Friday, July 25, 2003

While tracking down another problem I found this in my SMTP log. I'm using the SMTP server that ships with every copy of XP pro. It makes a lot sense to use it instead of the one from your connectivity provider. You might as well do your own mail routing on your laptop in a hotel rather than having to struggle to find a server that is willing to accept your connection and work around any "favors" the premises' local provider is foisting on you.

-, OutboundConnectionResponse, 7/25/2003, 15:08:55, SMTPSVC1, 

    , -, 16, 0, 74, 0, 0, -, -, 554- (RTR:BB)  The IP address you are using to connect to AOL is a dynamic,

    , -, 234, 0, 78, 0, 0, -, -, 554- (residential) IP address.  AOL will not accept future e-mail transactions, -, 

This seems to have started on 24-July-2003

There is little I can say now � I�m too angry and I find this very offensive. Static IP addresses, like Ethernet MAC addresses are mechanisms associated with the plumbing of the Internet. Dynamic addresses are the only ones that scale (I plan to post more about this soon). AOL is either a demonstrating a misunderstanding of the fundamental nature of the Internet or it is opting out of real participation. Maybe it has decided that the Internet is too messy and innovative and might offer its users too much freedom. Their advertising emphasize safetly not opportunity. No wonder they are losing subscribers -- once people discover the Internet AOL is just an obstacle to real participation.

Wednesday, July 23, 2003

Dan's comments illustrate two problems. One is the problem of the reuse of the DNS entry as in WebPhotoJournals. The problems with this should be so obvious that not understanding it means not understanding the Internet itself.

The more subtle problem should still be obvious but people don't seem to about think it. The whole idea of hierarchical naming is flawed because it creates unnecessary dependencies. Instead of using paths, each should be globally unique so that it can be found without such a dependency. That's the kind of link you would put in a permanent directory. It could then point to a hierarchical mechanism like the DNS which is basically a tool for managing the Internet configuration information. Why do you have to change your email address just because your provider changes their advertising campaign and wants to use a different name? Why does ones email address depend on the name of ones service provider? (Why does a women's name depend on the husband du jour?)

It's a bit complicated since, as perhaps in the Robb example, there is the question of who owns the pages and such issues stay ambiguous until there is a forcing factor as in this case. Things get more complicated since passive URLs aren't the only kind of linkage. There are also implementation details that are solvable but I won't go into here.

For now, as I point out in my dotDNS proposal - simply separating the DNS from social policy is a JDI - Just Do It. Since it can coexist with the .com there is no reason not to provide an opportunity for escaping from the fundamentally flawed and dangerous mechanism. Dangerous - yes - just a simple typo can bring you to a hijacked fake site and exploit your trust.

Tuesday, July 22, 2003

I just posted some comments from the Always-On conference I just attended. I�d hoped to post it a few days ago but I�ve been delayed for two reasons. One is that I�m trying to do a not-too-bad job of writing which introduces delays due to thing like editing and thinking about what I want to say and making some attempt to explain rather than rant.

But rants are necessary and the current model of computing deserves a rant and I just wrote one but � well, even a rant takes a little thinking so I�ll post this and then try to wrestle the rant into shape so I can post it.

The rant is inspired by the other reason for the delay -- the recalcitrant PC. It seems to think I'm trying to do too much. The PC has lost its way and na�ve solutions like hierarchical trust models are going very much in the wrong direction.

The always-on model of connectivity is a step in the right direction but we must apply it even within a single computing system as well as to environments (such as the home) that are part of the connected community that we call the Internet. I�ll try to post the rant and a more sober set of comments on these concepts as soon as I can which may not be very soon.