Full marks for completely missing the point, and thus to a large degree proving it.
It's pointless to attempt to reply item-by-item. The premise of the presentation is that computers have largely failed to deliver on the percieved promise of increasing productivity and making our lives easier/better.
Your points, by and large, miss this, and are a defence of computers as computers. Sure, guis and advanced programs have shifted our perspective and changed our baseline expectations, but I think the assumption that this has paid off with *more* or *better* productivity is invalid. Sure, Betsy the admin can use excel to update sales forecasts, but that doesn't make them more accurate or better. Sure, we can use Word to quickly cut/copy/paste etc, but that doesn't necessarily make the cycle time for a full document any shorter. Sure we can 'analyse numbers with impressive ease', but this doesn't improve the knowledge or understanding of those numbers. By giving us superficially accessable functionality, they distract us from the underlying task/goal. In essence, the Sirius Cybernetics Corp made real - the superficial design flaws are obscuring the inherent design flaws.
Maybe restating his point would help - we have let the practical issues of computers blind us from the larger motivations - computers have become a world and a justification unto themselves. We are 'spending' the efforts devoted to computers in the wrong places; rather than more complicated solutions which is what we're getting, we really need simpler ones that get the job done. This is as much a social point as a technical one, but it's a powerful one nonetheless. Faster compiles are all well and good, but *FEWER* compiles is really what we need.
I think most folks here are missing the point of fallacy 10. His whole presentation is about how computers have failed to deliver on the promise of making things better.
#10 is simply saying that Open Source is not magically going to make this happen. I agree. Useability and productivity of computers in the hands of the *ordinary user* are not elements that are well-handled by Open source in it's current forms, because the model IS wrong. The vast majority of OS development is done by techies doing things because they want to. They're either making tools for themselves and folks like them, or exploring technology for it's own sake. Useability in the hands of Aunt Bess is simply not something that appears on the radar screen for OS developers. This is why source code is useless - sure, if I have the source to something I can 'fix' it, but it doesn't really help in changing the authors conception of the problem or the design of the solution, which is the real problem.
The real crux, IMHO, is that OS is about flexibility, freedom and choice. Unfortunately, these are elements that detract from useability and productivity. Give someone 1 clear way to solve a problem, and it'll get solved quickly. Give them 10, and they'll never get anywhere, because they'll spend all their time evaluating/deciding. Give 10 people 10 choices, and ask them to work together/be consistent - forget it. This, ultimately, is Michi's point, near as I can tell - we are all spending WAY too much time and effort sifting through all the choices/flexibility/options, and nowhere near enough time actually doing meaningfully productive work.
For what it's worth, I really don't think he's saying that OS is _bad_, or that proprietary/commercial dev is better, just that it's not a panacea. Maybe I'm just being too generous.
A dedicated USB bus is fine for stereo, and might even handle 5.1, but the problem is that people have keyboards, mice, scanners, printers, vidcap units etc all hanging off the same 11Mbps channel. Guess what - no sound when you're printing. No sound when you're scanning. Potential dropouts when using your USB game controller.
I can't believe they didn't at least use USB 2.0 for this.
Very little of the CG for the next movies has been done. All the live-action footage is done, b ut the post-production and CG effects have not, at least as far as I understand it. They have the ability (and one would assume, the expectation) to stay current and improve their internal techniques.
It might work, but shipping bitmaps over the network is basically what VNC does, and if folks are complaining about X being slow, let's not bring up VNC.
X works very well remotely because it was designed that way - ie ship the commands, not the result of the commands because the commands are much smaller. (much the same principle as HTML or XML)
X and FB can co-exist, but not by using FB as the abstraction, because it's too low level. You need a higher-level abstraction that can render to either X or FB. This is why the port of GTK etc is a critical part of the effort. Essentially, what it will probably entail is a switch to an X server that renders only to FB, and either an FB-aware "Window Manager" or possibly local proxies to remote X clients. Could work pretty well, but will take a fair amount of work.
Use one of these Flash to IDE to set up a small Flash boot drive, then mount over the network. Shouldn't need more than 8-16 meg, so it might not be too expensive.
Video should be ok, if you get a DVD/CD-RW combo drive. Jacks the price up, but still feasible. Firewire storage would be the way to go. The biggest problem for video is how to get capture _and_ 5.1 digital out. Either/or is possible, but unless you go Firewire for audio (Motu 828 - not supported in Linux, sadly) or for capture (gotta be a solution floating around the Mac world that can be adapted), you might be out of luck.
For audio, though, it looks pretty good. PCI slot for a high-end soundcard, files either over the net or on firewire, big CPU and little else inside for DSP work. Very promising, and more affordable than I was expecting.
The big problem as I see it with trying to cram an Athlon into something like this is cooling. I have a 1.2G Athlon, and even in a mid-tower it wants 3 case fans to be completely happy. In a tiny cramped box like this, it would be trouble.
Couple that to the increased power supply as well, and making one that is even remotely quiet is going to be tough. Quiet would appear to be one of the selling points of this type of design.
Maybe the new generation of Athlons will be better in this regard.
Okay, these things aren't beautiful, but I don't find them as ugly as most. Not compared to the average case, anyway. Paint the front black and stick it behind a smoked glass door, and you're in business.
There are three things that this has above the Cappucino
- pci slot
- firewire
- cost
This is the best candidate so far for my dedicated audio box. Might not be ideal, but better than anything I've seen so far. If it's as quiet as it looks like it should be, we may have a winner.
What's the problem? Drop a better sound card in the PCI slot, and off you go. With SB Live's running $30, it's hardly a problem, and with the integration you don't need the slot for anything else.
The problem with something like a PC/104 custom component is that it probably won't have the volume to compete on price - those custom embedded boards are pricey.
Slashdot Mirror
Full marks for completely missing the point, and thus to a large degree proving it. It's pointless to attempt to reply item-by-item. The premise of the presentation is that computers have largely failed to deliver on the percieved promise of increasing productivity and making our lives easier/better. Your points, by and large, miss this, and are a defence of computers as computers. Sure, guis and advanced programs have shifted our perspective and changed our baseline expectations, but I think the assumption that this has paid off with *more* or *better* productivity is invalid. Sure, Betsy the admin can use excel to update sales forecasts, but that doesn't make them more accurate or better. Sure, we can use Word to quickly cut/copy/paste etc, but that doesn't necessarily make the cycle time for a full document any shorter. Sure we can 'analyse numbers with impressive ease', but this doesn't improve the knowledge or understanding of those numbers. By giving us superficially accessable functionality, they distract us from the underlying task/goal. In essence, the Sirius Cybernetics Corp made real - the superficial design flaws are obscuring the inherent design flaws. Maybe restating his point would help - we have let the practical issues of computers blind us from the larger motivations - computers have become a world and a justification unto themselves. We are 'spending' the efforts devoted to computers in the wrong places; rather than more complicated solutions which is what we're getting, we really need simpler ones that get the job done. This is as much a social point as a technical one, but it's a powerful one nonetheless. Faster compiles are all well and good, but *FEWER* compiles is really what we need.
I think most folks here are missing the point of fallacy 10. His whole presentation is about how computers have failed to deliver on the promise of making things better.
#10 is simply saying that Open Source is not magically going to make this happen. I agree. Useability and productivity of computers in the hands of the *ordinary user* are not elements that are well-handled by Open source in it's current forms, because the model IS wrong. The vast majority of OS development is done by techies doing things because they want to. They're either making tools for themselves and folks like them, or exploring technology for it's own sake. Useability in the hands of Aunt Bess is simply not something that appears on the radar screen for OS developers. This is why source code is useless - sure, if I have the source to something I can 'fix' it, but it doesn't really help in changing the authors conception of the problem or the design of the solution, which is the real problem.
The real crux, IMHO, is that OS is about flexibility, freedom and choice. Unfortunately, these are elements that detract from useability and productivity. Give someone 1 clear way to solve a problem, and it'll get solved quickly. Give them 10, and they'll never get anywhere, because they'll spend all their time evaluating/deciding. Give 10 people 10 choices, and ask them to work together/be consistent - forget it. This, ultimately, is Michi's point, near as I can tell - we are all spending WAY too much time and effort sifting through all the choices/flexibility/options, and nowhere near enough time actually doing meaningfully productive work.
For what it's worth, I really don't think he's saying that OS is _bad_, or that proprietary/commercial dev is better, just that it's not a panacea. Maybe I'm just being too generous.
A dedicated USB bus is fine for stereo, and might even handle 5.1, but the problem is that people have keyboards, mice, scanners, printers, vidcap units etc all hanging off the same 11Mbps channel. Guess what - no sound when you're printing. No sound when you're scanning. Potential dropouts when using your USB game controller.
I can't believe they didn't at least use USB 2.0 for this.
Very little of the CG for the next movies has been done. All the live-action footage is done, b ut the post-production and CG effects have not, at least as far as I understand it. They have the ability (and one would assume, the expectation) to stay current and improve their internal techniques.
It might work, but shipping bitmaps over the network is basically what VNC does, and if folks are complaining about X being slow, let's not bring up VNC.
X works very well remotely because it was designed that way - ie ship the commands, not the result of the commands because the commands are much smaller. (much the same principle as HTML or XML)
X and FB can co-exist, but not by using FB as the abstraction, because it's too low level. You need a higher-level abstraction that can render to either X or FB. This is why the port of GTK etc is a critical part of the effort. Essentially, what it will probably entail is a switch to an X server that renders only to FB, and either an FB-aware "Window Manager" or possibly local proxies to remote X clients. Could work pretty well, but will take a fair amount of work.
Use one of these Flash to IDE to set up a small Flash boot drive, then mount over the network. Shouldn't need more than 8-16 meg, so it might not be too expensive.
Video should be ok, if you get a DVD/CD-RW combo drive. Jacks the price up, but still feasible. Firewire storage would be the way to go. The biggest problem for video is how to get capture _and_ 5.1 digital out. Either/or is possible, but unless you go Firewire for audio (Motu 828 - not supported in Linux, sadly) or for capture (gotta be a solution floating around the Mac world that can be adapted), you might be out of luck. For audio, though, it looks pretty good. PCI slot for a high-end soundcard, files either over the net or on firewire, big CPU and little else inside for DSP work. Very promising, and more affordable than I was expecting.
The big problem as I see it with trying to cram an Athlon into something like this is cooling. I have a 1.2G Athlon, and even in a mid-tower it wants 3 case fans to be completely happy. In a tiny cramped box like this, it would be trouble. Couple that to the increased power supply as well, and making one that is even remotely quiet is going to be tough. Quiet would appear to be one of the selling points of this type of design. Maybe the new generation of Athlons will be better in this regard.
and a big, fat, melon-spitter exhaust, too!! (uh, to help with cooling, of course :-)
Okay, these things aren't beautiful, but I don't find them as ugly as most. Not compared to the average case, anyway. Paint the front black and stick it behind a smoked glass door, and you're in business. There are three things that this has above the Cappucino - pci slot - firewire - cost This is the best candidate so far for my dedicated audio box. Might not be ideal, but better than anything I've seen so far. If it's as quiet as it looks like it should be, we may have a winner.
What's the problem? Drop a better sound card in the PCI slot, and off you go. With SB Live's running $30, it's hardly a problem, and with the integration you don't need the slot for anything else. The problem with something like a PC/104 custom component is that it probably won't have the volume to compete on price - those custom embedded boards are pricey.