Comments from Frankston, Reed, and Friends
Thursday, May 16, 2002
BobF at 5:10 PM [url]:
Two quick follow-ups to my comments on special wires.
First is Bluetooth. Bluetooth is a topic in its own right. It is touted as the way we are going to be able to connect all those gadgets we have without wires. But Bluetooth is extremely special. You aren't even allowed the option of having a wire. I use the term "Redclaw" for the red piece of plastic on your laptops - the infrared port you never use. It is a first version of the Bluetooth protocols and is unused because it doesn't do the one thing you need - provide a simple Internet (or IP) transport. Line-of-site limitations are a factor but that's a secondary issue.
The good news is that 802.11 is being rapidly deployed. While Bluetooth only allows you to connect a limited number of nearby devices, 802.11 does that and, via the Internet, any other device in the entire world! The continued interest in Bluetooth reflects the momentum of those leading the marketing campaign and the uncritical acceptance of the wondrous claims of the proponents. As I've observed there is a tendency for people (and the press as their agent) to focus on simple promises rather than the more powerful but less specific concept of the opportunity to solve problems. In looking at the New York Times Circuits Article listing Bluetooth products the limitations of the products are striking. Why would Sony want to use a slow Bluetooth connection for video when 802.11a is 54mbps vs sharing less than one megabit per second among seven devices?
Perhaps they really believe we are going to walk around with our cell phones having video conversations. It's much nicer to video the exception so I don't have to make it obvious when I use the bathroom in the midst of a long long long conversation.
The second problem is a more dramatic example. David Reed sent me a pointer to Local governments sue FCC over cable ruling losses. Apparently towns and cities are worried about losing their revenue from cable fees. In effect they become coconspirators in maintaining the very concepts that tie them to the past.
Tuesday, May 14, 2002
DPR at 10:24 AM [url]:
Communications, information and truth
Thomas Friedman wrote a column in the NY Times that takes a quite dark view of the spread of disinformation on the Internet. My friend Dave Weinberger blogs that the issue is serious.
I think Friedman missed a more fundamental explanation of the reason for misinformation propagation. I once called it "Reed's Second Law": "Communications media exist to confirm the prejudices of their audience." It occurred to me as an explanation of the tremendous redundancy of the bits in communications space, where Shannon had defined that "real information" consisted only of those bits that told you something you didn't know, implying that any other bits would be useless, and could be compressed away.
But it seems to me that the real value of a bit is in its role as a "resistance to challenge" or in its contribution to comforting stasis. Shannon assumed that there is "one truth" for all recipients. When there are competing truths, it is essential to continually reconfirm one's own "truth" with new bits that fit into ones beliefs.
The conditions that would allow a communications medium to drive towards a "shared universal truth" are not inherent in the medium. They would be inherent in the communicating entities, if they are present at all. And it's not obvious to me that those conditions exist.
Just a reminder that information, in the technical sense, is not the same as truth.
Monday, May 13, 2002
DanB at 2:48 PM [url]:
The Address Bar bottleneck
As Bob Frankston has pointed out on this web site and in other essays, we have real problems with having only the current one-level DNS as a mechanism for turning URLs into IP addresses. We need to experiment with other ways of getting to web sites.
The Internet isn't just for "famous names" like Coca Cola, unless they want to pay for the whole thing and pay us to use it (which we won't if it's only for them). They're riding on something that serves everybody, not just them.
The current system won't pass the "side of a truck or bus" test Clay Shirky proposes. We need a way to tell people something quickly and concisely that allows them to find us in a known way. "Able Pest Control. Look under 'Pests' in the Yellow Pages." Of course, in real life the sign on the side of the truck or bus doesn't pass this test, either. If I take a delivery truck with "Four Corner's Pizza, Tel: 965-6565" written on its side to another city the "obvious" bindings don't work. (Only the locals know where "Four Corners" is -- the four gasoline stations there are now down to two -- and what about the area code? 965-6565 is not unique in the world.) Assuming the domain name "www.fourcornerspizza.com" is just as bad. There may be hundreds of "Four Corners", each with a pizza place nearby, each with a telephone exchange of "965". It turns out that "fourcornerspizza.com" is owned by someone in Phoenix, Arizona, not Newton Highlands, Massachusetts, where my Four Corners is. (At least his phone number ends in "65", too.) If you search Google for "four corners" as a phrase, you get 286,000 items -- with places in the USA, Germany, Australia, etc.
We need a way to experiment with different ways of naming things on the Internet in addition to the "unique text to IP address" bindings of the current use of DNS technology. It is crucial to making the Internet work for small businesses and individuals.
Whatever we use should probably work in places that include the Address Bar (also known as the "Location Toolbar" to Netscape users) in browsers. We also know that to do such experimentation, we need to let all comers try their hands, using something like a plug-in architecture or other open API. The users and marketplace will choose the method (or methods) that work best for the various needs.
That brings me to today's news. RealNames is out of business (News.com, Scripting News, Keith Teare's personal account). They were an attempt by an outside company to provide a way of naming web sites that works in addition to the normal DNS way. To implement it, they needed the permission and help of the browser manufacturer, Microsoft in this case. Microsoft apparently decided that they don't like that way of resolving names (a Microsoft person decided, not the marketplace) and anyway it seems Microsoft only wants things like this that they can control. This is not a good sign for resolving of the naming problem nor for advancing other important architectural issues. Hopefully Microsoft will make amends for this by completely opening up an API for address resolution in a way that does not leave themselves as a bottleneck nor as a toll taker. This is fundamental to the Internet advancing. Microsoft has a duty as the leading company in the client-side world (with officially a monopoly whether they like it or not) to do things that are in the world's maximal interest even if it's not in their specific maximal interest.