Exactly how do you back this up? Graphcially, it has changed drastically over time. The stories told are alwaysmostly distinct, with a few common themes. And the battle system, though similar to FF4 has changeed ever time. You can especially see this with non FF games, where they have tried all sorts of experiments with battle systems and story telling. Read the article and you'll get a good overview of this. The problem for Square is that every time they try something new, their tried and true customers hate it because it's not like the previous release. Final Fantasy 8 is arguably the most different game of the series, going away from the importance of money and equipment, and this got a lot of criticism. When you get to really different games (SaGa Frontier) they get seriously criticized from a lack of narrative cohesion among a number of different plots. The moral of the story is don't fix what isn't broken. The Final Fantasy model works well, and the gameplay is smooth. It's not gameplay that is the focus of updates to the series, but rather story and graphic content. Try playing the FFX demo sometime if you can. Roughly the same battle system, but with improvements. The graphics are freaking amazing, and the first demo hints at a rather intriguing plot (even if blitzball seems like a dumb idea). I particularly thought Auron's voice was well-cast.
Just once still I suppose, they never claimed otherwise in the other two movies. The question should be, how many times is John Conner gonna need his ass saved by some futuristic cyborg?
Easy. One, 60 bucks an hour is a lot higher than I would expect to get for a regular job... Maybe for service calls or very short-term contracts, but on the order of 30 an hours sounds about right for a regular job.
In any case, for those that pursue this, it isn't money, it's really good fun. We have much more highly powered equipment to do what we want, but it's fun to get cheap hardware and see how hard we can push it.. For example, I think the dreamast, at 50 bucks, is a fantastic deal (especially since many can boot off of CD-R without any hardware modificaiotn). That's the main reason why I bought a dreamcast. I have linux running on it occasionally, but without any storage space nor any network connectivity, it is mostly useless, just kinda neat. This was a bit too easy, but it illustrates the point..
Also note that if your nVidia has TV capture ability, it's not likely to have Video4Linux compatibility. ATI All-in-Wonder cards have historically had this ability, through the GATOS project.
As far as TV-Out goes, it really annoys me how ATI protects this for the express purpose of protecting the bits that control MacroVision. I mean, under Windows there are always hacks for MacroVision even when there is no documentation available, and if you are a linux head and want to do this, you can use the framebuffer and XFree FBDev and mplayer in console mode to acheive TV-out that sucks for normal usages (unaccelerated) but suffices for those who want to copy stuff to tape.
All that said, the TV-Capture capabilities combined with a really nice 3D-chipset and open drivers make me want a Radeon AIW 8500DV. I'll wait til GATOS has official support though.
The problem with binary drivers is that:
1) You are stuck with Linux on x86. No *BSD, no alpha/powerpc/etc.
2) You are stuck with what linux kernel that nVidia deems ok. This may be fine for now, but when nVidia releases new products, and cease support of older ones, when you upgrade your distro to something with, say kernel 3.0, your screwed because they only support the GeForce 4 and newer (hypothetical future)
Even with corporate edition there is no guarantee there is no spyware stuff going on. I for fun did a corp ed. install somewhere behind a NATed private network. I logged all attempts by the installing computer to contact any machine on MSes subnet, and guess what, there were about 12 connection attempts throughout the install process, about 8 of which had no warning nor rationale, the others were along the lines of "testing your network connectivity".
I really ought to more officially document it, including tcpdumps of whatever the system is sending to MS and what is in the response, if any, from MS. Just because you aren't sending in for Product Activation doesn't mean it's not spyware. Also, I re-installed the OS later after formatting and both logged and blocked to see if it changed the install at all, and the install performed almost exactly as it had before, except a bit slower in places where it would timeout on a connection and try another IP until the list of IPs was exhausted. So the net connects don't seem to positively affect the install whatsoever.
Don't believe me? Try the experiment for yourself. Being a sys-admin I can toy around with the company firewall and license, but at home you can set up something similar.
You need to have perspective. Flash back to the days of Windows 95 and think about the state of Linux. Is anyone going to help you install something that ancient? Hell no. Or even Windows 98. If someone told you that their Graphics card didn't work in RedHat 5.2, what would you do? Tell them to upgrade to something that isn't so ancient. Under linux, upgrade paths are cheap, so it isn't such a big deal to upgrade, and thus the Linux community maintains a much shorter lifecycle in general than MS. Don't pretend to say that it won't go out of date, it will go out of date much faster than windows. It's just that upgrading to keep up is actually financially feasible with Linux distros.
Keep running Win95. Just because MS says it won't support it any more, doesn't mean that it will magically stop working.
Alternatively, upgrade your graphics card or look harder on the net for drivers. Then you can run, say, Win98, but you won't see much of a difference except IE and now newer DirectX availability, which I doubt matters on a 100 MHz machine which won't run most any DX8.1 applications anyway.
The lifecycle on their products is relatively long and overlay a great deal. As such, even though Windows 95 is officially unsupported, they still must support Windows 98, which was compatible with Windows 95 nearly completely. They are still stuck with the win32 API. When 2003 rolls around and Win98 is axed, then WinME will carry the support, then when ME is axed, XP will be supported, etc....
Though they can make minor changes, but for now backwards compatibility prevents them from axing any backwards compatiblity.
The problem for Wine remains the same, the API is huge and not well documented, and while not deprecating calls, they are still adding calls every release. As far as releases not being made with Win95 in mind by 3rd party companies, that has been and will remain their pergative. Some already say "no, we don't support that" Others will continue to test against it even if MS says it's unsupported.
Wine is catching up really fast, and the Win32 API is changing slowly (not a bad thing). I doubt MS sees Wine as that much of a threat right now. Wine is only useful for Desktop-level applications, and MS's only real threat is in the server arena, where all applications are run natively and thus wine becomes a moot point. The relatively small segment of Desktop linux users doesn't cut much into MS's bottom line. That's the whole deal with the.NET strategy, leveraging the Desktop dominance to get more major companies running.NET servers...
Why do you want a remote control starter in the first place? This isn't even a case of lazniess, you *have* to sit in the car in order to do anything useful with it. And you say you feel uncomfortable about leaving a key in your unit in order to make it functional, but if they are already in your car and that far anyway, what would they need your *key* for? Starting the car? Well, if they are that far in they would have hot-wired if you didn't have a system in place.
Besdies, a remote control car starter just sounds like a *really* bad idea. No benefit, all kinds of possible security breaches. You want to make your car easier to steal for no good reason?
I'll admit that the fancy electronics are pushing out the really small-scale mechanics, but it is by no means microsoft tactics. They want to improve cars, make them harder to steal, more convenient and efficient. Yes, you may have a bit more proprietary stuff in each car, but I'll wager that even if the accessories are produced by a single company now, in the future other companies will have the circuitry for the different models just like they do forconventional parts.
This is one of the most oddball ideas I've seen on Slashdot.
While using the win32 libs is still a very popular widespread method, ffmpeg and opendivx native linux cores are often used to play back divx content. My DivX with MP3 audio avi files play without any win32 stuff needed (so non-x86 DivX playback is possible). I keep win32 stuff around for Indeo avis (Xanim has a native module, but doesn't support mp3 audio tracks) and other random stuff, but for the DivX I don't need dlls...
lamp? Is that what you use? You might want to check out PythonTheater (http://xtheater.sourceforge.net/) or mplayer (http://mplayer.sourceforge.net/) if you don't want any gui.
My experience with MPG is that mplayer and SMPEG-based projects play more of my MPGs than either WMP or Xing. between those two I can play them all, but I have one MPG that only works with Xing and smpeg...
Anyway, avifile-0.6 based products cover all my wmvs, avis, and asfs, and mplayer does the same.
For Quicktime, I use Xanim (http://xanim.va.pubnix.com/) when they are older, and for the Sorenson based stuff I resort to wine to run the Windows Quicktime player, which works ok, except the Interface gets a bit garbled, though the movie looks fine. CodeWeavers CrossOver plugin seems to resolve these issues if you're willing to pay.
Multimedia playback under linux is great. The only format that I have not been able to view under linux one way or another is VIVO, and mplayer seems to be trying to get that working even.
Another piece of advice when dealing with realmedia under linux, have both the Realplayer 8 installed and the RealOne alpha installed. Use the alpha when you can get away with it as it takes advantage of the XVideo extension and does fullscreen well. Real8 is needed to fall back on when RealOne flakes out (as it often does)
Read carefully, he says to *succeed* as a standard. Yes, it can be considered a standard, even if it isn't widely used, but then it wouldn't be successful. Of course to even suggest that MPEG-4 is not really called for is silly. I mean look at all the DivX and related MPEG-4 based codecs that came out because people were essentially to wait for the standard to be finalized. Judging by these codecs, MPG-4 is a *really* necessary thing. High quality and low space usage. Maybe it doesn't scale as high as MPG-2, but for space-critical multimedia, MPG-4 is an excellent compromise.
Doesn't matter, the Quicktime format is generally made publicly known, and MPEG-4, of course, will be as well. We already have Quicktime for Linux. It's just that they only have the more open codecs which are not popular. For older Quicktime closed codecs there is Xanim. I would wager that the Quicktime for Linux project will be able to add MPEG-4 support and then get a *lot* more useful. Until then there is always Wine (which does work if you coax it enough). If you can't figure out how to make Wine do it, there is always CodeWeavers Crossover plugin. Of course the Wine solutions are x86 only, but with the use of MPEG-4 on the horizon, maybe we will have a more cross-platform style. Of course, it seems silly that all these formats are converging on the same codecs. Nice for developers, but then why bother with.mov files when.avi files will have identical codecs for the most part?
But they were encoded as normal CDs mostly. They just had an extra country code and bad blocks, and it was up to the hardware of the Playstation to explicitly check for these things. The modchip just feeds a false positive for these checks. If people swapped discs with the door sensor propped "closed" just after the code and blocks were verified but before boot data was loaded, then no modchip is necessary, but makes things much more convenient. So if it can load PS1 cds, it can *physically* load normal CD-ROMS, just as it can play normal music CDs... I would imagine a modified driver could likely bypass these checks, and the possibility of this happening makes them try to keep the CD/DVD drivers as far away from the hacking community as possible.
Actually, as of late a number of non-x86 only solutions are available. OpenDivX and ffmpeg for two are cross-platform, and have been used along with cross-platform mp3 codecs to provide playback of DivX avis on more platforms. Try the CVS version of avifile-0.6, or mplayer. I haven't had a chance to try, but the included aviplay or another program (i.e. PythonTheater) should work on non-x86 for some avis...
Ummm... Drag and Drop *does* pretty much work that way. XMMS supports drops from at least Nautilus and GMC, and I think Konqueror as well as ROX.
For the end user, Drag and Drop is very much a reality.
For Developers, there are some issues, I admit.
I for one, in my projects test my DnD against releases of Nautilus, Konqeuror, GMC, and Rox. If you can accept from Nautilus and from Konqueror, you will be able to accept from most anywhere. The problem I see from a developer standpoint is that while all the DnDs are now more similarly implemented, there are still enough differences so that implementing DnD interoperability with other apps is not as trivial as it should be. For example, the action type maybe copy from one app and move from another, even in the exact same context. Also, the formats of filenames differs. It could be prefixed with file:, file:/, file://, or nothing at all depening on the app. Could be suffixed with \r,\n, a space, an additional embedded NULL, simple NULL termination, or any combination of the afformentioned terminators. And in encoding the string, some programs pass it as it is, some do URI encoding (space becomes %20, etc) with capital letters for hex letters, some use lower case.
The prefix is pretty easy to catch, correct, and understand, but the inconsistancy with action types, encoding, and the various weird string-termination schemes is really hard for a developer to catch or develop around, and there is no real good reason for so many different ways of ending a string. If only projects could converge on a simple standard for this small issue, it would be great.
I should've asked this in my comment, but does anyone know if there is a NAT implementation that allows you to specify more than one interface/address for a NAT rule?
Make it even better, use a full-fledged PC with three interfaces to serve as router (one address for each connection, and one internally). Though I don't know of any way to do it now, I would assume it could be a logical extension of NAT to NAT over two interfaces rather than one and use load balancing on outgoing traffic to figure out where to NAT the traffic through. If one went down, automatically put everything on the remaining connection. Higher throughput dynamically managed (more efficient than manual allocation) and failover, all without you needing to do a lot of manual work to keep things balanced and working right. All of this is assuming a non-routable private subnets, which is for many companies out there unacceptable...
Slashdot Mirror
Try the FFX demo. I don't know about frequency of random battles, but the flow is much quicker and starts really quickly..
Exactly how do you back this up? Graphcially, it has changed drastically over time. The stories told are alwaysmostly distinct, with a few common themes. And the battle system, though similar to FF4 has changeed ever time. You can especially see this with non FF games, where they have tried all sorts of experiments with battle systems and story telling. Read the article and you'll get a good overview of this. The problem for Square is that every time they try something new, their tried and true customers hate it because it's not like the previous release. Final Fantasy 8 is arguably the most different game of the series, going away from the importance of money and equipment, and this got a lot of criticism. When you get to really different games (SaGa Frontier) they get seriously criticized from a lack of narrative cohesion among a number of different plots. The moral of the story is don't fix what isn't broken. The Final Fantasy model works well, and the gameplay is smooth. It's not gameplay that is the focus of updates to the series, but rather story and graphic content. Try playing the FFX demo sometime if you can. Roughly the same battle system, but with improvements. The graphics are freaking amazing, and the first demo hints at a rather intriguing plot (even if blitzball seems like a dumb idea). I particularly thought Auron's voice was well-cast.
Just once still I suppose, they never claimed otherwise in the other two movies. The question should be, how many times is John Conner gonna need his ass saved by some futuristic cyborg?
Easy. One, 60 bucks an hour is a lot higher than I would expect to get for a regular job... Maybe for service calls or very short-term contracts, but on the order of 30 an hours sounds about right for a regular job.
In any case, for those that pursue this, it isn't money, it's really good fun. We have much more highly powered equipment to do what we want, but it's fun to get cheap hardware and see how hard we can push it.. For example, I think the dreamast, at 50 bucks, is a fantastic deal (especially since many can boot off of CD-R without any hardware modificaiotn). That's the main reason why I bought a dreamcast. I have linux running on it occasionally, but without any storage space nor any network connectivity, it is mostly useless, just kinda neat. This was a bit too easy, but it illustrates the point..
My experience is that it trys three times and gives up. At least at install time, haven't bothered to check at subsequent bootups..
Also note that if your nVidia has TV capture ability, it's not likely to have Video4Linux compatibility. ATI All-in-Wonder cards have historically had this ability, through the GATOS project.
As far as TV-Out goes, it really annoys me how ATI protects this for the express purpose of protecting the bits that control MacroVision. I mean, under Windows there are always hacks for MacroVision even when there is no documentation available, and if you are a linux head and want to do this, you can use the framebuffer and XFree FBDev and mplayer in console mode to acheive TV-out that sucks for normal usages (unaccelerated) but suffices for those who want to copy stuff to tape.
All that said, the TV-Capture capabilities combined with a really nice 3D-chipset and open drivers make me want a Radeon AIW 8500DV. I'll wait til GATOS has official support though.
The problem with binary drivers is that:
1) You are stuck with Linux on x86. No *BSD, no alpha/powerpc/etc.
2) You are stuck with what linux kernel that nVidia deems ok. This may be fine for now, but when nVidia releases new products, and cease support of older ones, when you upgrade your distro to something with, say kernel 3.0, your screwed because they only support the GeForce 4 and newer (hypothetical future)
Even with corporate edition there is no guarantee there is no spyware stuff going on. I for fun did a corp ed. install somewhere behind a NATed private network. I logged all attempts by the installing computer to contact any machine on MSes subnet, and guess what, there were about 12 connection attempts throughout the install process, about 8 of which had no warning nor rationale, the others were along the lines of "testing your network connectivity".
I really ought to more officially document it, including tcpdumps of whatever the system is sending to MS and what is in the response, if any, from MS. Just because you aren't sending in for Product Activation doesn't mean it's not spyware. Also, I re-installed the OS later after formatting and both logged and blocked to see if it changed the install at all, and the install performed almost exactly as it had before, except a bit slower in places where it would timeout on a connection and try another IP until the list of IPs was exhausted. So the net connects don't seem to positively affect the install whatsoever.
Don't believe me? Try the experiment for yourself. Being a sys-admin I can toy around with the company firewall and license, but at home you can set up something similar.
You need to have perspective. Flash back to the days of Windows 95 and think about the state of Linux. Is anyone going to help you install something that ancient? Hell no. Or even Windows 98. If someone told you that their Graphics card didn't work in RedHat 5.2, what would you do? Tell them to upgrade to something that isn't so ancient. Under linux, upgrade paths are cheap, so it isn't such a big deal to upgrade, and thus the Linux community maintains a much shorter lifecycle in general than MS. Don't pretend to say that it won't go out of date, it will go out of date much faster than windows. It's just that upgrading to keep up is actually financially feasible with Linux distros.
Keep running Win95. Just because MS says it won't support it any more, doesn't mean that it will magically stop working.
Alternatively, upgrade your graphics card or look harder on the net for drivers. Then you can run, say, Win98, but you won't see much of a difference except IE and now newer DirectX availability, which I doubt matters on a 100 MHz machine which won't run most any DX8.1 applications anyway.
The lifecycle on their products is relatively long and overlay a great deal. As such, even though Windows 95 is officially unsupported, they still must support Windows 98, which was compatible with Windows 95 nearly completely. They are still stuck with the win32 API. When 2003 rolls around and Win98 is axed, then WinME will carry the support, then when ME is axed, XP will be supported, etc.... .NET strategy, leveraging the Desktop dominance to get more major companies running .NET servers...
Though they can make minor changes, but for now backwards compatibility prevents them from axing any backwards compatiblity.
The problem for Wine remains the same, the API is huge and not well documented, and while not deprecating calls, they are still adding calls every release. As far as releases not being made with Win95 in mind by 3rd party companies, that has been and will remain their pergative. Some already say "no, we don't support that" Others will continue to test against it even if MS says it's unsupported.
Wine is catching up really fast, and the Win32 API is changing slowly (not a bad thing). I doubt MS sees Wine as that much of a threat right now. Wine is only useful for Desktop-level applications, and MS's only real threat is in the server arena, where all applications are run natively and thus wine becomes a moot point. The relatively small segment of Desktop linux users doesn't cut much into MS's bottom line. That's the whole deal with the
Why do you want a remote control starter in the first place? This isn't even a case of lazniess, you *have* to sit in the car in order to do anything useful with it. And you say you feel uncomfortable about leaving a key in your unit in order to make it functional, but if they are already in your car and that far anyway, what would they need your *key* for? Starting the car? Well, if they are that far in they would have hot-wired if you didn't have a system in place.
Besdies, a remote control car starter just sounds like a *really* bad idea. No benefit, all kinds of possible security breaches. You want to make your car easier to steal for no good reason?
I'll admit that the fancy electronics are pushing out the really small-scale mechanics, but it is by no means microsoft tactics. They want to improve cars, make them harder to steal, more convenient and efficient. Yes, you may have a bit more proprietary stuff in each car, but I'll wager that even if the accessories are produced by a single company now, in the future other companies will have the circuitry for the different models just like they do forconventional parts.
This is one of the most oddball ideas I've seen on Slashdot.
You may wish to try a newer build. I know that OpenDivx plays Divx3 content under linux and windows on my sys, encoding, however, is another story
While using the win32 libs is still a very popular widespread method, ffmpeg and opendivx native linux cores are often used to play back divx content. My DivX with MP3 audio avi files play without any win32 stuff needed (so non-x86 DivX playback is possible). I keep win32 stuff around for Indeo avis (Xanim has a native module, but doesn't support mp3 audio tracks) and other random stuff, but for the DivX I don't need dlls...
lamp? Is that what you use? You might want to check out PythonTheater (http://xtheater.sourceforge.net/) or mplayer (http://mplayer.sourceforge.net/) if you don't want any gui.
My experience with MPG is that mplayer and SMPEG-based projects play more of my MPGs than either WMP or Xing. between those two I can play them all, but I have one MPG that only works with Xing and smpeg...
Anyway, avifile-0.6 based products cover all my wmvs, avis, and asfs, and mplayer does the same.
For Quicktime, I use Xanim (http://xanim.va.pubnix.com/) when they are older, and for the Sorenson based stuff I resort to wine to run the Windows Quicktime player, which works ok, except the Interface gets a bit garbled, though the movie looks fine. CodeWeavers CrossOver plugin seems to resolve these issues if you're willing to pay.
Multimedia playback under linux is great. The only format that I have not been able to view under linux one way or another is VIVO, and mplayer seems to be trying to get that working even.
Another piece of advice when dealing with realmedia under linux, have both the Realplayer 8 installed and the RealOne alpha installed. Use the alpha when you can get away with it as it takes advantage of the XVideo extension and does fullscreen well. Real8 is needed to fall back on when RealOne flakes out (as it often does)
Read carefully, he says to *succeed* as a standard. Yes, it can be considered a standard, even if it isn't widely used, but then it wouldn't be successful. Of course to even suggest that MPEG-4 is not really called for is silly. I mean look at all the DivX and related MPEG-4 based codecs that came out because people were essentially to wait for the standard to be finalized. Judging by these codecs, MPG-4 is a *really* necessary thing. High quality and low space usage. Maybe it doesn't scale as high as MPG-2, but for space-critical multimedia, MPG-4 is an excellent compromise.
Doesn't matter, the Quicktime format is generally made publicly known, and MPEG-4, of course, will be as well. We already have Quicktime for Linux. It's just that they only have the more open codecs which are not popular. For older Quicktime closed codecs there is Xanim. I would wager that the Quicktime for Linux project will be able to add MPEG-4 support and then get a *lot* more useful. Until then there is always Wine (which does work if you coax it enough). If you can't figure out how to make Wine do it, there is always CodeWeavers Crossover plugin. Of course the Wine solutions are x86 only, but with the use of MPEG-4 on the horizon, maybe we will have a more cross-platform style. Of course, it seems silly that all these formats are converging on the same codecs. Nice for developers, but then why bother with .mov files when .avi files will have identical codecs for the most part?
But they were encoded as normal CDs mostly. They just had an extra country code and bad blocks, and it was up to the hardware of the Playstation to explicitly check for these things. The modchip just feeds a false positive for these checks. If people swapped discs with the door sensor propped "closed" just after the code and blocks were verified but before boot data was loaded, then no modchip is necessary, but makes things much more convenient. So if it can load PS1 cds, it can *physically* load normal CD-ROMS, just as it can play normal music CDs... I would imagine a modified driver could likely bypass these checks, and the possibility of this happening makes them try to keep the CD/DVD drivers as far away from the hacking community as possible.
I think the reports of problems have been greatly exagerrated. I've played many DVDs on my PS2 without issue..
Actually, as of late a number of non-x86 only solutions are available. OpenDivX and ffmpeg for two are cross-platform, and have been used along with cross-platform mp3 codecs to provide playback of DivX avis on more platforms. Try the CVS version of avifile-0.6, or mplayer. I haven't had a chance to try, but the included aviplay or another program (i.e. PythonTheater) should work on non-x86 for some avis...
Ummm... Drag and Drop *does* pretty much work that way. XMMS supports drops from at least Nautilus and GMC, and I think Konqueror as well as ROX.
For the end user, Drag and Drop is very much a reality.
For Developers, there are some issues, I admit.
I for one, in my projects test my DnD against releases of Nautilus, Konqeuror, GMC, and Rox. If you can accept from Nautilus and from Konqueror, you will be able to accept from most anywhere. The problem I see from a developer standpoint is that while all the DnDs are now more similarly implemented, there are still enough differences so that implementing DnD interoperability with other apps is not as trivial as it should be. For example, the action type maybe copy from one app and move from another, even in the exact same context. Also, the formats of filenames differs. It could be prefixed with file:, file:/, file://, or nothing at all depening on the app. Could be suffixed with \r,\n, a space, an additional embedded NULL, simple NULL termination, or any combination of the afformentioned terminators. And in encoding the string, some programs pass it as it is, some do URI encoding (space becomes %20, etc) with capital letters for hex letters, some use lower case.
The prefix is pretty easy to catch, correct, and understand, but the inconsistancy with action types, encoding, and the various weird string-termination schemes is really hard for a developer to catch or develop around, and there is no real good reason for so many different ways of ending a string. If only projects could converge on a simple standard for this small issue, it would be great.
ISS systems? So the International Space Station systems are overintegrated? :)
I mean, look at this logo on a nokia phone. As soon as you see this logo on a phone, you know trouble is coming. I think it is some sort of curse :)
Btw, if you actually want this logo, go here.
I should've asked this in my comment, but does anyone know if there is a NAT implementation that allows you to specify more than one interface/address for a NAT rule?
Make it even better, use a full-fledged PC with three interfaces to serve as router (one address for each connection, and one internally). Though I don't know of any way to do it now, I would assume it could be a logical extension of NAT to NAT over two interfaces rather than one and use load balancing on outgoing traffic to figure out where to NAT the traffic through. If one went down, automatically put everything on the remaining connection. Higher throughput dynamically managed (more efficient than manual allocation) and failover, all without you needing to do a lot of manual work to keep things balanced and working right. All of this is assuming a non-routable private subnets, which is for many companies out there unacceptable...
Robots already have idols: Battlebots
Granted, they are Remote controlled rather than AI, but they should suffice as Lego Mindstorm idols..