Tuesday, May 20, 2008

Why it's good to be messy

Perhaps developer number one will get a brainstorm and develop a clean, well-defined application which is intended to do certain things.

Meanwhile, developer number two will get a brainstorm and develop a messy, ill-defined application which could do some things, or maybe could do others, or maybe everything. (The kitchen sink approach.)

In which Ontario Emperor provides a personal example

So anyways, I was reading a FriendFeed item by Britney Mason in which she asked:

So where is the blog post about what is wrong with twitter the company? Whats going on over there? Why has no writtern something about this? I dont mean the twitout either.

Now I have written *a* blog post about what is wrong with Twitter the company, but I probably haven't written what Britney would consider *the* blog post - after all, I'm not regularly referenced in TechMeme, so I doubt that Ms. Mason has ever heard of me.

So I couldn't rightfully interject myself into Ms. Mason's conversation and say, "I have written *the* blog post about what is wrong with Twitter." Then I thought of this:

Too bad I can't d (direct message) Britney with a link to my April post.

This happens to be one of the differences between FriendFeed and Twitter. In Twitter, you can send private messages to people (as long as they follow you). In FriendFeed, you can't.

In which we postulate that FriendFeed is messy, Twitter isn't

As some of us prepare for tomorrow's Twit-Out, a few of us are going to try to use FriendFeed as a replacement for Twitter. As a result of this, I bet a bunch of us will get all sorts of ideas of things to add to FriendFeed. Actually, people are already getting ideas.

Why is it so easy to come up with ideas for FriendFeed? Because FriendFeed is perceived as "messy" - something that has a lot of things already, so adding one more thing won't mess it up too much.

Now contrast that with Twitter, which was designed for a very specific purpose - sending 140 character messages. Sure, there are wrinkles - you can send public messages to one person, or private messages to one person, and you can add stars to items, but it's still a fairly dedicated service with a very strong mission statement.

Think about some of the ideas that have been proposed for FriendFeed, and now imagine adding them to Twitter.
  • Making Twitter like a blog? No way! That breaks the all important 140 character limit, and would definitely cause a dramatic change in the service that would decrease its value for others.

  • Threading the conversations in Twitter? While in retrospect it would have been a good idea to implement threading on day 1, I think it would cause problems to introduce it at this late date.

  • Integrated interfaces with other applications? (In other words, your Twitter tweets can show up in things other than aggregators.) In the case of Twitter, this doesn't even make sense. You can't tweet to a Flickr picture. And would you want to?

  • How about the Twitter equivalent of a FriendFeed "super like" - in the case of Twitter, a "super star"? Again, this starts to break down the simplicity of Twitter. It works better in boolean, rather than having gradations of everything.
I'm going to advance the hypothesis that a messy application is better in the long run, because it's more expandable than an excellent single-purpose application.

Let's look at Twitter for a moment. If Twitter is still around in some form five years from now, it will probably be somewhat similar to the way that Twitter is today. Sure there will be a few new wrinkles, but it will still be the same basic application.

Contrast this to FriendFeed. I'd be willing to bet five super likes that FriendFeed in 2013 will be vastly different from FriendFeed in 2008. Who knows, maybe we'll be doing secure banking on FriendFeed in 2013. I doubt we'll be doing secure banking on Twitter in 2013.

In which we consider other revolutionary applications

But let's expand this beyond FriendFeed and Twitter. Are the revolutionary applications the messy ones, or the clean ones? Using PCWorld's 2007 list of the 50 best products of all time as a starting point, let's look at some revolutionary applications of the past:
  • Number 1 on InfoWorld's list was Netscape Navigator, which by its very nature was merely a presentation engine that could show all sorts of content, and which could be expanded to show other sorts of content.

  • The Apple II was, fittingly, number 2 on the list. Again, it was just an engine upon which people could put all sorts of things. A dedicated computer that could only run VisiCalc wouldn't have gone so far.

  • Number 3 is TiVo which, in my mind, is a pretty straightforward non-messy application. Sure, you can use it to store pornography or Davey & Goliath cartoons, but it's pretty much a single purpose device.

  • Number 4 is the original Napster which, in my mind, is also not that messy. People weren't about to use Napster to trade books, and I don't even think it could handle movies. Just songs. That's it.

  • Number 5 is Lotus 1-2-3, which PCWorld characterized as being better than VisiCalc. Around that time I was at a trade show, hawking a word processor and spreadsheet for the THEOS operating system, when a guy came up to me and insisted that our software had to be "integrated." He didn't even have a definition of the term, but he needed our stuff to be integrated. That's the floodgate that Lotus 1-2-3 opened. As D.J. Power noted, "Lotus 1-2-3 made it easier to use spreadsheets and it added integrated charting, plotting and database capabilities. Lotus 1-2-3 established spreadsheet software as a major data presentation package as well as a complex calculation tool. Lotus was also the first spreadsheet vendor to introduce naming cells, cell ranges and spreadsheet macros." You can see the messiness in all this, throwing charts and macros into the mix.
In which we conclude stuff

So, in my unscientific survey, it seems that more of the leading applications were "messy" rather than "clean." A messy application allows for growth, evolution, and continuing relevance, while a clean application stays locked into its original vision.

Am I off base?

Sphere: Related Content
blog comments powered by Disqus