kevin cabral's blog: bits+thesis

The excellent Barbarian Blog has a great piece up on why Twitter is not simple. That’s been apparent to me since I began using it myself and even more so when I tried to teach my web/blog-savvy mother how to use it.

I’d like to extend the dialogue by speaking to why Twitter is complicated and providing my thoughts on whether to try to “fix” it.

First, product design necessarily tackles three dimensions: business objectives, design hypotheses and user interaction with them. It’s hard to guess Twitter’s business “objectives” since they self-avowedly have no business model. But we can safely assume that, like most start-up companies, the initial design reflects business objectives more than it does customer-centric design. Businesses almost always have to grow with their customers. Take Amazon, which invested millions to hire literary editors and library scientists to provide its customers with a well-organized browsing experience before they found that the search box was enough for 75% of its users.

Twitter’s design reflects several design hypotheses:

• Personal networks are most important: contact list import is sign-up default, suggested users is an after-thought
• Broadcasting comes first: the input box is at top
• Broadcasting is ambiguous by default: there are no categorization options, no “send to” option
• Short is best: 140 characters or less
• Messages should be personal: “what are you doing?” is the default prompt
• Timing is everything: messages are arranged chronologically with no other filtering options
• Searching is secondary to streaming

However, these have been rejected by most Twitter users in their interactions with the site:

• 60% of new users don’t have anyone to follow and they aren’t finding them
• Updates received by Twitter users out-number updates sent by a 100:1 ratio indicating following is more important.
• 140 characters is not enough. Most Tweets are multi-tweet thoughts. Few users actually send-and-receive on mobile phone SMS where the 140 character limit matters.
• Twitter users have introduced a whole language of hash-tags, re-tweets, @tweets, bit.ly links and so forth to add specificity and meta-data to messages.
• Anecdotally, most Twitter messages aren’t personal. This is consistent with Technorati’s report on blogging where 79% of bloggers blog to “speak their mind” versus only 32% who blog to keep friends & family updated.
• “Real-time search” seems to be the most frequently referenced application for Twitter at least by the TechCrunch and digital journalism crowds

So should Twitter alter its product design to address these issues? Ultimately, this is a matter of business objectives. We can infer a few objectives from their design:

• Twitter wants users to contribute content to the network. Thus search and following are de-emphasized.
• Twitter wants mystique, such that knowing how to use it identifies one as hip. e.g. easy-to-use sounds too much like MSN or AOL.
• Twitter wants to create an alternative language: short, dense, notable, brandable?

Are these the best objectives? I think the answer boils down to whether these can be linked to the most valuable business model. They have served Twitter well to build a network with appeal to a small percentage of well-networked tech saavy: the core of these seems to be about 2 million strong. But “messaging” isn’t inherently an application that wants to be limited to technorati so this is where I think Twitter has a fork in the road:

1. Twitter can seek to become a basic, open messaging protocol like SMS with some network services built it. This will imply a lowest-common denominator feature set and a quest for distribution embedding Twitter within more devices, web sites, software and services. This is the path which Twitter is going down now, but from the revenue perspective it seems weak.

   OR

2. Twitter can try to challenge blogging as another service which facilitates structured communication. In that case, it will need to attract large “follower” audiences which don’t necessarily write updates and focus on helping paying broadcasters to be discovered by search engines and share multimedia content, track readership, etc. Twitter can make money by charging commercial users for helping them to influence but the egalitarian, utopian nature of Twitter will be lost and the feature-set will need to become more responsive to its customer base.

Personally, I think great businesses always start with a vision that seems a little utopian and they try to nudge users in the direction of “what they should do” rather than “what they actually would like to do.” But any product needs to evolve its design over time to learn from customers actual behavior and reconcile the needs of different customer segments.

Complexity then is the natural result of popular interest and pleading. So as Twitter continues to gain popular interest, it will need to be more customer-centric to capture that audience. That  probably means it will need to spawn concepts like “channels” which organize content and an API which enables some users to experience “Twitter for technorati” and others to experience “Twitter for corporations” and “Twitter for the rest of us.”

Balancing the complexity introduced by actual users with a business model that serves a target market which is not too broad or too narrow is a matter of art informed by science. On the science side, there’s plenty that can be done to crunch usage data, develop perceptual maps, look over the shoulders in ethnographic research, perform conjoint studies (a simulated, more versatile and lower cost “bucket test”) and incrementally test the business models which product design supports.

But if I’m certain about one thing: it would be a shame if Twitter tries to stay “simple” forever because right now, it’s anything but.