A case study in improving software: What Office 2010 can learn from Notion 3

On the surface, this is a review of a music composition product entitled Notion 3, from Notion Software, priced at $249 suggested retail, born out of the original VirtuosoWorks product produced in 2005 by music professor Dr. Jack Jarrett, and which produces realistic orchestral sound from precisely notated sheet music on a standard Windows-based PC or Mac. But if you've never composed music before, and even if you don't plan on doing so in the future, I urge you to read on anyway, because this is about the business that we are all engaged in.

This is a story about a software company managing the process of starting all over again, putting its past behind it and crafting a completely new product. Notion 3 is billed as an upgrade (Notion 2 owners can purchase it for $8), but it is entirely new.

When Microsoft first announced the tremendous changes it had planned for Office 2007, including the premiere of the innovative ribbon, it was turning its back on over a decade of adherence to the spirit -- though typically not the letter -- of the Common User Access guidelines. This meant retraining millions of users on a completely new way of working -- a process that is still going on in many enterprises, even now. Microsoft knew that to embrace the future, it needed to change. But managing the transition, to this day, has been difficult.

Notion is not the dominant music composition product -- though market studies have yet to be made official, it's probably the #3 product in its category. The top two are bona fide professional music composition market leaders -- Sibelius and Finale -- trading off with one another like Coke and Pepsi. By contrast, the original Notion had been positioned as a beginner's product, a step up from something called "Music Printer Plus" that seemed about as likely to dethrone one or the other market leader as Print Shop Deluxe was to knocking off QuarkXPress.

What Notion had going for it was two things: a surprisingly agile sound reproduction system that reproduced a respectable sounding studio orchestra, and a front end that was not only easy to learn, but educational for those who wouldn't know a fermata from a marcato. But against these two giants -- the proton and neutron of the composer's world -- Notion 2's market penetration had pretty much run its course by the middle of this year.

Thus the bold decision to redo everything. Notion Music would replace the sound system and the user interface, and both would have to be worlds better.
"It was a hard decision, to tell you the truth, that we didn't take lightly," said Lubo Astinov, Notion Music's product manager, in an interview with Betanews. "We pondered on it for a few months before we finally decided to go ahead and switch. But basically, we decided in one swift move to go to a new code base...and take everything that we could from Notion 2 in features, rebuild them onto the new code base, and on top of them, build an abundance of new features. We tell people that Notion 3 is indeed a new product; it's not an update, it's a new piece of software."

With Notion 3 only having been released late in September, the jury is still out as to whether the product has even begun to achieve the market penetration it needed. So it's premature to talk about the Notion 3 "success story." But just as the recent evolution of Microsoft Office has been about pulling off a successful remodeling job while keeping all its users comfortable -- a process which isn't over yet -- the Notion 3 story is about rethinking everything, and challenging old "notions" about how a very complex piece of software should work.

The people leading the discussion for Notion were musicians first who became developers later. Astinov is a studio guitarist skilled in many styles, including jazz and flamenco, though he's also composed production music for PBS programs. He was a graduate of the Berklee College of Music where Dr. Jarrett taught.

"Mostly the discussion was driven by, 'Where do we need to be?'" Astinov told us. "We had to lay out a plan of not only, 'How do we get there?' but, 'In what time can we get there?' We realized that those were the goals that we set two years ago...that we needed to embrace a code base that is more flexible, easily updatable, and easily manageable."

Lesson 1: Rebuild the code base to be taken apart

Years ago, a point release for software suggested a completely replaced code base. But as the "engines" of applications became precious intellectual property for their publishers, point releases eventually became new life support systems for old code. Sure, there were new feature sets, and sometimes a revised and updated look and feel. But under the hood, you'd often find a living, breathing, single-threaded creature of the pre-OO era, not so much a component as a mashup.

"We needed to have a more general way of controlling elements, a more...well, [Notion 2] was object-oriented before, but [we needed to] really utilize the power of generalized object-oriented languages to their extreme, making sure everything is nicely structured, utilizing best software development practices, to create an architecture that is easily updatable, easily changeable."

Sometimes, the problem with the "engine" mentality in a development shop is that it creates an artificial barrier between what can be changed and what cannot be touched, lest the ghost of past developers haunt their sleep. In the new reality of software, even past the RTM date, problems and bugs crop up that deserve to be addressed immediately. If those problems exist within an untouchable zone of software sanctity, customers come to view them as defects endemic to the code base rather than as unforeseen circumstances curable with minor surgery.

"Although we had a brilliant team working on Notion 2, and it achieved what it set out to achieve, developing a software product is always a learning in progress," remarked Astinov. "You learn as you do it, and a couple of months later, you come back and you say, 'You know, I wish we coded that differently.' Then little by little, we came to the realization that, it would be harder to update the Notion 2 code base, fix all the problems -- whether they're design problems or coding problems in the software -- and get to where we want to be at the same time with allowing us to quickly enhance the software for future features."

Next: Building the user experience down...