Comments from Frankston, Reed, and Friends
Thursday, May 09, 2002
BobF at 1:46 PM [url]:
Despite the success of the Internet we are still seeing effort to create special networks for special bits. While some of the special networks exist for historic reasons, there are still new efforts to create special solutions.
The word "special" in this context is used in its negative sense of arbitrarily different. But product designers have a different way of saying it. They claim to be delivering a complete solution to a problem so that the consumer doesn't deal with things too complicated for his or her pretty little head.
But special networks are an invitation to failure since the failure of any aspect of the system means that the whole system fails. And, more problematic, the designers need to get it right the first time in order to have a coherent system. Thus the design cycle is long.
USB, Universal Serial Bus is a good example. Typically when I plug my iPAQ into my PC I get told that the device is not recognized. Yet if I take the same iPAQ and use an 802.11 card to connect it to my network it can always synchronize. Even when I use USB the system uses an IP-based connection for synchronization. So why does the PC have to know all about the iPAQ in order to allow the USB connection? What's so special about USB that requires this effort to meddle on my behalf instead of simply connecting? USB 2.0 is faster but is still the same architecture.
IEEE-1394, AKA FireWire has its own host of problems. It was designed for the oh-so-special requirements of consumer electronics. It is designed around the concept of isochronous or lock-step messaging in which every package of bits must arrive within a few millionths of a second of its designated time. Why? That's the way things had to be done in the 1930's since there was no way to buffer a signal. But as we've seen with streaming data over the Internet it is just a matter of buffering and bounding the jitter and covering lost packets. But TV can't risk losing any packets can it? Of course it can, that's inherent in MPEG2!
Alas, one byproduct of all this effort at strict timing is that if I have a digital camera (DV) and connect it to my PC and try to "upload" the video, I can lose frames because if we don't keep up with the precisely timed bits from the camera we have no way to recover because it can't happen except when it does. And there is no way to go faster because the assumption is that someone is watching the stream on a television not one of those new fangled PC things. Upload? Why is one end of the transfer more special than the other?
The idea of having my disk drives outside the PC so I can mix and match is very appealing so I have USB and 1394 drives. It actually works well because it was designed for this purpose - I plug in the disk drives and they are automatically recognized. Well, usually. Sometimes the PC gets confused. I've found the odds of this working are a tad higher on USB than 1394 but haven't looked into why.
1394 has one advantage over USB in that it is a peer network and I can connect multiple PC's to the same bus and the PC's will recognize this and create a network. That means my disk drives, cameras and other devices are available to all the devices on the network. But it isn't a network, it's a bus! So we shouldn't have to think about such things. And thus we aren't given a way to think about these things. Instead you play disk drive (or camera or ...) roulette.
There is actually a native 1394 disk drive format that is meant to be used without a PC - you just get out your huge remote control and wrestle it to the ground. You can then point to it on your television and with the finesse you've developed "programming" your VCR you can transfer the contents of a video tape to your disk drive and honor all of the rules that MPAA and others have coded into their content. You can even transfer the broadcast content (I think it is called television) to the disk drive. That is, of course, assuming that you have the right streams directed through the appropriate descramblers and don't want to watch something else on the TV at the same time and you don't have a family yelling at you while they try to watch something else.
Which brings me to today's Circuit's section in the New York Times and their story on "The Entertainment Server" "The Entertainment Server"). While the articles says that the devices will be "networked" but I can't help but chuckle (ok, groan) at the quote "'There's certainly the expectation among consumers that an TCA device or a Sony device will be relatively straightforward to use,' said Susan Kevorkian, an analyst at IDC". More telling is "'Cable companies can come in and install a media server in your home,' said Adi Kishore, a Yankee Group analyst. 'They provide it, run it and generate revenue from it.'". Not only don't they want you to worry your pretty little head, they will lose money if you can take charge and that, in their omniscience, they can make it all work the one right way. Remember these are the people who blessed you with a plethora of remote controls because they had to make sure that the set top box could enforce all of their restrictive policies.
Too bad Microsoft seems to suffer from what I call TV envy (really closed "solution" envy). Apple long ago succumbed to providing very well-crafted solutions that worked as long as you lived within its rules. While Apple, with OS/X seems to be opening up more, Microsoft's Freestyle is taking the opposite tack and giving you a crafted solution while leaving their platform lame. I don't want to be limited to Microsoft's "vision" of a home entertainment system. I would much rather they fix their "special" sound support. For some reason my ATI All-in-wonder board feeds the sound out the back and I must plug it into my sound board. After years (yes, years) of playing with this stuff I finally figured out that the "line-in" control on the playback panel allows me to take that sound and pipe it to the speakers on that board. But I can't route the signal to my headset! Instead of giving me a closed solution, I need the ability to manage these streams.
What is most disappointing is that Microsoft is doing itself a great disservice by depriving themselves of the hobbyist and developer community that has added so much value to Microsoft's systems in the past.
While many of the problems come from an attempt to maintain control (AKA forced customer loyalty) a bigger motive may be the attempt to do good to people. It is an attempt to solve the PLH (Pretty Little Head) problem by creating auto-magical solutions. If I buy a "media server" and a "TV" (I use quotes since the term TV will soon be as dated as "typewriter") they are supposed to automatically find each other. How? Don't worry your PLD - it's magic. But what if I want to choose who can get to what files or try to access may neighbor's server? PLD! Just wait for us to deliver the solution - you can't very well do it yourself. Otherwise you'll get chaos like the unruly Internet and people will get the silly idea they can do things themselves. Imaging millions of people contributing solutions!
Yeah - just imagine that.
Too bad there are so many people focused on giving us "solutions". Unfortunately when we have interacting systems there are no real solutions and any attempts simply create new limitations and frustration. It is far more valuable to just give us the means to solve our own problems. Generic connectivity is one part of it. Another part is giving us devices that simply provide interfaces that allow use to define our own controls. While it would be nice to have some standards, the best ones arise after we�ve had experience. We are at the start of an exciting learning process and companies should welcome our willingness to add value to their products.
I realize it is hard to ask companies to not give the consumers what they are asking for. It's very easy to do a marketing study and delivery exactly what people say they want even though, when they get it, it wasn't really what they meant. Giving them the means to find their own solutions is actually a safer course though it requires the bold step of rethinking product boundaries and seeing the users as partners not just "the market".
Tuesday, May 07, 2002
DPR at 12:51 PM [url]:
Time for a Silicon Tax
Remember you heard it here first on SATN...
The digital revolution is out of control. Change is rampant, and all information-based industries are being disrupted. At the root of it all is the silicon economy.
It's obvious that whoever controls silicon controls our destiny.
So it's time for Congress to act. This natural resource, silicon (not silicone), must be managed and protected. Before it's too late, we need to make sure that our national infrastructure is managed efficiently. Consequently, we must immediately begin to manage the production of this essential resource, for the public good.
We must form, as a key independent authority created with the help of the Department of Commerce, the Federal Silicon Resources Management Authority (FSRMA). This authority will be chartered to manage our limited silicon resources, issuing licenses to VLSI foundries and chip design firms for raw silicon, and licenses to glassmakers who make optical fiber and other critical resources out of silicon oxides. It will be answerable to Congress - deciding the national priorities for services that are based on silicon, such as computing, communications, etc. Of course, the silicon needs of national security will be handled directly by the DoD, coordinating through the President as needed.
The FSRMA will determine, subject to oversight by appropriate Congressional committees, whether to use market based approaches or regulation-based approaches to resolve the public good. Industry leaders will be consulted to determine the best VLSI architectures, the best computer designs, and the most productive and economically beneficial applications of silicon devices; the goal of the FSRMA will be to direct silicon resources to the best uses. To make sure that key industries are represented, appointment of commissioners to the FSRMA will be bipartisan. Industry leaders will be encouraged to serve as commissioners, in order to take full advantage of their expertise in these deeply technical questions.
Economists strongly urge the FSRMA to use market-based approaches. Since the creation of new fabrication processes and architectures involve huge risk that require large companies and huge investment, FSRMA will maintain a 10-year technology roadmap of the sorts of chips and architectures that scientists predict will be important. Then for each class of chip, the FSRMA will auction licenses to one or two vendors, who will produce the designs in such a way that they are completely compatible, to maximize the benefit to the public. In this way, the risk of wasteful competition will be reduced.
The FSRMA is also tasked with making sure that the intellectual property rights of all creative people are assured. We are assured that digital processing architectures allow each bit of information to be tagged with an author. The FSRMA will manage the evolution of computer architectures to ensure that every time a tagged bit is used or copied, a small payment is made to an account on behalf of the author.
To pay for its work, the FSRMA will tax the use of silicon in production. This taxing process will be simplified since only licensees can produce silicon devices, so there need not be a large taxation infrastructure. Similarly, the payments to authors for use of their bits will be taxed, again via the licensees.
The licensing of silicon can be used to prevent criminal use of silicon based devices and services, as well. Licensed manufacturers and service providers will have their licenses suspended or revoked if they are found to be supporting information-based criminal activities, ranging from accounting fraud to virtual child pornography. The threat of license revocation is a powerful one, and will focus the attention of our society's best academic and industrial researchers to solve these problems using new digital architectures.
Finally, the FSRMA can be used as a model for other countries and regions to manage their scarce silicon resources. The FSRMA will be encouraged to work with the Department of State to create a World Silicon Licensing Regime (WSLR). This truly 21st century political body will ensure that the silicon architectures are harmonized across national boundaries, to ensure that free interchange of information is preserved, while at the same time making sure that licensees in each country are working in the interest of all world citizens.
Coupled with the initial revenue from licensing silicon services, these taxes will grow as our digital economy grows. Wasteful competition will be reduced, because efficient use of silicon is encouraged.
When you vote in the next election, make sure you know where your candidates stand on the Silicon Tax issue. If they don't have a position, urge them to take a stand to regulate this industry. Already many corporate "good citizens" have pointed out how unregulated digital technology is disrupting American business and harming many of our business and cultural icons, destroying the economic franchises of brand names known to every household in America. If we are to remain competitive on the world stage, it is crucial to manage our silicon resources effectively.
PS: Bob Frankston reminded me that silicon is unevenly distributed, with some of the most accessible silicon dioxide deposits near the surface in many Middle Eastern nations. It's clear that American national interests in silicon, as well as oil, call for American, European, Japanese and Russian unity in stabilizing unrest there. And of course, the shoreline of many countries and island nations harbors silicon dioxide in easily extracted form, but these scarce resources are vulnerable to ocean surface increases due to global warming.