Most software is designed to satisfy the needs, as best it can, for 80% of its targeted users. If those users happen to be consumers, then in general that 80% figure will mostly include people with limited knowledge of how computers work.

You can see how this works today. Take a look at MSN Spaces itself. Specifically, take a look at the themes that are provided by default. You have the ability to select from a list of pre-designed themes. This list is large enough to satisfy most of the common wants. But then consider the true power user. I am not talking about the developer, who could create their own web site for posting photos and blog entries, customized to their hearts content (but may choose not to, because there are more efficient uses for their time). Rather, there is an audience of people who may not understand all of the finer details of CSS styles, javascript, and other web goodies, but who nonetheless are sophisticated enough to understand what the capabilities of the web are, and would like to do some customization. What is being done to satisfy those users? Unfortunately, they are left unserved. They can either do everything on their own, or they can accept a default.

I am reminded of this because of a recent experience with Hotmail. A user pointed out that they wanted to customize how thier name appeared in the From line of email that they sent. They couldn't figure out how to change this. What Hotmail created is a leaky abstraction. In reality, this is just an SMTP header that must abide by certain standards-based rules. The 80% user isn't interested in this detail of implementation, they would just like a name to appear so it looks like other email that they send and receive instead of just having an email address. Hotmail just pops their name (collected earlier) into this header, and most users are fine. If you want to customize the contents of this header, then you have to customize the name and understand the correlation. The 20% user will likely eventually find this. But this 20% user tried to eliminate this auto-filled name altogether, and it turns out that just removing the name from the Hotmail configuration does NOT remove the name. The business logic, apparently, is to fill the header from the Hotmail name, and if nothing is found there then just take the Passport name and use it.

Now, the 20% user would have no idea how this is implemented, and should not have to be concerned with how many and what sorts of data stores are sitting on a remote server somewhere. In fact, most would probably expect only one. So, the real task is to understand where the abstraction is leaky, and enable these 20% users to achieve what they want, without making them go all of the way to being a developer or else just settling for what the 80% want even when they are more sophisticated.

This, to me, is a huge challenge for UI design. Much energy is focussed on the 80%, for obvious reasons. If more resources are applied, they go here. But 20% of everyone is an extremely large number, and I think it's very important to accomodate this important group of people. Not only to satisfy a very large number of people, but also to challenge the abstractions that are created to hopefully make them even better in the long run.

It's hard work, and often neglected. I think that's unfortunate. The audience for software is not polar (consumer or developer). We should be seeking to serve all audiences, and not make the audience conform to our expectations even if that makes the job easier. For a v1 application, such as MSN Spaces, it makes sense that not everything has been developed yet. However, Hotmail has been around for a while. It should be eaiser for that 20% to achieve their goals.