"I remember being similarly annoyed in the 95/98/early-NT timeframe when Windows apps stopped displaying ".." in the directory pane (easy, just PgUp to get to the top of the file listing and hit Enter to go up one) and forced me to switch from keyboard to mouse and click on the up-pointing green arrow."
As far as I know, you've always been able to type ".." into the file name bar and go up a directory. It seems to be that would be a more likely place for the focus if you're using a keyboard anyway.
Vista actually makes a huge improvement for this sort of navigation too in the save dialog. Under XP and before, if you entered a folder name and hit enter it would navigate to that folder, but the folder name would remain in the text box. This meant that you had to retype the file name you wanted, copy and paste the file name you wanted, or (I don't know why I didn't think of this until now) hit ctrl-Z to go back to what it was before. (It doesn't revert your folder location, just the text. Downside: only works if you go only one folder.) In Vista, if you navigate by typing a folder name, it will navigate to that folder then restore the original file name.
Anyway, I'm a big fan of letting people pick the way that's right for them. I missed the up button too, and sometimes still do. So part of me sympathizes with you. At the same time, from my perspective I am glad that ".." isn't shown in directory listings.
Actually, the complaint is about how IE renders HTML that does conform to standards.
Actually, while that's the more common complaint, if you were to look through the archives you will find plenty of complaints about how the fact that IE allows poorly coded HTML to render discouraged developers from coding reasonably, and creating sites that don't work in other browsers. This is probably less of a problem now, since Firefox has taken off pretty well, but 3 or 4 years ago this was a big problem, and there were complaints.
Nature, you have once again awed me with your incredible weirdness.
Every time I read a story like this, I can't help but think of the following quote from Hitchhiker's:
There is a theory which states that if ever anyone discovers exactly what the Universe is for and why it is here, it will instantly disappear and be replaced by something even more bizarrely inexplicable.
There is another theory which states that this has already happened.
I sometimes wonder if perhaps there is a God, and he is changing the rules on us. Like maybe until the time of Newton, it was the case that mass didn't vary with velocity, and that you could continually apply a force and accelerate to infinite speeds. And then Newton came around, discovered those rules, and God supplanted them with relativity. And then Einstein came along, and God said "hmm, let's add dark matter and make things more interesting."
Sort of like when I was a kid playing battleship, I would track my opponent's guesses on my lower board. And if they guessed where one of my ships was, I would discretely move the ship to somewhere they didn't guess yet and announce a miss. Except with physical laws.
I don't think this is a reasonable thing to actually believe, but it would be amusing. If I were God I would totally do that.
It's not exactly like the Wii... the Wii tracks IR emitters, whereas motion capture more commonly uses reflective spheres and a separate light source. Also the camera is stationary, and the light sources are moving in motion capture, whereas in the Wii it's the other way around.
But I would say the two techniques are a lot closer than you seem to think.
I wouldn't want to do this. There are plenty of reasons why a company would, very legitimately IMHO, want to register multiple domains. First there are typos; I don't think it's reasonable to expect Google to pay through the nose to get "typoed" domains like googel.com or gogole.com (both of which they have). Maybe the need for these domains would go down a lot if there weren't squatters who would snap them up if Google didn't though, so maybe this isn't terribly convincing. (That said, Google apparently currently has 520 domains, though a ton of those are through acquisitions of other companies.)
The second big reason is that I don't think it's reasonable to prevent registering different domains for either different services or different products. For instance, msoffice.com redirects to the Office page at microsoft.com.
Now you could double to a limit, but I wouldn't make it much above $1000.
It would also cut me out. You don't need to make them terribly expensive... just $25 or so would probably be sufficient. It would at least cut that percentage down a ton.
The gifting system for the Orange Box was only set up to work for HL2 and Episode 1. I don't know if this was a deliberate decision on their part, or if they didn't bother to make the mechanism work for the other titles because they figured no one would be daft enough to buy Portal/Ep2/TF2 alone and then buy the Orange Box.;-)
(For my case, I bought Portal separately. At that point I hadn't done Ep1 yet. I then went through Ep1, said "wow, I really want Ep2 now", and about the same time saw that Best Buy was selling the Orange Box for $25 on Black Friday. So I bought it.)
I just had a problem last night starting EP2. I think it wanted to update itself, and the servers were busy. I turned off automatic updating (didn't try offline mode though), but that didn't work.
A minute later it did successfully start though, so this wasn't too bad.
The HUGE annoyance I had with Steam was when I bought Portal not long after its release. For several straight hours, their servers were too busy to serve it to me.
The new annoyance I have is that you can't give a game to someone else. I bought Portal standalone, then later got the Orange Box. So I should have an extra Portal license I can give away, right? No.
Only kernel mode drivers can cause a blue screen. Other drivers can take out access to a component (e.g. the sound), but they can't bring down the system.
I don't know if it would change your conclusion or not, but remember that if you are comparing to 2001 you need to be comparing to XP before SP2. There's a ton of improvement just to SP2, probably the equivalent of one OS X version change.
That said, I do agree that the time elapsed since last version makes the improvements in Vista disappointing.
(The 32 bit version should never have existed, IMO. Microsoft should have used that as the dividing line. "Want a computer with less resources? Use XP. Want to use 4GB of RAM? Use Vista." Its not like Vista on a 1GB Laptop is going to work worth a damn anyway.)
This isn't a bad idea. The other thing it touches on is the fact that the 32-bit editions of XP and Vista don't support PAE (at least officially), so are limited to 4 GB of RAM. (And really, 4 GB minus your PCI address space, which includes all of your video card RAM. So for someone who would have that much system RAM, we're talking under 3.5 GB.)
Well, guess what: we're rapidly approaching the point where that limit is getting reached. And a lot of people aren't going to be happy to find that if they have a computer with 2 GB and decide to do a 2 GB upgrade, suddenly they have to upgrade to the 64 bit version to access all of it.
I see the presence of a 32-bit Vista without PAE support to be a pretty big error on the part of MS, especially if most of the retail computers are shipping with the 32-bit version. (I don't actually know if that's typical or not.)
"You open the IIS manager Cancel or Allow Well duh! I Open the IIS Manager by clicking the Icon I would Think I want access to the App."
What if that wasn't you clicking on the icon, but rather was a piece of malware sending a "double click" message to the icon, and wanted to then send a series of other messages to change the settings to something insecure that it wants?
(That this may be possible to do is another issue entirely. Vista makes a small amount of progress towards doing something about it, but fixing the issue completely would require breaking far more programs than MS is willing to.)
"if a shoddy driver is affecting stability beyond the device in question it's not the driver that's to blame. You may as well also subscribe to the BS microsoft spewed about old file formats being insecure rather than the program that reads them."
I mostly disagree with this on a number of points.
(1) Every other remotely common OS -- the various Unixes, Linux, OS X -- is just as susceptible as Windows is. They all use the same architecture: the driver runs in the kernel. Once you have that, an unstable driver can easily crash the system. Guess what: there are rootkits for Linux too, and they use the exact same principles as Windows ones: once you are installed as a driver, you are God.
(2) The main reason that this has been done is that it's hard to do well another way. Until relatively recently, the only mechanism that provided protection against a badly behaved driver was to run it in its own protection domain. This means a context switch whenever the kernel wants to call the driver, and a system call when the driver wants to call the kernel or return. For many drivers, the overhead here has been unacceptable. In the last several years there have been a couple new ideas for how to provide protection with lower overhead, but (1) they remain in the research state and haven't made it to real-world products, and (2) they too have overheads that are not trivial.
(3) MS is actually doing MORE to move drivers out of the kernel than the other mainstream OSes. Linux has some examples, for instance FUSE, but Vista introduces a new driver model that strongly encourages user mode drivers. (For instance, sound drivers are often written with the UMDF.) Performance critical drivers, such as parts of video card drivers, still run in the kernel.
For instance, MS watered down greatly the file associations tab. Before you could, open up the dialog, add a new file type, say.py for python, and associate it with "python.exe -i %1" or something like that in order to run it. You could also add a number of other actions that would show up in the right click menu. AFAICT, there is no longer any ability to add a file type from that dialog. (You have to make a file, double click it, choose the program, and tell it to always use that.) There is also no ability to make multiple actions. There is no ability to specify the command line; you can just specify the program, which means you can't, for instance, pass -i to python. If you want to do the stuff XP let you do, you have to manually edit the registry.
That said, I do think it's an improvement. There are some things that I've noted that make a big difference. For instance, in XP and previous versions, when Explorer tried to display a preview of some video files, it would crash. In Vista, the preview is done in a separate process, so you just get a dialog saying that the "COM Surrogate" has crashed. This is a huge improvement in this respect. (Though they could make it better by eliding the crashing dialogs for the COM surrogate, as it will crash once for every crashy video you have.)
There are some new features that I like. For instance, I actually *like* UAC. I think the complaints are vastly overstated, especially as they often come from people who run Linux/Unix and have to run "sudo" everywhere. Over about a month long period (maybe 3/4 of which I was in Windows) I kept a log of every UAC prompt I got. Almost all of them were for things that would have required root privileges on Linux. The only thing that Linux would have going for it at that point is that you can open a root prompt and do a number of things without giving permission. This is balanced somewhat by the fact that, under the default Vista setup, the user is running as a close-to-admin user, so the UAC prompt doesn't prompt for a password. Again, the mechanism isn't perfect; for instance, Vista will elevate installers whether they actually need it or not. But I think the complaints are overblown.
I haven't had an issue with the driver signing thing because I don't run x64. I'm not totally convinced it's because of DRM though. I suspect it's more a combination of (1) DRM, (2) generally keeping more control away from the user, (3) malware concerns, and (4) providing accountability in the face of poorly written drivers.
Agreed. Also Portal is a puzzle game. HL2 is a FPS with occasional, pretty simple puzzles. Forcing puzzles into HL2 or a significant FPS part into Portal I don't think would work. Having the portal gun in HL2 would be neat, but it wouldn't be Portal, it would be HL2 with the portal gun.
Nonsense. If IBM didn't own OS/2 they couldn't have sold it to Serenity Systems. Open sourcing is the equivalent of selling the source for zero dollars. IBM can set whatever price they like.
What if IBM's agreement with MS included additional restrictions? For instance, that IBM could sublicense OS/2, but only under an agreement that required royalties to be paid to MS for each copy distributed. Is Serenity Systems paying royalties to MS? Maybe the license only allows certain other kinds of redistribution. You certainly couldn't add a "if you copy this software you need to send $1 to Microsoft" clause to any license and have it accepted by the OSI, accepted by Debian, or combined with any GPL'd software.
I have no clue what IBM's agreement with Serenity Systems said, but somehow I doubt you do either, and your post leaves an awful lot implicit.
I guess that's why IBM did not develop the cell processor which is therefor not used in PS3s or why no supercomputer is built using it.
While true, this is a little bit of a red herring for this discussion. Sony gets away with using the cell because it's a console system. That games from one generation of consoles are not compatible with other generations is largely expected. Thus they can go and change the architecture willy-nilly. The same is true for Microsoft and Nintendo. (I'm not sure whether they are backwards compatible or not. But I assert that if they weren't, their sales wouldn't be substantially hit.)
OTOH, backwards compatibility with PCs are expected. The work MS has put into backwards compatibility, maintaining bug-for-bug behavior to keep even badly behaved programs working between versions is one of the major reasons for Windows's success. (It's also one of the major reasons they are hurting in security and complexity, but that's another issue.) If Dell stopped selling x86 computers and only, I dunno, Itanium, people would stop buying it. (Or, more likely, they would buy, then return it when they found out that none of their old software works.)
This isn't to say architecture changes are impossible; look at Apple's switch. There were a couple things that made this happen. First is that Apple did maintain backwards compatibility through Rosetta, and second is that Apple controls all of Apple's distribution and could therefore unilaterally make the switch. (The gaming companies also have this second reason. You don't have Microsoft making a version of the XBox 360 that isn't backwards compatible because they changed architectures and some other company staying with the x86 that was in the first version (I think that's right) so that you can continue to run old games.)
I think one of the two would be needed before you saw another architecture supplant x86 on people's desks. If AMD decided to scrap x86 and release a new chipset, say Uberisa, I don't think it would see much success. They would need to either convince MS to implement a VM (or do it themselves) that would let people run x86 code on it at a good speed, or convince all of the big manufacturers to drop x86.
Without one of those two, AMD could introduce the Uberisa, but uptake would be very slow. It might pick up steam eventually, but I predict that it would remain specialized. I mean, look at Transmeta. They've been around for a while, and provide a way to run x86. But who actually uses a Transmeta chip?
Of course, all this assumes you're going after the PC market. If you're not, then there is a lot more room for you to do what you want. But at the same time, if you're not going for the PC market, you're probably not too worried about supplanting x86.
And I would just like to point out that those are just still images, and none were of people (until you got to the bonus round). Start animating people, and you'll be able to tell the difference real quick.
Slashdot Mirror
"I remember being similarly annoyed in the 95/98/early-NT timeframe when Windows apps stopped displaying ".." in the directory pane (easy, just PgUp to get to the top of the file listing and hit Enter to go up one) and forced me to switch from keyboard to mouse and click on the up-pointing green arrow."
As far as I know, you've always been able to type ".." into the file name bar and go up a directory. It seems to be that would be a more likely place for the focus if you're using a keyboard anyway.
Vista actually makes a huge improvement for this sort of navigation too in the save dialog. Under XP and before, if you entered a folder name and hit enter it would navigate to that folder, but the folder name would remain in the text box. This meant that you had to retype the file name you wanted, copy and paste the file name you wanted, or (I don't know why I didn't think of this until now) hit ctrl-Z to go back to what it was before. (It doesn't revert your folder location, just the text. Downside: only works if you go only one folder.) In Vista, if you navigate by typing a folder name, it will navigate to that folder then restore the original file name.
Anyway, I'm a big fan of letting people pick the way that's right for them. I missed the up button too, and sometimes still do. So part of me sympathizes with you. At the same time, from my perspective I am glad that ".." isn't shown in directory listings.
Actually, the complaint is about how IE renders HTML that does conform to standards.
Actually, while that's the more common complaint, if you were to look through the archives you will find plenty of complaints about how the fact that IE allows poorly coded HTML to render discouraged developers from coding reasonably, and creating sites that don't work in other browsers. This is probably less of a problem now, since Firefox has taken off pretty well, but 3 or 4 years ago this was a big problem, and there were complaints.
Every time I read a story like this, I can't help but think of the following quote from Hitchhiker's:
I sometimes wonder if perhaps there is a God, and he is changing the rules on us. Like maybe until the time of Newton, it was the case that mass didn't vary with velocity, and that you could continually apply a force and accelerate to infinite speeds. And then Newton came around, discovered those rules, and God supplanted them with relativity. And then Einstein came along, and God said "hmm, let's add dark matter and make things more interesting."
Sort of like when I was a kid playing battleship, I would track my opponent's guesses on my lower board. And if they guessed where one of my ships was, I would discretely move the ship to somewhere they didn't guess yet and announce a miss. Except with physical laws.
I don't think this is a reasonable thing to actually believe, but it would be amusing. If I were God I would totally do that.
I'm not sure what your point is... A camera recording positions of bright dots is one of the most common motion capture technologies.
It's not exactly like the Wii... the Wii tracks IR emitters, whereas motion capture more commonly uses reflective spheres and a separate light source. Also the camera is stationary, and the light sources are moving in motion capture, whereas in the Wii it's the other way around.
But I would say the two techniques are a lot closer than you seem to think.
I wouldn't want to do this. There are plenty of reasons why a company would, very legitimately IMHO, want to register multiple domains. First there are typos; I don't think it's reasonable to expect Google to pay through the nose to get "typoed" domains like googel.com or gogole.com (both of which they have). Maybe the need for these domains would go down a lot if there weren't squatters who would snap them up if Google didn't though, so maybe this isn't terribly convincing. (That said, Google apparently currently has 520 domains, though a ton of those are through acquisitions of other companies.)
The second big reason is that I don't think it's reasonable to prevent registering different domains for either different services or different products. For instance, msoffice.com redirects to the Office page at microsoft.com.
Now you could double to a limit, but I wouldn't make it much above $1000.
It would also cut me out. You don't need to make them terribly expensive... just $25 or so would probably be sufficient. It would at least cut that percentage down a ton.
With Vista, Windows Update is a separate program, no longer done through IE.
The gifting system for the Orange Box was only set up to work for HL2 and Episode 1. I don't know if this was a deliberate decision on their part, or if they didn't bother to make the mechanism work for the other titles because they figured no one would be daft enough to buy Portal/Ep2/TF2 alone and then buy the Orange Box. ;-)
(For my case, I bought Portal separately. At that point I hadn't done Ep1 yet. I then went through Ep1, said "wow, I really want Ep2 now", and about the same time saw that Best Buy was selling the Orange Box for $25 on Black Friday. So I bought it.)
That's true. You're thinking playing time, I'm thinking sales. I think both make sense.
I just had a problem last night starting EP2. I think it wanted to update itself, and the servers were busy. I turned off automatic updating (didn't try offline mode though), but that didn't work.
A minute later it did successfully start though, so this wasn't too bad.
The HUGE annoyance I had with Steam was when I bought Portal not long after its release. For several straight hours, their servers were too busy to serve it to me.
The new annoyance I have is that you can't give a game to someone else. I bought Portal standalone, then later got the Orange Box. So I should have an extra Portal license I can give away, right? No.
They don't really have anything to worry about- their madly popular titles are all multiplayer...
You mean "except for Half Life, Half Life 2, Episode 1 and 2, and Portal", don't you?
Only kernel mode drivers can cause a blue screen. Other drivers can take out access to a component (e.g. the sound), but they can't bring down the system.
I don't know if it would change your conclusion or not, but remember that if you are comparing to 2001 you need to be comparing to XP before SP2. There's a ton of improvement just to SP2, probably the equivalent of one OS X version change.
That said, I do agree that the time elapsed since last version makes the improvements in Vista disappointing.
(The 32 bit version should never have existed, IMO. Microsoft should have used that as the dividing line. "Want a computer with less resources? Use XP. Want to use 4GB of RAM? Use Vista." Its not like Vista on a 1GB Laptop is going to work worth a damn anyway.)
This isn't a bad idea. The other thing it touches on is the fact that the 32-bit editions of XP and Vista don't support PAE (at least officially), so are limited to 4 GB of RAM. (And really, 4 GB minus your PCI address space, which includes all of your video card RAM. So for someone who would have that much system RAM, we're talking under 3.5 GB.)
Well, guess what: we're rapidly approaching the point where that limit is getting reached. And a lot of people aren't going to be happy to find that if they have a computer with 2 GB and decide to do a 2 GB upgrade, suddenly they have to upgrade to the 64 bit version to access all of it.
I see the presence of a 32-bit Vista without PAE support to be a pretty big error on the part of MS, especially if most of the retail computers are shipping with the 32-bit version. (I don't actually know if that's typical or not.)
Because anyone who (disagrees with you/likes Vista) must be a shill?
Seriously, a "resource monitor that has been added in the NT task manager" as a marketing point that MS shills would hit?
But it does have cool features such as ... the ability to load Windows updates withotu installing them, ...
XP (at least SP2 forward) had this feature. I know because I used it constantly.
"You open the IIS manager Cancel or Allow Well duh! I Open the IIS Manager by clicking the Icon I would Think I want access to the App."
What if that wasn't you clicking on the icon, but rather was a piece of malware sending a "double click" message to the icon, and wanted to then send a series of other messages to change the settings to something insecure that it wants?
(That this may be possible to do is another issue entirely. Vista makes a small amount of progress towards doing something about it, but fixing the issue completely would require breaking far more programs than MS is willing to.)
"if a shoddy driver is affecting stability beyond the device in question it's not the driver that's to blame. You may as well also subscribe to the BS microsoft spewed about old file formats being insecure rather than the program that reads them."
I mostly disagree with this on a number of points.
(1) Every other remotely common OS -- the various Unixes, Linux, OS X -- is just as susceptible as Windows is. They all use the same architecture: the driver runs in the kernel. Once you have that, an unstable driver can easily crash the system. Guess what: there are rootkits for Linux too, and they use the exact same principles as Windows ones: once you are installed as a driver, you are God.
(2) The main reason that this has been done is that it's hard to do well another way. Until relatively recently, the only mechanism that provided protection against a badly behaved driver was to run it in its own protection domain. This means a context switch whenever the kernel wants to call the driver, and a system call when the driver wants to call the kernel or return. For many drivers, the overhead here has been unacceptable. In the last several years there have been a couple new ideas for how to provide protection with lower overhead, but (1) they remain in the research state and haven't made it to real-world products, and (2) they too have overheads that are not trivial.
(3) MS is actually doing MORE to move drivers out of the kernel than the other mainstream OSes. Linux has some examples, for instance FUSE, but Vista introduces a new driver model that strongly encourages user mode drivers. (For instance, sound drivers are often written with the UMDF.) Performance critical drivers, such as parts of video card drivers, still run in the kernel.
This is somewhat my reaction too.
.py for python, and associate it with "python.exe -i %1" or something like that in order to run it. You could also add a number of other actions that would show up in the right click menu. AFAICT, there is no longer any ability to add a file type from that dialog. (You have to make a file, double click it, choose the program, and tell it to always use that.) There is also no ability to make multiple actions. There is no ability to specify the command line; you can just specify the program, which means you can't, for instance, pass -i to python. If you want to do the stuff XP let you do, you have to manually edit the registry.
Not everything is good about it.
For instance, MS watered down greatly the file associations tab. Before you could, open up the dialog, add a new file type, say
That said, I do think it's an improvement. There are some things that I've noted that make a big difference. For instance, in XP and previous versions, when Explorer tried to display a preview of some video files, it would crash. In Vista, the preview is done in a separate process, so you just get a dialog saying that the "COM Surrogate" has crashed. This is a huge improvement in this respect. (Though they could make it better by eliding the crashing dialogs for the COM surrogate, as it will crash once for every crashy video you have.)
There are some new features that I like. For instance, I actually *like* UAC. I think the complaints are vastly overstated, especially as they often come from people who run Linux/Unix and have to run "sudo" everywhere. Over about a month long period (maybe 3/4 of which I was in Windows) I kept a log of every UAC prompt I got. Almost all of them were for things that would have required root privileges on Linux. The only thing that Linux would have going for it at that point is that you can open a root prompt and do a number of things without giving permission. This is balanced somewhat by the fact that, under the default Vista setup, the user is running as a close-to-admin user, so the UAC prompt doesn't prompt for a password. Again, the mechanism isn't perfect; for instance, Vista will elevate installers whether they actually need it or not. But I think the complaints are overblown.
I haven't had an issue with the driver signing thing because I don't run x64. I'm not totally convinced it's because of DRM though. I suspect it's more a combination of (1) DRM, (2) generally keeping more control away from the user, (3) malware concerns, and (4) providing accountability in the face of poorly written drivers.
Agreed. Also Portal is a puzzle game. HL2 is a FPS with occasional, pretty simple puzzles. Forcing puzzles into HL2 or a significant FPS part into Portal I don't think would work. Having the portal gun in HL2 would be neat, but it wouldn't be Portal, it would be HL2 with the portal gun.
Nonsense. If IBM didn't own OS/2 they couldn't have sold it to Serenity Systems. Open sourcing is the equivalent of selling the source for zero dollars. IBM can set whatever price they like.
What if IBM's agreement with MS included additional restrictions? For instance, that IBM could sublicense OS/2, but only under an agreement that required royalties to be paid to MS for each copy distributed. Is Serenity Systems paying royalties to MS? Maybe the license only allows certain other kinds of redistribution. You certainly couldn't add a "if you copy this software you need to send $1 to Microsoft" clause to any license and have it accepted by the OSI, accepted by Debian, or combined with any GPL'd software.
I have no clue what IBM's agreement with Serenity Systems said, but somehow I doubt you do either, and your post leaves an awful lot implicit.
Actually that's not quite true either. They released the format specs a while ago. They've announced that they're making them easier to access.
I have no clue... I'm not a console gamer. ;-)
I guess that's why IBM did not develop the cell processor which is therefor not used in PS3s or why no supercomputer is built using it.
While true, this is a little bit of a red herring for this discussion. Sony gets away with using the cell because it's a console system. That games from one generation of consoles are not compatible with other generations is largely expected. Thus they can go and change the architecture willy-nilly. The same is true for Microsoft and Nintendo. (I'm not sure whether they are backwards compatible or not. But I assert that if they weren't, their sales wouldn't be substantially hit.)
OTOH, backwards compatibility with PCs are expected. The work MS has put into backwards compatibility, maintaining bug-for-bug behavior to keep even badly behaved programs working between versions is one of the major reasons for Windows's success. (It's also one of the major reasons they are hurting in security and complexity, but that's another issue.) If Dell stopped selling x86 computers and only, I dunno, Itanium, people would stop buying it. (Or, more likely, they would buy, then return it when they found out that none of their old software works.)
This isn't to say architecture changes are impossible; look at Apple's switch. There were a couple things that made this happen. First is that Apple did maintain backwards compatibility through Rosetta, and second is that Apple controls all of Apple's distribution and could therefore unilaterally make the switch. (The gaming companies also have this second reason. You don't have Microsoft making a version of the XBox 360 that isn't backwards compatible because they changed architectures and some other company staying with the x86 that was in the first version (I think that's right) so that you can continue to run old games.)
I think one of the two would be needed before you saw another architecture supplant x86 on people's desks. If AMD decided to scrap x86 and release a new chipset, say Uberisa, I don't think it would see much success. They would need to either convince MS to implement a VM (or do it themselves) that would let people run x86 code on it at a good speed, or convince all of the big manufacturers to drop x86.
Without one of those two, AMD could introduce the Uberisa, but uptake would be very slow. It might pick up steam eventually, but I predict that it would remain specialized. I mean, look at Transmeta. They've been around for a while, and provide a way to run x86. But who actually uses a Transmeta chip?
Of course, all this assumes you're going after the PC market. If you're not, then there is a lot more room for you to do what you want. But at the same time, if you're not going for the PC market, you're probably not too worried about supplanting x86.
And I would just like to point out that those are just still images, and none were of people (until you got to the bonus round). Start animating people, and you'll be able to tell the difference real quick.