I had occasion to confess my undying appreciation for xml-rpc this weekend at Drupal Camp Chicago. I might have been a little more expressive. I also might have explained this appreciation as being a product of my general disdain for web-based interfaces. Which we have to admit is kind of awkward at a conference of web-based developers.

Clearly I was being a little over the top, but it’s more or less true. I’ve come to ways of using the internet and digital data that means I spend as little time in a web browser as I have to. While the obvious--and predominant--reason for this is generally that I think most browsers are utter failures, there’s another reason that I’ve not written about explicitly: browsers lead to inefficient data consumption methods.

Browsers, with the exception of some very subtle trickery are an entirely “pull” based technology. That is to say, there’s nothing in your web-browser that comes to you and says “hey there’s something new out there for you, come look.” It’s as if the entire web is out there, just waiting for us to check it out and see if something new has changed.

This is a huge issue with the internet, particularly as data begins to migrate from our own computers to the internet--by way of web-based applications. Web programmers/content creators have a couple of ways of counteracting this problem. First, they can do little programing hacks to make the content rearrange itself even if there’s little if any new content, so it looks like it changes. Adding “interesting dynamic” content is a sort of “user hack” that gets people to check back more often with a website, and is pretty low tech solution to the problem, and sites like facebook and twitter are really good at this.

The truth is, that if you have a pull technology, where users have to check to see if there’s anything new, the only way to make it work like a “push” technology (one that notifies you when there’s new content), is to check it again and again and again. If it’s automatic the frequency is high enough it almost looks like a push. This is how most twitter-apps work, for instance. This is also how most email clients work. Though email is a push (because other people push email to your server) in most cases you have to pull email yourself, but that moves us more from the “usability effects” to “technological semantics,” and I digress.

While this might seem like a really minor complaint about how we access technology it has a profound impact on the communities that we form on the internet and how we use technology on “web applications.” Here are a few of examples:

  • “Critical mass” for online communities that are push-based (twitter, IRC channels, listservs) are much smaller than for pull-based communities (web forums, blogs). An IRC channel seems active if there are 25 people in the room and 5 people really active. A listserv with 200 subscribers is lively and bustling. A blog needs thousands of regular readers before comment threads start to seem lively, and most “successful” websites have tens of thousands of regular visitors.
  • Pull based applications must be both “useful” and “compelling,” and I often find that the later interferes with the former. This is mostly a personal preference issue, but I think it’s important, and I’d like to think that my concerns are rooted in efficiency rather than curmudgeonliness.
  • Data that we have to go out and look for means that we spend our time on the computer looking for new data rather than consuming data. I, begrudgingly, spend a lot of my internet time for instance reloading my LiveJournal friends page, because there’s no way for me get that feed all rolled up and pushed to me (even semi-pushed, like RSS.) I feel the same way about twitter, frankly.

The astute among you will realize that RSS works over pull technology, and I accept this but I think that the interaction paradigm for RSS is more like push based services. Because while your RSS reader has to pull from a lot of feeds every time you check for updates, you probably only see “unread articles,” if there are new items in your feed reader. So maybe the push/pull--while rooted in a technological division--also represents an independent interaction paradigm.

Maybe it’s not that I don’t like web-technologies (because I so clearly do), but rather that I want them to suck a lot less? It’s not so much that the applications are “in the cloud,” but that the ways we interact with them (like browser based interfaces versus desktop apps built around XML-RPC) are flawed.

Onward and Upward!