What I think of as the Microsoft approach to product development is to cram as many features as possible into each product. Interface is an afterthought: All those features exist primarily for the benefit of the feature list, not the user, so who cares if the feature is buried three menu levels deep? If users complain enough we’ll just add another cryptic icon to a button bar. No room on the button bar? We’ll add another button bar, piling up shortcuts to the most popular features until the interface threatens to overwhelm the app. Maybe users don’t want so many buttons? Give them the option to remove whatever buttons they want. (Another feature!) Hey, we’re giving you features! Why are you complaining? All you have to do is go to Tools > Options > Customize and you can make the interface look just the way you want!
Somewhere along the line this became the default mode for software development. Open source development works exactly the same way. Somebody wants a feature? Great, let’s just add it. Stick it on to a menu (or its equivalent in web applications, a console page) and move on to the next item on the checklist.
I see this in a microcosm surveying the world of mailing list managers (for use on my haiku site, of course). Mailman, PHPlist, Majordomo, Dada Mail aka Mojo Mail … they’re all powerful, capable programs in their way. And their all fantastically complicated to use, because they have so many options. The thought given to interface is minimal, because the focus is on features.
What I want is something like the iPod Shuffle of mailing list managers. I want a simple signup process, preferably with a confirm step thats compatible with SMS. I want to be able to send list members just the content I specify, nothing more or less. And I want an easy console that doesn’t take a week to master. Is that so much to ask?
Hypothesis: In web software as in appliances, the autocratic, top-down approach produces much more usable products than a bottom-up, consensus-based approach. Somebody’s got to get in there and start saying “No” to the engineers. It takes a Jobs to make an iPod. It takes a Blake Ross and Dave Hyatt to make a Firefox.
I think you’re being too hard on MS. Yes, feature-itis is a real issue, but you asked for a mailing list manager with SMS support. That’s one specific feature. I want a mailing list manager that has good reporting (who opened, who bounced). And we are on the road to features. Yikes.
I think the best product design has good features that do not overwhelm the user. Firefox is actually a great example of this. There are TONS of features in Firefox and with the add-ins, the number skyrockets even more. So for a power user like me, I am happy. For the casual users, they are happy.
You Mon, I am no doubt being too hard on MS to make a rhetorical point. But the point is still valid: Throw in too many features and things get confusing. Firefox is a good example: It’s relatively simple, but it’s extensible — which means, if you want something, you go out and plug it in. This is a much more intelligent approach than cramming everything in from the get-go.
Also, not to pick nits, but SMS support is not really a feature of the mailing list manager. All the list manager needs to do is *not* add a lot of crap onto the beginning or end of the message, and it can send out SMS messages via email the same as any other message. So really all I need is a mailing list manager that doesn’t gunk up my messages.
Most arguments in the open source development world are solved in the same way: “Make it an option!” So screenshots of Open Office’s deep menus look similar. Simple truth is that everyone wants something different from the same piece of software. I’m not sure this one is ultimately solvable, though Apple software does a damn good job of stripping complex interfaces to the minimum.
Well, I’d love to know what’s fantastically complicated about Dada Mail – the second to last release introduced no new features, but 50+ bug fixes. The last update introduced one new feature and the throwing out of about 5+ other features it replaced.
The install process to get it up and running has been the same as it has been in 5 years. Sending a message requires you to fill out a Subject and a Message and click a green button.
But honestly, if you don’t give back feedback to a project like Dada Mail, the developers have nothing to go on when making decisions. Autocratic is fine, but the king has to listen to its people, for its the people’s country at the end. Sometimes, the king gets tired of wearing all these different crowns: researcher, developer, tester, designer, sales-person, QA, web designer, etc, etc, etc. A little help goes a long way, as long as the help is a positive crit. Being all these things as a sole developer, I can tell you that I’m honestly dry-mouthed thirsty for this type of thing.
Scot, you may be right that the problem is intractable.
Justin, I was probably harsh on Dada mail. I haven’t used it in a couple of years. Dada is actually closer to the autocratic model, isn’t it, since you’re the primary if not sole developer?
I’d be happy to give some constructive, positive feedback to you! I’ll install the latest version and test it out.
and it goes all the way around…
http://tech.cybernetnews.com/2006/12/06/cybernotes-200-firefox-extensions-installed-at-one-time/
My goodness — 200 extensions! I guess you can’t stop some people.
At least Firefox doesn’t ship with all 200 extensions installed and enabled, eh?
Worth reading:
http://joelonsoftware.com/items/2006/12/09.html
You Mon, that was an excellent read. Sums up the paradox pretty well. Simplicity is only an apparent goal, but actual simplicity is too easy, too copy-able, and too limiting. Wide functionality is the real desire; making complex software simple to use is the real act of genius, and where the really hard development work goes.
You Mon, thanks for that link. I agree with Scot that it is a great read. My complaint is that Joel somehow manages to blame *journalists* for the problem of software being too complicated. In fact I know the scenario he describes intimately — in the early 1990s the editors I worked with at PC/Computing were always hammering on Microsoft and every other word processor vendor about their word-count feature. Of course that is one of the features that only journalists care about. But to generalize from that is risky.
In reality, journalists are no better than the programmers at figuring out what a good interface is. I think it\’s a cop-out to say that \”simplicity\” or \”a clean interface\” is a \”feature\” because the difficulty of creating a clean interface (one which perhaps conceals a wealth of features) is so great. It\’s not just a matter of adding another feature to the checklist: \”Simple interface? Great — got that!\” You have to devote hard thought to figuring it out. User testing is probably also mandatory.
In fact, interface design is a completely different discipline than programming, and I think it\’s programmers who think they can do interface design who are probably responsible for the most horrific examples of interfaces. But it\’s also the case that too many features can make creating a clean, simple interface *impossible.* Just look at Yahoo.