Trading one centralized net for another? 

Interesting question came up in response to yesterday's piece about naming. Since name resolution is inherently centralized, aren't we trading one vulnerable system for another? Couldn't a hacker take down the name server just as easily as they swamp twitter.com with zombie requests?

First, I'm not an expert on this subject, so take everything I say with a grain of salt.

Second, let's summarize the problem we're solving.

We're trying to turn a mnemonic like judy into a URL like looselycoupled.net/judy.xml.

If the name server were to go down, one thing we lose is the ability to find new people. But our clients still know how to find all the people we were subscribed to before the attack. We also lose the ability to relocate feeds that move. But that's only temporary, and we still have HTTP-level redirects, and we don't have that ability in the RSS network today and we seem to get along okay.

Another observation. Somehow the domain name system survives these kinds of attacks (knock wood). That's all we're talking about here, a service that's no more complex than DNS. And like DNS, the net could operate indefinitely even if it went down, though we'd lose the ability to find new people or ones that moved.

One more comment, unrelated to the centralization question. A question that keeps coming up. Why not have the name be like an email address, like judy@looselycoupled.net instead of judy. There are several reasons why. 1. Twitter has nice short names, and if you want to be as easy as Twitter, you have to have short names too. 2. The goal is to map a name to a URL. If the name has all the info the URL does, why bother mapping at all? No point, just skip the whole thing and have users enter URLs.

I posted a video excerpt from Oliver Stone's movie Any Given Sunday where Al Pacino tells the team they have to fight for every inch if they want to win. I've learned that simplicity in protocols is the same. You don't get many breakthroughs, most of the simplicity comes from fighting for every inch. Stone, speaking through Pacino, tells the story infinitely better than I can.

How many RSS updates per second? 

Jim Posner asked an interesting question via email.

"I was wondering about your estimate re: the number of rss updates per second there are on the web. Crazy question I know but just trying to understand orders of magnitude.

"If I were to subscribe to every available feed on the net do you think that would be 1000/updates a sec or closer to 1M/sec. Any guidance appreciated."

I responded that I don't have any idea. But I bet it's 1000 times what Twitter does. And it never goes down and isn't subject to a DOS attack.