idea that we're on some straight line to an AI revolution driven just increases in processing power, and he does it mainly to sell books to the masses. It will happen at some point (my own prejudices there:), but predicting when is possibly futile.
The main reason that it will be solved is that a) I don't buy the weak AI arguments, and b) it is the most interesting hard problem facing mankind at the moment, and that tends to motivate a lot of smart people into working on it.
Ok, according to moore's law we will get there, with a transistor based computer.
It's nice that you're taking the time to point out the flaws in the original troll post's religion. But it is a shame that you have to resort to another religion to do so. Without getting into the argument about whether or not it is wise to extrapolate short-term trends on curved graphs, and ignoring the curve-jumping shit that Kurzweil has resorted to, lets just pretend that we will have hte computing power that you assume in a few decades.
Even if I give you an infinitely powerful chunk of hardware, it doesn't get you a single step closer to solving the problems in AI. Hardware power is not the problem, and this whole idea that transistor based computing merely needs to catch up with the brain is a red herring. The transhumanists looks at other engineering problems that we have solved, and assumes that AI is the same sort of problem. It is not. It is not something that can be solved by throwing time, money, or computing power at it. The hardest problems in AI can be boiled down to things that are incredibley simple and fundamental, and yet are completely beyond our current understanding. We don't even know how to express the problems in a way that we can attack them, let alone work on ways of solving them.
If you believe the transhumanist manifesto then I would recommend that you spend some timing looking at AI and machine learning research. Perhaps (re)read Godel, Escher and Bach. Hoffstader does a very good job of explaining why the hardness of some problems is measured on a different scale.
You make a good point, and I'm not going to disagree with you over whether or not WoWGlider is cheating. I don't play WoW, as I think grinding is not really gameplay (see the other reply that I made with a link to Gamasutra). If you see the game as a timesink, and progress as dependent on time "invested" rather than skill (which I guess you do) then sure, it's cheating.
But you can't sue someone over allowing people to cheat in your game. You can sue people over copyright violation, and that is what Vivendi has tried to do. So whether or not WoWGlider is cheating, I still think that his (legal) case is pretty tight.
Also, how is this different than say, cracking DVD decryption or bypassing copyright schema on software?
Well, the those are examples of copyright protection - they stop people from copying the product. WoWGlider is an example of software that people cheat with (I'm going with your argument on this one). Cheating is not a copyright violation.
I don't like grinding. I don't play MMOs - which I stated, so what's your point?
Perhaps you've decided that current MMO design is the absolute nadir, and that it cannot be progressed any further. Other people have decided that it can be improved. One of the issues with this WoW autopilot is that it is needed at all. Games are supposed to be fun, MMOs that require the player to grind to get anywhere are not fun - they're an addiction. In fact they are pretty close to optimal in terms of the reward schedule, which is why they are popular.
Before you hit the reply button to slate me because I disagree with you, try reading this. It does require a free-registration hassle, or a Bug me not. If you are at all interested in MMO design then it is an interesting read, and the opinions in it are a good reflection of how online gaming will shift over the next five years. The most important point is that grinding is not an addition to gameplay - it's a barrier
While this is all true and good they are not banning the defendant. He hasn't violated their terms and conditions as he is not playing the game. They are trying to sue him for copyright infringement because he makes and sells a bot. The pdf is quite interesting (although it uses the worst font I've ever read) and it sounds like he has a very tight case. Mainly because Vivendi are misrepresenting their position - they thought that a threat to file suit would make him fold. Instead it seems like he has explained the details of the case to a lawyer pretty well, and the document that he has filed seems to tear their case apart.
As far as cheating goes - bots for grinding in MMO games are an interesting case. This isn't an aimbot that helps you beat other players, or improves your abilities. It doesn't hack the client into thinking that you have more gold / resources that you really do. It just takes the tedious repetitive actions in the "game" and plays them through. It's an autopilot. The real question for me: is a game that requires an autopilot actually fun enough to play?
(I would say no, but given that he's made a profitable business out of it, lots of people must say yes).
No it definitely isn't good enough. It's the researcher calling it a forcefield that annoyed me in the first place, as strangely enough - it is not a fucking forcefield. But hey, call me picky.
Well done you. You quoted PBS nova, surely a great scientific publication and source of fact. And because you believe in the power of citation so much, you did so as AC. But lets look at your quote in more detail shall we:
Force Field: from a macroscopic perspective, the means by which a force communicates its influence; described by a collection of numbers at each point in space that reflect the strength and direction of the force at that point
This does not mean that any collection of numbers that describes force in a volume is a force-field, but rather that it can describe one. Much as a picture on my monitor can describe a Ferrari, but is rather hard to drive.
While your points are all true, as a scientist it was the description of "force-fields" in the article that really pissed me off. So we map points in the city into "force-fields", and then we "simulate" the "force-fields" with haptic input. So.... in effect there are no force-fields - only geometry which the haptic device then interacts with. Whoever wrote this article is a first class dickwad.
On the subject of visible assumptions, we have a blind guy doing some research into haptic interfaces as part of his PhD. Every so often the department gets the chance to try one of his experiments and the results are odd to say the least. As someone with sight I would assume that most information comes from shape and size, apparently these are secondary cues to the user of a haptic interface. I shouldn't really go into too many details as I'm not sure what he's published and what he hasn't.
Hmmm, I have to agree with the wiki article as I worked on ISDN and SS#7 switches and gateways for five years. Haven't poked around inside the depths of an ISDN stack I can tell you that it is definitely a circuit-switched network in the manner that the wiki describes. Are you perhaps thinking of something else?
No it wouldn't. The internet is a packet-switched network designed for computers. ISDN is a circuit-switched network designed to carry calls. They are very different.
Well I haven't made it work yet:) But I have a very weird bug in the nvidia driver that only seems to affect my configuration out of the whole world. Basically if you can't switch SBA off in the driver then hibernation won't work. If you (like the rest of the world) can then hibernation is very easy to setup and works likes a charm. I haven't played with the WakeOnLAN part as currently it would be a little pointless for me.
I'd agree with your points about the cost of making a new machine versus the savings; but instead of saying don't bother I'd recommend playing with the software a little. Lots of people have "24/7" servers at home for the same applications, but how many people actually use them 24/7? With me it is more a case that I want access 24/7 even if I just use it a few hours a day on average. Have a look at a hibernation kernel and WakeOnLAN. If you can bring the server up to a good point remotely with just 30sec latency then the real power saving is having it shutdown 80% of the time. Using a longhaul governer and scripts you can ensure that it stays up when it needs to (ie during a bittorrent download) but then shuts down when there are no active processes on the system.
Yes they really would care. It would completely destroy casual gaming. Clicking on an icon for a quick 20min blast versus having to reboot to play the game. From my experience of a dual-boot with cedega on the linux system I can say that I play games more if they less hassle to launch. Bootable games disks would be a throwback to the stoneage.
That probably isn't a bug. Most file-systems don't lock files that you are executing, so they can be overwritten whilst mapped into memory. This can abused in lots of amusing ways.
That's not what he means by scatter. GPUs perform stream processing, so imagine a stream of data coming into a GPU program and a stream of data coming out. The basic "shape" is something like this:
for(int i=0; i<size; i++)
out[i] = f( in[i] );
Where f is some arbitrary function. Note that the positions in the array cannot be changed - these corrospond to the xy positions of the target pixels. For a scatter operation (very useful in simulations, sieving etc):
Now the results of each computation can be directed to where they need to go. Without this operation you either need to recode your algorithm to work inplace, where the performance hit comes, or resort your data afterwards. On a GPU sorting is very expensive so most people recode. This problem should disappear on DX10 hardware because the new geometry shaders can be used to do real scatter operations. At the moment people try and fake this with multiple pixel->vertex>pixel->vertex passes but the point cloud performance sucks on a device optimised for rasterisation.
Yes. Yes it is. Out there are people willing to spend money on gold-plated scart connectors. I say the more overhyped, overpriced pieces of junk on the market to separate these fools and their cash the better.
I have to ask if you are trolling now or are actually being serious. I hope that you are trolling and I have to say that you made me laugh out loud. In particular your misuse of statistical significance is a comedic masterpiece.
If I am looking at a dataset containing several million items and I'm trying to establish that X is popular, finding that it has occured at least once is not evidence. I think that you'll find this is the common dictionary definition; observations that support a hypothosis. Or maybe you're not using the dictionary definition of popularity?
So if I buy a new computer and swap the CPU for one from an old machine, then does the combination become the old computer?
If it isn't obvious why it takes more thought than that then consider your own body. You shed cells all the time and grow new ones. At what point (if any) do you stop being the "old you"?
Your first response came across as rude and arrogant. But you're right that name-calling is childish and unnecessary. My bad.
I'm sure that my system is the exception, but that was my point. I think that the AC who replied grasped it fairly well. The article covers why when you take averages and trends from any complex data it is possible to make them support any point that you want. In this case there is some fairly good evidence that Apple has been on the receiving end of these dodgy statistics.
The point that I was replying (and taking exception) to was attempting to defend Apple's sales figures with a single arbitrary datum. Anecdotal evidence is largely useless for addressing bad statistics and there are better ways of doing it.
While I mentioned that my "system" had been around for eight years, what I didn't mention is that it hasn't always been working. Any electronic device that I've personally tinkered with is inherently dangerous, and sometimes I do long for the uncomplicated Apple box that simply works. But then part of me realises that it just wouldn't be as much fun for me... Either that or it's just the maschistic part of me that insists that I run Gentoo on the box made from random parts.
Dear oh dear. Where to start. Caustically perhaps?
Anecdotal Evidence is an oxymoron and should always be treated as such. If I know a single person who has had X happen to them this does not give me any evidence to infer the popularity of X. In short, my comparison of "anecdotal evidence" with a singular datum is entirely correct. You seem to have difficulty with logical reasoning, so perhaps justifying statistics is a little bit too advanced for you.
If I mention that X happened exactly once, then it is true that it does not imply that X has not happened multiple times. However it is also true that if I do not mention X at all then it does not imply that X has not happened multiple time. In fact me mentioning X, supposedly as evidence of the frequent occurance of X, is completely unrelated to these other alleged data points. If you could understand logic then I would be able to explain to you that one occurance of X is independent of the frequent occurance of X. But we've already established there is no point in doing so. This is why annecdotal "evidence" is worthless for anything other than sheer entertainment.
Finally, your reading comprehension is a wee bit shit. I have not had the same case for 8 years, as I clearly stated. Factually I have never replaced the computer at all; merely all of the components within in. The AC who replied first understood the point far more clearly than you ever will.
It is interesting that if you pull a figure out of your arse (4 times) and then use it in an equation (8 years / 4 times) then you can appear to make a point. At least this explains your defense of ancedotal evidence. Given that the article was about the market share of Macs being underestimated by dodgy statistics then I must assume that you graduated in advanced irony.
No. It means that I've chosen to replace components because as commodity goods it's cheap enough to do this.
When you claim that my anecdotal evidence supports anything it means that you weren't capable of understanding the message that you've replied to. Makes me wonder if you even know what QED means, or if its just something you thought would look cool on the end of a message. Are you a Queen Examining Dicks perhaps?
Slashdot Mirror
This is quite a reasonable description. It clearly will be something new rather than an adaption of something known. But just because we've tried all of the known things for the past fifty years. The interface in machine learning between logic and stats is quite interesting at the moment, with some pretty cool work coming out of it. So the "breakthrough" (probably won't be such a singular thing, if you'll forgive the terrible pun
Thanks for the reasoned response. I had written you off as yet another singularitist with blinkers, but you've proven me quite wrong on that one. Personally (when I'm not playing devil's advocate) I come down on the strong-AI side of the debate, it's just the timing part that pisses me off. Kurweil uses this idea that we're on some straight line to an AI revolution driven just increases in processing power, and he does it mainly to sell books to the masses. It will happen at some point (my own prejudices there
The main reason that it will be solved is that a) I don't buy the weak AI arguments, and b) it is the most interesting hard problem facing mankind at the moment, and that tends to motivate a lot of smart people into working on it.
It's nice that you're taking the time to point out the flaws in the original troll post's religion. But it is a shame that you have to resort to another religion to do so. Without getting into the argument about whether or not it is wise to extrapolate short-term trends on curved graphs, and ignoring the curve-jumping shit that Kurzweil has resorted to, lets just pretend that we will have hte computing power that you assume in a few decades.
Even if I give you an infinitely powerful chunk of hardware, it doesn't get you a single step closer to solving the problems in AI. Hardware power is not the problem, and this whole idea that transistor based computing merely needs to catch up with the brain is a red herring. The transhumanists looks at other engineering problems that we have solved, and assumes that AI is the same sort of problem. It is not. It is not something that can be solved by throwing time, money, or computing power at it. The hardest problems in AI can be boiled down to things that are incredibley simple and fundamental, and yet are completely beyond our current understanding. We don't even know how to express the problems in a way that we can attack them, let alone work on ways of solving them.
If you believe the transhumanist manifesto then I would recommend that you spend some timing looking at AI and machine learning research. Perhaps (re)read Godel, Escher and Bach. Hoffstader does a very good job of explaining why the hardness of some problems is measured on a different scale.
But you can't sue someone over allowing people to cheat in your game. You can sue people over copyright violation, and that is what Vivendi has tried to do. So whether or not WoWGlider is cheating, I still think that his (legal) case is pretty tight.
Well, the those are examples of copyright protection - they stop people from copying the product. WoWGlider is an example of software that people cheat with (I'm going with your argument on this one). Cheating is not a copyright violation.
I don't like grinding. I don't play MMOs - which I stated, so what's your point?
Perhaps you've decided that current MMO design is the absolute nadir, and that it cannot be progressed any further. Other people have decided that it can be improved. One of the issues with this WoW autopilot is that it is needed at all. Games are supposed to be fun, MMOs that require the player to grind to get anywhere are not fun - they're an addiction. In fact they are pretty close to optimal in terms of the reward schedule, which is why they are popular.
Before you hit the reply button to slate me because I disagree with you, try reading this. It does require a free-registration hassle, or a Bug me not. If you are at all interested in MMO design then it is an interesting read, and the opinions in it are a good reflection of how online gaming will shift over the next five years. The most important point is that grinding is not an addition to gameplay - it's a barrier
Exactly, how many people would pay for an .unsafe tld?
.com.safe, .gov.safe, .net.safe....
.unsafe domain and we're done!
So once 95% of all websites decide that they want to be safe, how do organise the namespace? How about
Then all we do is turn off the
While this is all true and good they are not banning the defendant. He hasn't violated their terms and conditions as he is not playing the game. They are trying to sue him for copyright infringement because he makes and sells a bot. The pdf is quite interesting (although it uses the worst font I've ever read) and it sounds like he has a very tight case. Mainly because Vivendi are misrepresenting their position - they thought that a threat to file suit would make him fold. Instead it seems like he has explained the details of the case to a lawyer pretty well, and the document that he has filed seems to tear their case apart.
As far as cheating goes - bots for grinding in MMO games are an interesting case. This isn't an aimbot that helps you beat other players, or improves your abilities. It doesn't hack the client into thinking that you have more gold / resources that you really do. It just takes the tedious repetitive actions in the "game" and plays them through. It's an autopilot. The real question for me: is a game that requires an autopilot actually fun enough to play?
(I would say no, but given that he's made a profitable business out of it, lots of people must say yes).
Nobody is claiming to have invented, or patented, a font. At least read the summary if you're not going to read the article.
No it definitely isn't good enough. It's the researcher calling it a forcefield that annoyed me in the first place, as strangely enough - it is not a fucking forcefield. But hey, call me picky.
Well done you. You quoted PBS nova, surely a great scientific publication and source of fact. And because you believe in the power of citation so much, you did so as AC. But lets look at your quote in more detail shall we:
Force Field: from a macroscopic perspective, the means by which a force communicates its influence; described by a collection of numbers at each point in space that reflect the strength and direction of the force at that point
This does not mean that any collection of numbers that describes force in a volume is a force-field, but rather that it can describe one. Much as a picture on my monitor can describe a Ferrari, but is rather hard to drive.
While your points are all true, as a scientist it was the description of "force-fields" in the article that really pissed me off. So we map points in the city into "force-fields", and then we "simulate" the "force-fields" with haptic input. So.... in effect there are no force-fields - only geometry which the haptic device then interacts with. Whoever wrote this article is a first class dickwad.
On the subject of visible assumptions, we have a blind guy doing some research into haptic interfaces as part of his PhD. Every so often the department gets the chance to try one of his experiments and the results are odd to say the least. As someone with sight I would assume that most information comes from shape and size, apparently these are secondary cues to the user of a haptic interface. I shouldn't really go into too many details as I'm not sure what he's published and what he hasn't.
Hmmm, I have to agree with the wiki article as I worked on ISDN and SS#7 switches and gateways for five years. Haven't poked around inside the depths of an ISDN stack I can tell you that it is definitely a circuit-switched network in the manner that the wiki describes. Are you perhaps thinking of something else?
No it wouldn't. The internet is a packet-switched network designed for computers. ISDN is a circuit-switched network designed to carry calls. They are very different.
Well I haven't made it work yet :) But I have a very weird bug in the nvidia driver that only seems to affect my configuration out of the whole world. Basically if you can't switch SBA off in the driver then hibernation won't work. If you (like the rest of the world) can then hibernation is very easy to setup and works likes a charm. I haven't played with the WakeOnLAN part as currently it would be a little pointless for me.
I'd agree with your points about the cost of making a new machine versus the savings; but instead of saying don't bother I'd recommend playing with the software a little. Lots of people have "24/7" servers at home for the same applications, but how many people actually use them 24/7? With me it is more a case that I want access 24/7 even if I just use it a few hours a day on average. Have a look at a hibernation kernel and WakeOnLAN. If you can bring the server up to a good point remotely with just 30sec latency then the real power saving is having it shutdown 80% of the time. Using a longhaul governer and scripts you can ensure that it stays up when it needs to (ie during a bittorrent download) but then shuts down when there are no active processes on the system.
Yes they really would care. It would completely destroy casual gaming. Clicking on an icon for a quick 20min blast versus having to reboot to play the game. From my experience of a dual-boot with cedega on the linux system I can say that I play games more if they less hassle to launch. Bootable games disks would be a throwback to the stoneage.
That probably isn't a bug. Most file-systems don't lock files that you are executing, so they can be overwritten whilst mapped into memory. This can abused in lots of amusing ways.
That's not what he means by scatter. GPUs perform stream processing, so imagine a stream of data coming into a GPU program and a stream of data coming out. The basic "shape" is something like this:
for(int i=0; i<size; i++)
out[i] = f( in[i] );
Where f is some arbitrary function. Note that the positions in the array cannot be changed - these corrospond to the xy positions of the target pixels. For a scatter operation (very useful in simulations, sieving etc):
for(int i=0; i<size; i++)
out[ g(i) ] = f( in[i] );
Now the results of each computation can be directed to where they need to go. Without this operation you either need to recode your algorithm to work inplace, where the performance hit comes, or resort your data afterwards. On a GPU sorting is very expensive so most people recode. This problem should disappear on DX10 hardware because the new geometry shaders can be used to do real scatter operations. At the moment people try and fake this with multiple pixel->vertex>pixel->vertex passes but the point cloud performance sucks on a device optimised for rasterisation.
Yes. Yes it is. Out there are people willing to spend money on gold-plated scart connectors. I say the more overhyped, overpriced pieces of junk on the market to separate these fools and their cash the better.
I have to ask if you are trolling now or are actually being serious. I hope that you are trolling and I have to say that you made me laugh out loud. In particular your misuse of statistical significance is a comedic masterpiece.
If I am looking at a dataset containing several million items and I'm trying to establish that X is popular, finding that it has occured at least once is not evidence. I think that you'll find this is the common dictionary definition; observations that support a hypothosis. Or maybe you're not using the dictionary definition of popularity?
Yes. It was in my comment. If you look up now you might just catch it as it sails over your head...
So if I buy a new computer and swap the CPU for one from an old machine, then does the combination become the old computer?
If it isn't obvious why it takes more thought than that then consider your own body. You shed cells all the time and grow new ones. At what point (if any) do you stop being the "old you"?
Your first response came across as rude and arrogant. But you're right that name-calling is childish and unnecessary. My bad.
I'm sure that my system is the exception, but that was my point. I think that the AC who replied grasped it fairly well. The article covers why when you take averages and trends from any complex data it is possible to make them support any point that you want. In this case there is some fairly good evidence that Apple has been on the receiving end of these dodgy statistics.
The point that I was replying (and taking exception) to was attempting to defend Apple's sales figures with a single arbitrary datum. Anecdotal evidence is largely useless for addressing bad statistics and there are better ways of doing it.
While I mentioned that my "system" had been around for eight years, what I didn't mention is that it hasn't always been working. Any electronic device that I've personally tinkered with is inherently dangerous, and sometimes I do long for the uncomplicated Apple box that simply works. But then part of me realises that it just wouldn't be as much fun for me... Either that or it's just the maschistic part of me that insists that I run Gentoo on the box made from random parts.
Dear oh dear. Where to start. Caustically perhaps?
Anecdotal Evidence is an oxymoron and should always be treated as such. If I know a single person who has had X happen to them this does not give me any evidence to infer the popularity of X. In short, my comparison of "anecdotal evidence" with a singular datum is entirely correct. You seem to have difficulty with logical reasoning, so perhaps justifying statistics is a little bit too advanced for you.
If I mention that X happened exactly once, then it is true that it does not imply that X has not happened multiple times. However it is also true that if I do not mention X at all then it does not imply that X has not happened multiple time. In fact me mentioning X, supposedly as evidence of the frequent occurance of X, is completely unrelated to these other alleged data points. If you could understand logic then I would be able to explain to you that one occurance of X is independent of the frequent occurance of X. But we've already established there is no point in doing so. This is why annecdotal "evidence" is worthless for anything other than sheer entertainment.
Finally, your reading comprehension is a wee bit shit. I have not had the same case for 8 years, as I clearly stated. Factually I have never replaced the computer at all; merely all of the components within in. The AC who replied first understood the point far more clearly than you ever will.
It is interesting that if you pull a figure out of your arse (4 times) and then use it in an equation (8 years / 4 times) then you can appear to make a point. At least this explains your defense of ancedotal evidence. Given that the article was about the market share of Macs being underestimated by dodgy statistics then I must assume that you graduated in advanced irony.
No. It means that I've chosen to replace components because as commodity goods it's cheap enough to do this.
When you claim that my anecdotal evidence supports anything it means that you weren't capable of understanding the message that you've replied to. Makes me wonder if you even know what QED means, or if its just something you thought would look cool on the end of a message. Are you a Queen Examining Dicks perhaps?