6.1 for MacOS X was released a few months ago, but 6.1 for Windows was only yesterday (Monday). Also released yesterday was 6.1.1 for MacOS X. MacOS 8/9 is still at 6.0.2. No official word from Apple that I know of, but I imagine that might be the terminal release for classic.
Alas, the release notes for 6.1 are the same as 6.0, which is odd given the amount of time between releases.
DVI input to CRT monitors
on
LCD Price Fixing?
·
· Score: 3, Insightful
The best thing about LCD's is that the display is perfect with a DVI input. Perfect pixel alignment, and no analog artifacts. LCD + VGA is almost worse than CRT + VGA, since analog errors look a LOT worse on a digital display.
Alas, i do a lot of video testing, so I need a display that is analog resizable - a CRT. But that analog noise in VGA always worries me, since it's hard to prove what image errors are due to compression, and which are due to the cable. So, what I want is a DVI CRT! Something like a LaCie ElectronBlue 22".
Anyone making anything like that. I don't mind if it's significantly more expensive than a normal monitor.
Well, I have a variety of social contracts, at different levels, with my family, my community, my industry, my nation, and humanity in general.
To the extent global society evolves in a way where everyone on the planet feels a social contract with everyone else, this is a good thing. India is an especially important place to see succeed, since it's structurally a pluralistic democracy, with lots of pressures in other directions, and nukes. Letting smart people used to working with Americans get rich and powerful there seems like a very good thing to me.
A heated argument, but completely not founded in economic reality. Nikes aren't cheap, but this isn't because of the money saved by manufacturing overseas. They're cheap because of the huge numbers of people they employ in design and marketing, and the huge fees they pay to sports figures. I'd sure that 90% of Nike's revenue stays in the USA. Here in Portland, Oregon, my neighborhood is chock full of very well paid Nike shoe designers, with wives who did graphic design for one of Nike's ad agencies before having kids. Seriously, I know five famlies within four blocks of me meeting this profile.
These are jobs I'd much rather have in the US than, say, gluing leather together. You seriously want Nike factories here?
As for the standard of living thing, you've got it backwards. Our standard of living is higher because our productivity is higher, not the other way around.
As for fleecing the consumer, how many companies are posting record profit margins these days? Investors have gotten royally fleeced these last few years, but consumers have it pretty darn good.
For specific examples of products that are cheaper due to oversease manufacturing how about, oh, everything dicsussed on Slashdot. Ever notice how kick-ass a laptop you can buy for $1000 now? Ever notice how many of those parts were built in the USA? Even as computers get better, they get cheaper, both due to outsourcing, and the pressure on prices of outsourcing.
High tariffs would erode your standard of living more via higher prices than you'd gain in job protection. Ponder the most effective free trade agreement of all time - the Constitution of the United States of America. Lots of the same arguments you're making now were made then, but I'd have to say allowing stuff into Oregon from California without us being able to charge a tariff has been a good thing. I sure wouldn't want to force every state to have its own complete supply chain for everything. Take the same principal, and apply it to the rest of the world.
So, basically, the problem is that overseas workers can in many cases produce more good code per $ than US workers? Adding a tariff will protect some jobs here, but at the greater cost of increased administrative costs of handing (and working around) the tariff, and higher software costs in general. Basically what the tariff does is make outsourcing more expensive than using local talent. Sounds nice in general, but unless the USA spends more money on software, the net effect of this is that our software dollars will buy less productivity, and hence we'll have less software. So, instead of saving an American job, it might just mean that the software project doesn't happen wholesale.
A much better approach is to find a way to justify the higher wages of American works. In order to not get outsourced, find a way to prove you're worth the money. Is that sheer code productivity, product insights, or what? There needs to be an angle. If you look at laptop manufacture, while most of the work is done around the South China Sea, product management, marketing, sales, etcetera all happens in the US. And while it sucks to be a laid off engineer, it certainly rocks to be able to buy an excellent laptop for half would an equivalent product would have cost a few years ago.
The thing about free trade is that its benefits are diffusely spread, but those it hurts are highly concentrated. So, even though steel tariffs cost more American jobs than they save, the steel industries and unions are more concentrated and vocal than the broader steel consumer industries, which is everyone from car manufacturers to can makers to appliance builders, or whatever. Higher steel costs might have saved a job in Pennsylvania, but they probably cost twice that many jobs in lots of other places. People were laid off from car dealerships because of them, even if the person laid off didn't know that was the root cause.
I've worked with several engineering projects where components were outsourced to India. The Indian engineers were are talented and motivated, and wrote good code. And, honestly, I can't say they need the money any less than "native" programmers. Our long term national interest is certainly aligned with India developming a modern, integrated economy! And I'm happy to be able to pay less for software, or buy packages that otherwise wouldn't have been written, than if all software had to be authored in the USA.
Sure, JPEG2000 certainly works better for synthetic images than JPEG did. But PNG is still better yet in my experience.
Also, since JPEG2000 doesn't use DCT, it doesn't show anywhere near the kind of blocking artifacts of JPEG. At equal SNR, JPEG2000 looks better, since the image gets softer instead of artifacty.
Most Mac browsers will let you use QuickTime to show embedded images, and so MacOS X will play JPEG 2000 just fine in browsers so configured. Of course, not many web sites are using them yet.
For high quality images. The lower the quality target, the better its relative efficiency, since it doesn't get blocking and ringing artifacts nearly as badly as the DCT based JPEG.
JPEG2000 is a LOT better for "good enough" images.
Hard, when you have infinitely variable rescaling, and add a couple hundred new photos in a bunch. Heck, even once the photos are cached, just the memory allocation to scroll around a window with several thousand of them at once is tricky.
Good question. MacOS X can import, export, embed in browers, whatever. But I haven't seen much on Windows yet.
I know there is overlap between QuickTime and JPEG, which is maybe why this has happened on Mac first. But even though QuickTime is multi-platform, it doesn't have JPEG2000 on Windows or MacOS 9.
The standard is long since complete - ISO standard since December 2000. QuickTime for MacOS X has a good implementation of it. And yes, it has both lossless and lossy modes. And yes, the core coding scheme is license and and royalty free.
http://www.jpeg.org/JPEG2000.html
I'm really looking forward to JPEG2000 for digital cameras, since instead of having to cache thumbnails, applications like iPhoto can just decode the wavelet subbands appropraite for the current resolution. Much faster than having to decode the whole JPEG and then cache a thumbnail. Browsing an iPhoto library with 2000+ files strikingly slow, and surprisingly fast considering the math that is going into it.
Still, PNG will probably be better for synthetic graphics like screen shots, where JPEG2000 will be better for natural images.
I've heard RMS does his remote email by downloading and uploading a gzipped mbox file. Given the volume of email he must get, I'm sure it's a pretty sizable file even after compression. Doing an entire mailbox will be more efficient than doing individual messages, since email messages tend to overlap in content quite a bit. Makes sense if you're going to sync only rarely.
Compression would matter to me quite a bit when I'm logging in on the road, where I might have 100 messages a day pile up. And with the trend of putting spam filtering locally ala Apple's Mail.app, making them download smaller would be nice. Heck, given how many copies of the same spam I get a day, compression would help a LOT:).
gzip or equivalent of the HTML at the proxy (hence faster HTML and browser-based email, which is really HTML)
Recompression of images at a lower quality. Nasty quality hit there, but can be better for some stuff. If they were truly bold criminals, they could switch to formats with superior compression efficiency, like going GIF to PNG, or JPEG to JPEG2000.
But are POP, SMTP, and IMAP normally compressed? Since mail docs are mainly text and aren't latency sensitive, a 75% compression would be plausible for some kinds of content.
Still, this is obvious enough that I imagine at least some mail systems must be gzipping the connection.
This being Slashdot, I'm sure someone knows all about this. I'd love to hear details about where and how compressed mail is used.
Specifically, the movies seem to have been encoded with Discreet's Cleaner, using the "High Quality First Frame" option set to JPEG instead of Basic. This feature dates from the Cinepak CD-ROM era, where the Cinepak codec had low maximum quality for still images. This feature would make the first frame a JPEG, giving better quality for the one frame that tends to get stared at the most, especially with web-based progressive download.
Of course, since the movie uses Sorenson Video 3.1 Pro, which does just fine with still images, they should have either left the feature off, or just used the Basic mode, which would have made the first frame a really big keyframe, but in the same codec as the rest of the movie.
Still, this kind of movie is perfectly legal in QuickTime, and part of the open definition of the format. The format is so complex that no one has made much of an effort to implement more than the 5% or so most commonly used features.
Good points. One of the definite trends is the use of off the shelf PC/consumer equipment in these kinds of high end uses. Bluetooth? Ten years ago, the cable you used to connect a videocamera used for news to a deck was a bizzare things that no consumer would ever have seen. Looked like something that Fox Mulder would have been tortured with. Now, we're talking a cheap thing meant for cell phones.
Still, Bluetooth doesn't have the bandwidth for even DV video. It'd have to be a more compressed bitstream than that. But the consumer electronic industry has a lot of those in progress to!
Yeah, this happens with all kinds of enabling technologies, from the typewriter on down. When fewer people are required to make a given work of art, two things happen:
The percentage of good stuff goes down.
With fewer copyeditors, or more people graduating from art school, or whatever, more of the stuff that hits the market is stuff that wouldn't have survived the process before. Think of all the crappy movies that are being shot on DV now. How many of those would have been funded if they had cost as much as a 16mm production?
The absolute amount of good stuff goes up.
Still, most of the stuff that would have happened before still happens, plus some of the new lower budget stuff actually winds up being pretty good. So there's still more good stuff to watch/read/listen to than ever before, even if the filtering process is more challenging.
And actually quite possible with current technology. You can start uploading the movie as soon as it is acquired, in parallel with the editing. Once the editing is done, you send up what is basically an Edit Decision List, telling how the final movie should be put together.
This is trivial with either Avid or Final Cut Pro, both of which can export what's called a "Reference Movie" which doesn't duplicate any local data. It basically is a metafile which says which and where from other files the data should come from. The only unique local data would be audio and any frames that have been changed from the source, like transitions. Alternatively, you could just upload the project file itself.
I've consulted on a few systems like this before. This article conflates a few different flavors of real-time broadcasting via bird.
The traditional mode uses bidirectional communciation, where the anchor can ask questions of the on-location talent. This has the advantage of being immediate (mostly). However, due to the latency of the encode and transmission, there is always a noticeable delay. These systems tend to use standards-based videoconferencing codecs like H.263. Bang for the bit isn't very good, so the quality is poor over most connections.
The next is real-time unidirectional, like a standard internet live broadcast. The video is transmitted in real-time, but the encoder uses a buffer in order to control data rate better. There can be a 15-20 second delay between something happening at it being seen on television. More modern or even proprietary formats/codecs like MPEG-4, QuickTime, and Windows Media 9 can be used. Thus, quality will be better than the bidirectional mode.
The next is "fast" where a file is compressed locally, and uploaded as a file. Most of the examples from the article of this type, encoding with tools like Movie Maker or Cleaner. The plus of this is that you can use as many bits as you want, so quality can be great, if you can afford the increased upload time. Also, since it uses TCP/IP, there isn't a risk of data corruption from dropped packets. This is fine for anything that isn't breaking news - expect at least an hour or so delay. For video broadcast, ideally interlaced encoding would be used, but it doesn't sound like it is in these examples. Squeeze certainly can't handle interlaced output for QuickTime, although it can for MPEG-4. Getting the optimum settings for encoding is my area of specialty.
Still, only a few decades ago, the nightly news was produced by guys with film cameras shooting on actual film, and then rushing to get the film developed in time for broadcast. It's amazing how quickly things change.
Ten years from now, upload will probably be built into the cameras - no laptop needed, unless editing locally.
Yeah, but having to tune into someone else's schedule is contrary to the whole point of the internet.
I suppose you could have a system where you could flag a Blog for caching, and it'd assemble the thing and let you know when it's done, ala TiVo. Then maybe have a daily multicast of each blog. But then you'd have to pick your blogs in advance.
As someone who works full-time in compressed video delivery technologies, video Blogs seem like a solution in search of a problem.
Video conferencing is a sore spot for the Mac, certainly.
However, your Gigabit ethernet objection seems rather besides the point. G4's have been shipping with Gigabit installed for over two years now - my old G4 Dual 450 has Gigabit. Upgrading a machine more ancient than doesn't seem so critical that a single, good, vendor isn't enough.
Slashdot Mirror
6.1 for MacOS X was released a few months ago, but 6.1 for Windows was only yesterday (Monday). Also released yesterday was 6.1.1 for MacOS X. MacOS 8/9 is still at 6.0.2. No official word from Apple that I know of, but I imagine that might be the terminal release for classic.
Alas, the release notes for 6.1 are the same as 6.0, which is odd given the amount of time between releases.
The best thing about LCD's is that the display is perfect with a DVI input. Perfect pixel alignment, and no analog artifacts. LCD + VGA is almost worse than CRT + VGA, since analog errors look a LOT worse on a digital display.
Alas, i do a lot of video testing, so I need a display that is analog resizable - a CRT. But that analog noise in VGA always worries me, since it's hard to prove what image errors are due to compression, and which are due to the cable. So, what I want is a DVI CRT! Something like a LaCie ElectronBlue 22".
Anyone making anything like that. I don't mind if it's significantly more expensive than a normal monitor.
Well, I have a variety of social contracts, at different levels, with my family, my community, my industry, my nation, and humanity in general.
To the extent global society evolves in a way where everyone on the planet feels a social contract with everyone else, this is a good thing. India is an especially important place to see succeed, since it's structurally a pluralistic democracy, with lots of pressures in other directions, and nukes. Letting smart people used to working with Americans get rich and powerful there seems like a very good thing to me.
A heated argument, but completely not founded in economic reality. Nikes aren't cheap, but this isn't because of the money saved by manufacturing overseas. They're cheap because of the huge numbers of people they employ in design and marketing, and the huge fees they pay to sports figures. I'd sure that 90% of Nike's revenue stays in the USA. Here in Portland, Oregon, my neighborhood is chock full of very well paid Nike shoe designers, with wives who did graphic design for one of Nike's ad agencies before having kids. Seriously, I know five famlies within four blocks of me meeting this profile.
These are jobs I'd much rather have in the US than, say, gluing leather together. You seriously want Nike factories here?
As for the standard of living thing, you've got it backwards. Our standard of living is higher because our productivity is higher, not the other way around.
As for fleecing the consumer, how many companies are posting record profit margins these days? Investors have gotten royally fleeced these last few years, but consumers have it pretty darn good.
For specific examples of products that are cheaper due to oversease manufacturing how about, oh, everything dicsussed on Slashdot. Ever notice how kick-ass a laptop you can buy for $1000 now? Ever notice how many of those parts were built in the USA? Even as computers get better, they get cheaper, both due to outsourcing, and the pressure on prices of outsourcing.
High tariffs would erode your standard of living more via higher prices than you'd gain in job protection. Ponder the most effective free trade agreement of all time - the Constitution of the United States of America. Lots of the same arguments you're making now were made then, but I'd have to say allowing stuff into Oregon from California without us being able to charge a tariff has been a good thing. I sure wouldn't want to force every state to have its own complete supply chain for everything. Take the same principal, and apply it to the rest of the world.
Well isn't that embarassing! Thanks for catching that.
Fixed.
So, basically, the problem is that overseas workers can in many cases produce more good code per $ than US workers? Adding a tariff will protect some jobs here, but at the greater cost of increased administrative costs of handing (and working around) the tariff, and higher software costs in general. Basically what the tariff does is make outsourcing more expensive than using local talent. Sounds nice in general, but unless the USA spends more money on software, the net effect of this is that our software dollars will buy less productivity, and hence we'll have less software. So, instead of saving an American job, it might just mean that the software project doesn't happen wholesale.
A much better approach is to find a way to justify the higher wages of American works. In order to not get outsourced, find a way to prove you're worth the money. Is that sheer code productivity, product insights, or what? There needs to be an angle. If you look at laptop manufacture, while most of the work is done around the South China Sea, product management, marketing, sales, etcetera all happens in the US. And while it sucks to be a laid off engineer, it certainly rocks to be able to buy an excellent laptop for half would an equivalent product would have cost a few years ago.
The thing about free trade is that its benefits are diffusely spread, but those it hurts are highly concentrated. So, even though steel tariffs cost more American jobs than they save, the steel industries and unions are more concentrated and vocal than the broader steel consumer industries, which is everyone from car manufacturers to can makers to appliance builders, or whatever. Higher steel costs might have saved a job in Pennsylvania, but they probably cost twice that many jobs in lots of other places. People were laid off from car dealerships because of them, even if the person laid off didn't know that was the root cause.
I've worked with several engineering projects where components were outsourced to India. The Indian engineers were are talented and motivated, and wrote good code. And, honestly, I can't say they need the money any less than "native" programmers. Our long term national interest is certainly aligned with India developming a modern, integrated economy! And I'm happy to be able to pay less for software, or buy packages that otherwise wouldn't have been written, than if all software had to be authored in the USA.
Sure, JPEG2000 certainly works better for synthetic images than JPEG did. But PNG is still better yet in my experience.
Also, since JPEG2000 doesn't use DCT, it doesn't show anywhere near the kind of blocking artifacts of JPEG. At equal SNR, JPEG2000 looks better, since the image gets softer instead of artifacty.
Most Mac browsers will let you use QuickTime to show embedded images, and so MacOS X will play JPEG 2000 just fine in browsers so configured. Of course, not many web sites are using them yet.
For high quality images. The lower the quality target, the better its relative efficiency, since it doesn't get blocking and ringing artifacts nearly as badly as the DCT based JPEG.
JPEG2000 is a LOT better for "good enough" images.
Hard, when you have infinitely variable rescaling, and add a couple hundred new photos in a bunch. Heck, even once the photos are cached, just the memory allocation to scroll around a window with several thousand of them at once is tricky.
Good question. MacOS X can import, export, embed in browers, whatever. But I haven't seen much on Windows yet.
I know there is overlap between QuickTime and JPEG, which is maybe why this has happened on Mac first. But even though QuickTime is multi-platform, it doesn't have JPEG2000 on Windows or MacOS 9.
Exactly,
Any statistics on how many web sites are/aren't using this?
The standard is long since complete - ISO standard since December 2000. QuickTime for MacOS X has a good implementation of it. And yes, it has both lossless and lossy modes. And yes, the core coding scheme is license and and royalty free.
http://www.jpeg.org/JPEG2000.html
I'm really looking forward to JPEG2000 for digital cameras, since instead of having to cache thumbnails, applications like iPhoto can just decode the wavelet subbands appropraite for the current resolution. Much faster than having to decode the whole JPEG and then cache a thumbnail. Browsing an iPhoto library with 2000+ files strikingly slow, and surprisingly fast considering the math that is going into it.
Still, PNG will probably be better for synthetic graphics like screen shots, where JPEG2000 will be better for natural images.
I've heard RMS does his remote email by downloading and uploading a gzipped mbox file. Given the volume of email he must get, I'm sure it's a pretty sizable file even after compression. Doing an entire mailbox will be more efficient than doing individual messages, since email messages tend to overlap in content quite a bit. Makes sense if you're going to sync only rarely.
:).
Compression would matter to me quite a bit when I'm logging in on the road, where I might have 100 messages a day pile up. And with the trend of putting spam filtering locally ala Apple's Mail.app, making them download smaller would be nice. Heck, given how many copies of the same spam I get a day, compression would help a LOT
So, it sounds like they're doing
gzip or equivalent of the HTML at the proxy (hence faster HTML and browser-based email, which is really HTML)
Recompression of images at a lower quality. Nasty quality hit there, but can be better for some stuff. If they were truly bold criminals, they could switch to formats with superior compression efficiency, like going GIF to PNG, or JPEG to JPEG2000.
While it's possible to do 5:1 on some HTML stuff, that assumes that mod_gzip or other compression schemes aren't being used.
And for images, they're presumably just reencoding them at a higher compression ratio, ala AOL. Which can work if you prefer crappy quality, faster.
But TANSTAAFL always applies.
But are POP, SMTP, and IMAP normally compressed? Since mail docs are mainly text and aren't latency sensitive, a 75% compression would be plausible for some kinds of content.
Still, this is obvious enough that I imagine at least some mail systems must be gzipping the connection.
This being Slashdot, I'm sure someone knows all about this. I'd love to hear details about where and how compressed mail is used.
Specifically, the movies seem to have been encoded with Discreet's Cleaner, using the "High Quality First Frame" option set to JPEG instead of Basic. This feature dates from the Cinepak CD-ROM era, where the Cinepak codec had low maximum quality for still images. This feature would make the first frame a JPEG, giving better quality for the one frame that tends to get stared at the most, especially with web-based progressive download.
Of course, since the movie uses Sorenson Video 3.1 Pro, which does just fine with still images, they should have either left the feature off, or just used the Basic mode, which would have made the first frame a really big keyframe, but in the same codec as the rest of the movie.
Still, this kind of movie is perfectly legal in QuickTime, and part of the open definition of the format. The format is so complex that no one has made much of an effort to implement more than the 5% or so most commonly used features.
Good points. One of the definite trends is the use of off the shelf PC/consumer equipment in these kinds of high end uses. Bluetooth? Ten years ago, the cable you used to connect a videocamera used for news to a deck was a bizzare things that no consumer would ever have seen. Looked like something that Fox Mulder would have been tortured with. Now, we're talking a cheap thing meant for cell phones.
Still, Bluetooth doesn't have the bandwidth for even DV video. It'd have to be a more compressed bitstream than that. But the consumer electronic industry has a lot of those in progress to!
Yeah, this happens with all kinds of enabling technologies, from the typewriter on down. When fewer people are required to make a given work of art, two things happen:
The percentage of good stuff goes down.
With fewer copyeditors, or more people graduating from art school, or whatever, more of the stuff that hits the market is stuff that wouldn't have survived the process before. Think of all the crappy movies that are being shot on DV now. How many of those would have been funded if they had cost as much as a 16mm production?
The absolute amount of good stuff goes up.
Still, most of the stuff that would have happened before still happens, plus some of the new lower budget stuff actually winds up being pretty good. So there's still more good stuff to watch/read/listen to than ever before, even if the filtering process is more challenging.
Intriguing idea.
And actually quite possible with current technology. You can start uploading the movie as soon as it is acquired, in parallel with the editing. Once the editing is done, you send up what is basically an Edit Decision List, telling how the final movie should be put together.
This is trivial with either Avid or Final Cut Pro, both of which can export what's called a "Reference Movie" which doesn't duplicate any local data. It basically is a metafile which says which and where from other files the data should come from. The only unique local data would be audio and any frames that have been changed from the source, like transitions. Alternatively, you could just upload the project file itself.
I've consulted on a few systems like this before. This article conflates a few different flavors of real-time broadcasting via bird.
The traditional mode uses bidirectional communciation, where the anchor can ask questions of the on-location talent. This has the advantage of being immediate (mostly). However, due to the latency of the encode and transmission, there is always a noticeable delay. These systems tend to use standards-based videoconferencing codecs like H.263. Bang for the bit isn't very good, so the quality is poor over most connections.
The next is real-time unidirectional, like a standard internet live broadcast. The video is transmitted in real-time, but the encoder uses a buffer in order to control data rate better. There can be a 15-20 second delay between something happening at it being seen on television. More modern or even proprietary formats/codecs like MPEG-4, QuickTime, and Windows Media 9 can be used. Thus, quality will be better than the bidirectional mode.
The next is "fast" where a file is compressed locally, and uploaded as a file. Most of the examples from the article of this type, encoding with tools like Movie Maker or Cleaner. The plus of this is that you can use as many bits as you want, so quality can be great, if you can afford the increased upload time. Also, since it uses TCP/IP, there isn't a risk of data corruption from dropped packets. This is fine for anything that isn't breaking news - expect at least an hour or so delay.
For video broadcast, ideally interlaced encoding would be used, but it doesn't sound like it is in these examples. Squeeze certainly can't handle interlaced output for QuickTime, although it can for MPEG-4. Getting the optimum settings for encoding is my area of specialty.
Still, only a few decades ago, the nightly news was produced by guys with film cameras shooting on actual film, and then rushing to get the film developed in time for broadcast. It's amazing how quickly things change.
Ten years from now, upload will probably be built into the cameras - no laptop needed, unless editing locally.
Actually, with modern codecs, 110 is plenty for "entertainment quality" listening.
Some formats that sounds great at 96 Kbps:
AAC-LC (from MPEG-4) (the one in QuickTime 6.1 is pretty good, but not the one from 6.0)
WMA9 2-pass VBR
RealAudio 8 Stereo Music
Ogg Vorbis and a tweaked lame --abr can certainly do more than good enough for workout music at 96 Kbps as well.
Some of the next generation stuff, like AAC-SBR, shoot for "sounds like CD quality" at 48 Kbps or lower.
Yeah, but having to tune into someone else's schedule is contrary to the whole point of the internet.
I suppose you could have a system where you could flag a Blog for caching, and it'd assemble the thing and let you know when it's done, ala TiVo. Then maybe have a daily multicast of each blog. But then you'd have to pick your blogs in advance.
As someone who works full-time in compressed video delivery technologies, video Blogs seem like a solution in search of a problem.
Video conferencing is a sore spot for the Mac, certainly.
However, your Gigabit ethernet objection seems rather besides the point. G4's have been shipping with Gigabit installed for over two years now - my old G4 Dual 450 has Gigabit. Upgrading a machine more ancient than doesn't seem so critical that a single, good, vendor isn't enough.