I don't think that's a minor niggle; it's a very good point! I was pretty sloppy in what I said. I actually wasn't even thinking about the dimensions, I was just considering rotation around the same axis, like what you'd be doing if you were spinning something in graphics.
I'm not sure I understand what you mean about angles not being directly useful. Rotation a given angle seems fundamental. If you are doing a POV game, and the person starts turning left, you're rotating the frame or camera by a definite angle each time step. If you want to grab a molecule in a protein visualization and spin it around, you'd be working with angles. Are you saying that angles are used only at a minimum, for rotating a few vectors on to which everything else is projected?
I agree with you. I should've been more explicit and stated that I was talking about general purpose digital computing with optics. Having said that, even the Fourier transform stuff that once looked so promising really doesn't look so good. How do you get the data in and out of the system? In the end, you need a spatial light modulator, and those are so damn slow that you can always compute a fourier transform faster on a DSP chip, despite the parallelism of light.
A photon has a wavefunction, that's it. Assuming the photon has a relatively sharp energy, you can then say it has a frequency (energy) and a phase. Amplitude is meaningless when speaking about a single photon.
I can only wait for the development of photonic processors.
You'll be waiting a long time, I think. People have been trying this foolhardy idea of optical computing for a long time, and the field is pretty much dead. The problem is that to get light to modulate other light, you still have to move electrons around. Photons don't interact directly with each other. So if you're moving electrons around, for one, you're back to the old problem of finding materials with fast state transitions.
But most importantly, nonlinear optical effects with degerate wavelengths (where inputs and output are the same color) are by definition, at least third order effects (in terms of perturbation theory where light is the perturbation on a quantum system). In other words, they are weak effects, and take a rather large distance (relative to a wavelength of light) to switch. Thus, while people will brag about this modulation speed and that, what they don't tell you is that (a) the device is big (on the order of many microns) and therefore (b) the latency sucks. This means you can't compute with it since your answer takes longer to come out the other end then a transistor would've taken. Of course, nobody tells this to us poor people who venture unwittingly into the field of optics as freshly graduated students who don't know any better. Because of all of this, optics is great for communications (where you modulate a ton in series, but don't need to wait for the output to feed into a next computation step) but really pretty bad for computation. So, the originaly article is right about clocking being a good match for optics, but all the user comments about optical computing should be taken with a grain of salt, I'm afraid.
How did you get the quandrance components? That's the part that's missing from this guy's scheme. He assumes you already have everything specified in his terms. So he's just offloaded the hard parts to "outside" the domain of his problem. In the end, you'll have transcendental operations before and after the operations you quoted in your example get things into his scheme. After all, if you want to spin something, you want to spin it in equal angle increments, not equal ratios of quandrances. So at the beginning, you'll need to convert an angle using trig functions. At the end, you'll have to take a square root to get real distances. I guess it's possible you'll have less total transendental ops, but I'm fairly certain the people that write graphics libraries and build hardware aren't stupid--they know how to minimize their use of transcendental ops and probably have already discovered everything this guy claims to have invented. I believe this guy's point is more pedagogical than anything.
I'm not dismissing this guy's work at all, I'm just saying that you can only avoid classical trig in specific problem domains.
I doubt it. In the end, the numerics are probably the same. Inside the computer, nobody computes "sines" they compute truncations of infinite series. In general this guy's computations will also end up with infinite series that need to be truncated (for example taking the square root at the end). It doesn't really matter, therefore, when it comes to numerical computation. A square root and a sine are very similar if you're a computer.
Furthermore, a lot of what this guy did is kind of a trick. Using 'spreads' may work when given an explicit triangle, but the part he's skimming over is that spreads are missing a REALLY nice property of angles. They don't add. Angles are a very nice parameter for rotation because a rotation of 10 followed by a rotation by 20 is the same as a rotation by 30. This property is implicitly used all over the place in graphics. So, in the end you probably have to use some angle-like measure when doing computer graphics (which is all about transformations, not measurements of unknowns). And in doing so, I'm sure you end up computing sines and cosines to do projections based on those rotations.
In the end, you just can't cheat your way out of the fact that a projection based on a rotation is a transcendental operation that numerically requires computing a truncated infinite series.
Objective C may be ancient and cryptic but it is and always has been fully OO. It's dynamic which means you can generate classes / subclasses / methods on the fly. Not something that C# or Java could do.
That's what I meant by faking it. The object orientation is handled by a runtime, not by the compiler. In fact, my understanding is that after preprocessing a standard C compiler does the work. In my mind, if the compiler doesn't know about objects, the language isn't really object oriented. Anyway, at the very least it's fair to consider the object orientation of Obj-C to be an interpreted aspect of the language, and that bothers me. One shouldn't have to worry about calling methods too often in an object oriented language.
Wow. Excellent reponse. Thank you. I stand corrected on quite a few counts. I guess C# beauty is in the eye of the beholder, but I really can't argue with your knowledge of Apple's crew.
About spotlight: my understanding was that it was very similar to Microsofts existing index-based disk search. The one difference is that Apple allows for "plug-ins" to index arbitrary data files. Other than that, though, it's just keyword search. Am I missing something?
About AppleScript: I was aware of the inheritence, but I thought that was off-limits to the actual user. That is, a programmer can write a dictionary with elements that inhererit, but the user cannot actually make up elements using AppleScript itself. And about the UNIX shell Apple ships with: that's another huge complaint I've got. Apple, uncharacteristically, hasn't integrated the UNIX and Mac parts of the OS. They've kept resource forks, and the UNIX commands don't grok them. So if you make a tar file, it may not contain everything you think it does. And the whole.app bundle is one hell of a kludge that becomes apparent when using a UNIX shell.
Anyway, I'm rooting for Apple, but I would be happy to see Microsoft pull things off, too. Thanks very much for your post.
I'm so goddam sick of smallminded mods with an agenda downing everything as a "Troll" that doesn't agree with their little world view. I gave reasons for everything I said, and said so in a respectful manner. So, on the off chance the fool who modded me down is actually reading these threads: remember the part of the modding FAQ about "concentrate more on promotion than demotion"? That means YOU. Just because you don't like an opinion isn't grounds for modding down, you twit. If you think my post was a troll, you must've just gotten your AOL disk in the mail last week...
Maybe I'm wrong. I've programmed in all three, but am much more familiar with C# and C++. Could you please give me an example of something that you can do with Obj-C that you can't do with C#? Anyway, I didn't say it didn't ACT like an object oriented language, I was just being critical of the fact that it's sort of grafted on top of C in what I think is an inelegant way. I think method dispatch is particularly slow in Objective C, for example.
c# > obj-c? have you actually programmed in obj-c? Stop spreading rumours please
I've programmed in both. Have you? Just one person's opinion, but I don't think I'm alone. I like type safety and garbage collection, for starters. Maybe I'm crazy. I also like any language that doesn't look like scheme.
Well, I said they TRY.:-) I think their efforts with WinFS show that. They could've just slapped a nice interface on their existing keyword search engine, like Apple did, but they are instead trying to do something a bit more sophisticated.
Second, just look at C#. It's a beautiful language that addresses many of the problems people have with Java and C++. It's not perfect, but it represents a lot of what people have learned from experience. Apple is using a version of C that fakes object orientation through a runtime and some clever preprocessing. Nobody expected objective C to be around in the year 2005.
Third, I believe most of the Windows kernel was written at Microsoft, and while I'm not a kernel hacker, from what I know the kernel is quite nicely done and well optimized. Windows XP certainly feels very responsive. Apple, on the other hand, didn't even write their kernel (they just used Mach) and from the number of times I sit and watch the spinning beach ball, I'm not sure they did much optimization for the PPC.
Fourth, as a good example of what I'm talking about look at AppleScript versus the shell being prepared for Vista server (can't remember its name). That's all you need to know about the difference between the two companies in term of their approach. One is a cute language for scriping iPhoto with no discernable syntax rules and the other is a powerful administrative tool that uses XML standards. I'm not sure which one is better for consumer devices, though.
Finally, Apple is a hardware company. They've always said so. The OS for them is not the primary concern. You want to work at Apple if you're a designer. If you're a computer scientist, you want to work at Google, or maybe Microsoft, but I don't know any CS people who consider Apple the place to go to do cutting edge CS. Maybe that's changing, I hope it is, but I'm not sure Apple even has the people in-house to rewrite an OS. Microsoft does, and is.
Steve Jobs may be confident that the next generation of OSX will beat Vista in comparison reviews - hell, the current version (Tiger) has a lot of the features Vista is supposed to have already.
That's misleading, because Vista doesn't really have any features, itself, except pretty pictures and a nice new browser. If you're talking about the forthcoming WinFS and Indigo, which ARE significant things, you're also wrong. Spotlight isn't really a database file system, but is more of a quick solution thrown together that looks pretty, but really isn't anything we haven't already seen on Windows in various forms for quite some time now. It's just got a nice menu icon. Apple tends to be good at getting things out quicker because they are content to wrap a pretty interface on a quick and dirty fix. (Which actually sums up Mac OS X itself, pretty much, which is a total kludge of OpenStep on BSD on HFS, and none of the three are completely integrated yet.) However, if it looks good enough to please the art school crowd, it works for Apple and everybody forgets that the "new" operating system is really a 15 year old class library.
So, the REAL question is: when Microsoft finally gets their shit together (they will eventually) will Apple have something to really compete against WinFS, Indigo and C# (which is lightyears ahead of OpenSTEP's frameworks and 'objective' C)? My hope is they do (I'm typing this on a PB G4) but my point is that it's not fair to say Tiger already has any of this. Microsoft, for all their faults, tends to try to do things the right way, at least in terms of computer science, if not morality, whereas Apple just tries to get it done. In the end Microsoft will have an almost new operating system with some very modern ideas on file systems and distributed computing. For all the delays, it could be quite good, actually. Apple, on the other hand, really hasn't added anything of huge significance to the old OpenSTEP except pretty pictures and a really nice graphics system. Yet...
epidemiologists find the Danish incredibly useful for large scale demographic data sweeps
Are you suggesting that scientific studies are justification for this kind of thing? That people's right to privacy is subject to suspension if a social scientist think some good data can come of it?
it's a great idea, so long as there's rigorous statutory limitations on how the data may be used and accessed
Maybe Europeans are far enough evolved for their governments to never abuse their power. But it's pretty clear we Americans don't have such considerate politicians, and even if we did I'm not willing to take the risk just so we can produce some nice statistics.
I'm not sure I'd call them communities, then. If you don't have a reasonable chance of knowing each person, it's not a community, almost by definition. The word gets thrown around a lot (schools with 20,000 students often conceit to considering themselves a community, for example) but it's sort of a meaningless concept unless you know the people. Imagine how much more useful actually knowing people is than keeping a database on them. Strong community will do far more for children than a database. When I was a kid, I had about six extra mothers around town who knew me since I was a baby. There's no way in hell I was going to get away with anything without my real parents finding out.
I completely respect some people's desire to live in anonymity and devote their lives to work while sending their kids to daycare, but I highly resent their desire to compensate for the resulting lack of community and parenting by forcing a nanny state on everybody.
Sometimes things are just wrongheaded, no matter what the intentions. There's no reason the government should be taking this much personal interest in citizens. It's not just that it's ripe for abuse so much as it's an indicator that the Dutch have completely given up on taking care of themselves and their communities on any level other than a centralized beaurocracy. What does it say about a society when they feel the best way to Do The Right Thing is to keep a central database on each other and pay the government to track their children for them?
I hope I speak for a majority of voters in America when I respectfully say please keep that shit on the other side of the pond.
(To head off the knee-jerk reactions: Yes, I know America is a place devoid of compassion for the poor, and that we might do better with an Orwellian scheme like this than what we're doing now. I'm not saying America does things right, I'm just saying I hope that in attempting to fix our problems we don't go anywhere near this kind of 'solution'.)
Another way to look at it is that Americans tend to realize that the same restriction won't work for everybody, and it's better to let people have responsibility for their own ears than trying to shoehorn everybody into a single mandated solution.
I'm sorry about your hearing loss, but holy shit does your hearing aid sound cool. You're like the bionic man when it comes to hearing. I'd love to be able to reprogram my ears, and even add a feedback loop for noise cancelling. How much did your gizmo cost?
I though that the reason the high-freq hearing went first was that the hairs in the coclea that are resonant to high frequencies are more easily damaged since they are thinner?
Amen, brother. It all started when they switched to rubber buttons...
Now their scopes run linux and have a goddam scroll ball on the front. And generally look like they were designed by the people who made the Lady Schick Razor.
HP died a long time ago; the products they put on the market place these days are just nerves twitching.
Ok, I guess I misunderstood what the grandparent meant by open. If you mean "publish APIs and specs" then, for all intents and purposes, AIM is open. By closed, I was talking (perhaps erroneously) about access. The POTS system is (ok, maybe was) closed in the sense that if you want to ring your friend, you eventually need to go over the copper last mile which is owned by his Telco and they are under no obligation to let you or charge a fair price. For example, except for legal problems, telcos could just disallow IM->POTS. In other words, ALL physical networks with single line bottlenecks are closed in a sense. And that's how I thought the grandparent was talking about Skype being closed; they're certainly not going just give away their SkypeOut service (the only interesting part of Skype).
Anyone still use it? Its been replaced by jabber and Windows Messenger, and I think messenger will die from neglect eventually.
I use it. Most people that IM I know use it. I think you're jumping the gun. AIM will live because there's critical mass there. In the end, I use IM software to talk to people, not to massage my beliefs. I care about who uses it, not the openness of their service, whatever that means.
If you want to see what openness gets you, look at TCP/IP, SMTP, HTTP and HTML.
Is that really a fair comparison? HTTP is a protocol existing only in logic; Skype is a service requiring physical assets. Exactly how do you propose somebody do IM->POTS with a system that is open in a meaningful way? Sure, Skype could give you all the APIs you want, but if they don't let you use their network of POTS interfaces, who gives a damn that they inform you of their protocols? I guess I'm just confused at this point over what you all mean when you say 'open'.
Anyway, I'm not being beligerent; I'd honestly appreciate being corrected where I'm missing your point. I do seem to be in the minority here so I assume I'm just misunderstanding what people mean by open as it pertains to a message service.
Skype is not built on open standards like SIP and remains isolated to its own so-called "Peer to Peer" network. It is to the Gizmo Project as AIM is to Jabber.
I don't understand your logic. Why is it the automatic death of something to not be based on an open standard? An open standard is nice in some obvious ways, but I'd take something closed that works well over anything else, open or no.
Isn't the POTS system closed? Pretty successful, though, no? And AIM, for that matter, is still doing pretty well. Windows? Closed, sucks, AND very successful.
I think your reports of Skype's imminent demise are premature, given that it's already well recieved and actually quite well designed. I think you may be getting a bit religious-minded about open standards. I probably agree with you that it would be The Right Thing if everything were, but I think you've let the dogma get to you if you think something's going to die just because it isn't based on an open standard. Would that it were the case, though...
Slashdot Mirror
I'm not sure I understand what you mean about angles not being directly useful. Rotation a given angle seems fundamental. If you are doing a POV game, and the person starts turning left, you're rotating the frame or camera by a definite angle each time step. If you want to grab a molecule in a protein visualization and spin it around, you'd be working with angles. Are you saying that angles are used only at a minimum, for rotating a few vectors on to which everything else is projected?
I agree with you. I should've been more explicit and stated that I was talking about general purpose digital computing with optics. Having said that, even the Fourier transform stuff that once looked so promising really doesn't look so good. How do you get the data in and out of the system? In the end, you need a spatial light modulator, and those are so damn slow that you can always compute a fourier transform faster on a DSP chip, despite the parallelism of light.
It sounds like you just made a typo or something in talking about a single photon, when you meant photons. If so, my apologies.
A photon has a wavefunction, that's it. Assuming the photon has a relatively sharp energy, you can then say it has a frequency (energy) and a phase. Amplitude is meaningless when speaking about a single photon.
I can only wait for the development of photonic processors.
You'll be waiting a long time, I think. People have been trying this foolhardy idea of optical computing for a long time, and the field is pretty much dead. The problem is that to get light to modulate other light, you still have to move electrons around. Photons don't interact directly with each other. So if you're moving electrons around, for one, you're back to the old problem of finding materials with fast state transitions.
But most importantly, nonlinear optical effects with degerate wavelengths (where inputs and output are the same color) are by definition, at least third order effects (in terms of perturbation theory where light is the perturbation on a quantum system). In other words, they are weak effects, and take a rather large distance (relative to a wavelength of light) to switch. Thus, while people will brag about this modulation speed and that, what they don't tell you is that (a) the device is big (on the order of many microns) and therefore (b) the latency sucks. This means you can't compute with it since your answer takes longer to come out the other end then a transistor would've taken. Of course, nobody tells this to us poor people who venture unwittingly into the field of optics as freshly graduated students who don't know any better. Because of all of this, optics is great for communications (where you modulate a ton in series, but don't need to wait for the output to feed into a next computation step) but really pretty bad for computation. So, the originaly article is right about clocking being a good match for optics, but all the user comments about optical computing should be taken with a grain of salt, I'm afraid.
I'm not dismissing this guy's work at all, I'm just saying that you can only avoid classical trig in specific problem domains.
Furthermore, a lot of what this guy did is kind of a trick. Using 'spreads' may work when given an explicit triangle, but the part he's skimming over is that spreads are missing a REALLY nice property of angles. They don't add. Angles are a very nice parameter for rotation because a rotation of 10 followed by a rotation by 20 is the same as a rotation by 30. This property is implicitly used all over the place in graphics. So, in the end you probably have to use some angle-like measure when doing computer graphics (which is all about transformations, not measurements of unknowns). And in doing so, I'm sure you end up computing sines and cosines to do projections based on those rotations.
In the end, you just can't cheat your way out of the fact that a projection based on a rotation is a transcendental operation that numerically requires computing a truncated infinite series.
That's what I meant by faking it. The object orientation is handled by a runtime, not by the compiler. In fact, my understanding is that after preprocessing a standard C compiler does the work. In my mind, if the compiler doesn't know about objects, the language isn't really object oriented. Anyway, at the very least it's fair to consider the object orientation of Obj-C to be an interpreted aspect of the language, and that bothers me. One shouldn't have to worry about calling methods too often in an object oriented language.
About spotlight: my understanding was that it was very similar to Microsofts existing index-based disk search. The one difference is that Apple allows for "plug-ins" to index arbitrary data files. Other than that, though, it's just keyword search. Am I missing something?
About AppleScript: I was aware of the inheritence, but I thought that was off-limits to the actual user. That is, a programmer can write a dictionary with elements that inhererit, but the user cannot actually make up elements using AppleScript itself. And about the UNIX shell Apple ships with: that's another huge complaint I've got. Apple, uncharacteristically, hasn't integrated the UNIX and Mac parts of the OS. They've kept resource forks, and the UNIX commands don't grok them. So if you make a tar file, it may not contain everything you think it does. And the whole .app bundle is one hell of a kludge that becomes apparent when using a UNIX shell.
Anyway, I'm rooting for Apple, but I would be happy to see Microsoft pull things off, too. Thanks very much for your post.
What are things that will give you a virus?
I'm so goddam sick of smallminded mods with an agenda downing everything as a "Troll" that doesn't agree with their little world view. I gave reasons for everything I said, and said so in a respectful manner. So, on the off chance the fool who modded me down is actually reading these threads: remember the part of the modding FAQ about "concentrate more on promotion than demotion"? That means YOU. Just because you don't like an opinion isn't grounds for modding down, you twit. If you think my post was a troll, you must've just gotten your AOL disk in the mail last week...
Maybe I'm wrong. I've programmed in all three, but am much more familiar with C# and C++. Could you please give me an example of something that you can do with Obj-C that you can't do with C#? Anyway, I didn't say it didn't ACT like an object oriented language, I was just being critical of the fact that it's sort of grafted on top of C in what I think is an inelegant way. I think method dispatch is particularly slow in Objective C, for example.
I've programmed in both. Have you? Just one person's opinion, but I don't think I'm alone. I like type safety and garbage collection, for starters. Maybe I'm crazy. I also like any language that doesn't look like scheme.
Heh heh... Touche.
Second, just look at C#. It's a beautiful language that addresses many of the problems people have with Java and C++. It's not perfect, but it represents a lot of what people have learned from experience. Apple is using a version of C that fakes object orientation through a runtime and some clever preprocessing. Nobody expected objective C to be around in the year 2005.
Third, I believe most of the Windows kernel was written at Microsoft, and while I'm not a kernel hacker, from what I know the kernel is quite nicely done and well optimized. Windows XP certainly feels very responsive. Apple, on the other hand, didn't even write their kernel (they just used Mach) and from the number of times I sit and watch the spinning beach ball, I'm not sure they did much optimization for the PPC.
Fourth, as a good example of what I'm talking about look at AppleScript versus the shell being prepared for Vista server (can't remember its name). That's all you need to know about the difference between the two companies in term of their approach. One is a cute language for scriping iPhoto with no discernable syntax rules and the other is a powerful administrative tool that uses XML standards. I'm not sure which one is better for consumer devices, though.
Finally, Apple is a hardware company. They've always said so. The OS for them is not the primary concern. You want to work at Apple if you're a designer. If you're a computer scientist, you want to work at Google, or maybe Microsoft, but I don't know any CS people who consider Apple the place to go to do cutting edge CS. Maybe that's changing, I hope it is, but I'm not sure Apple even has the people in-house to rewrite an OS. Microsoft does, and is.
That's misleading, because Vista doesn't really have any features, itself, except pretty pictures and a nice new browser. If you're talking about the forthcoming WinFS and Indigo, which ARE significant things, you're also wrong. Spotlight isn't really a database file system, but is more of a quick solution thrown together that looks pretty, but really isn't anything we haven't already seen on Windows in various forms for quite some time now. It's just got a nice menu icon. Apple tends to be good at getting things out quicker because they are content to wrap a pretty interface on a quick and dirty fix. (Which actually sums up Mac OS X itself, pretty much, which is a total kludge of OpenStep on BSD on HFS, and none of the three are completely integrated yet.) However, if it looks good enough to please the art school crowd, it works for Apple and everybody forgets that the "new" operating system is really a 15 year old class library.
So, the REAL question is: when Microsoft finally gets their shit together (they will eventually) will Apple have something to really compete against WinFS, Indigo and C# (which is lightyears ahead of OpenSTEP's frameworks and 'objective' C)? My hope is they do (I'm typing this on a PB G4) but my point is that it's not fair to say Tiger already has any of this. Microsoft, for all their faults, tends to try to do things the right way, at least in terms of computer science, if not morality, whereas Apple just tries to get it done. In the end Microsoft will have an almost new operating system with some very modern ideas on file systems and distributed computing. For all the delays, it could be quite good, actually. Apple, on the other hand, really hasn't added anything of huge significance to the old OpenSTEP except pretty pictures and a really nice graphics system. Yet...
Are you suggesting that scientific studies are justification for this kind of thing? That people's right to privacy is subject to suspension if a social scientist think some good data can come of it?
it's a great idea, so long as there's rigorous statutory limitations on how the data may be used and accessed
Maybe Europeans are far enough evolved for their governments to never abuse their power. But it's pretty clear we Americans don't have such considerate politicians, and even if we did I'm not willing to take the risk just so we can produce some nice statistics.
I completely respect some people's desire to live in anonymity and devote their lives to work while sending their kids to daycare, but I highly resent their desire to compensate for the resulting lack of community and parenting by forcing a nanny state on everybody.
Sometimes things are just wrongheaded, no matter what the intentions. There's no reason the government should be taking this much personal interest in citizens. It's not just that it's ripe for abuse so much as it's an indicator that the Dutch have completely given up on taking care of themselves and their communities on any level other than a centralized beaurocracy. What does it say about a society when they feel the best way to Do The Right Thing is to keep a central database on each other and pay the government to track their children for them? I hope I speak for a majority of voters in America when I respectfully say please keep that shit on the other side of the pond. (To head off the knee-jerk reactions: Yes, I know America is a place devoid of compassion for the poor, and that we might do better with an Orwellian scheme like this than what we're doing now. I'm not saying America does things right, I'm just saying I hope that in attempting to fix our problems we don't go anywhere near this kind of 'solution'.)
Another way to look at it is that Americans tend to realize that the same restriction won't work for everybody, and it's better to let people have responsibility for their own ears than trying to shoehorn everybody into a single mandated solution.
I'm sorry about your hearing loss, but holy shit does your hearing aid sound cool. You're like the bionic man when it comes to hearing. I'd love to be able to reprogram my ears, and even add a feedback loop for noise cancelling. How much did your gizmo cost?
I though that the reason the high-freq hearing went first was that the hairs in the coclea that are resonant to high frequencies are more easily damaged since they are thinner?
Amen, brother. It all started when they switched to rubber buttons...
Now their scopes run linux and have a goddam scroll ball on the front. And generally look like they were designed by the people who made the Lady Schick Razor.
HP died a long time ago; the products they put on the market place these days are just nerves twitching.
Ok, I guess I misunderstood what the grandparent meant by open. If you mean "publish APIs and specs" then, for all intents and purposes, AIM is open. By closed, I was talking (perhaps erroneously) about access. The POTS system is (ok, maybe was) closed in the sense that if you want to ring your friend, you eventually need to go over the copper last mile which is owned by his Telco and they are under no obligation to let you or charge a fair price. For example, except for legal problems, telcos could just disallow IM->POTS. In other words, ALL physical networks with single line bottlenecks are closed in a sense. And that's how I thought the grandparent was talking about Skype being closed; they're certainly not going just give away their SkypeOut service (the only interesting part of Skype).
Anyone still use it? Its been replaced by jabber and Windows Messenger, and I think messenger will die from neglect eventually.
I use it. Most people that IM I know use it. I think you're jumping the gun. AIM will live because there's critical mass there. In the end, I use IM software to talk to people, not to massage my beliefs. I care about who uses it, not the openness of their service, whatever that means.
If you want to see what openness gets you, look at TCP/IP, SMTP, HTTP and HTML.
Is that really a fair comparison? HTTP is a protocol existing only in logic; Skype is a service requiring physical assets. Exactly how do you propose somebody do IM->POTS with a system that is open in a meaningful way? Sure, Skype could give you all the APIs you want, but if they don't let you use their network of POTS interfaces, who gives a damn that they inform you of their protocols? I guess I'm just confused at this point over what you all mean when you say 'open'.
Anyway, I'm not being beligerent; I'd honestly appreciate being corrected where I'm missing your point. I do seem to be in the minority here so I assume I'm just misunderstanding what people mean by open as it pertains to a message service.
I don't understand your logic. Why is it the automatic death of something to not be based on an open standard? An open standard is nice in some obvious ways, but I'd take something closed that works well over anything else, open or no.
Isn't the POTS system closed? Pretty successful, though, no? And AIM, for that matter, is still doing pretty well. Windows? Closed, sucks, AND very successful.
I think your reports of Skype's imminent demise are premature, given that it's already well recieved and actually quite well designed. I think you may be getting a bit religious-minded about open standards. I probably agree with you that it would be The Right Thing if everything were, but I think you've let the dogma get to you if you think something's going to die just because it isn't based on an open standard. Would that it were the case, though...
Heh heh. I didn't think of that. :-)