转自：http://blogs.msdn.com/chris_pratley/archive/2004/04/27/120944.aspx
So far this blog has mainly been about OneNote and general product design. I started that way because I love OneNote, and with its hard core fan base and relatively newness, I thought it would be a fun thing to blog about (and I will continue to do that). I've been staying away from the "elephant in the room" though. I also manage the program management teams for two other products you would recognize: Word and Publisher. Publisher I picked up last fall, so I am not terribly familiar with the details of its history, but Word I have worked on for 9 years.

I've been a little gun-shy of blogging about Word for fear of being inundated by what are as far as I can tell a gang of "net thugs" who roam the net making outrageous claims about Microsoft and its behavior, motives, etc in every public forum they find (none of which information they are privy to, little of which they have evidence for, and basically all of which I find personally offensive, not to mention incorrect - since they often are implicitly about me and therefore I for one know them to be incorrect). But enough about that - let's just dive in and see what happens. Hopefully the net-dwelling paranoid delusional conspiracy theorists won’t descend upon me… :-) I should note that anything I write below (or in my blog in general) is my own opinion and memory, and is in no way official Microsoft anything.



I started at Microsoft in June of 1994 on the Excel team, where I worked on Japanese, Korean, and Chinese Excel. I was a total Mac-bigot at the time. I thought Microsoft was, if not an evil empire, at least a maker of substandard products that didn't deserve its success. The elegance of the Mac appealed to my design sensibilities - I took joy from its apparent "perfection". I had been living in Japan, and was looking for a job either in France (I also speak French, being from Montréal, Canada), or on the West coast of North America: San Francisco, Vancouver B.C. or Seattle, which was just becoming known as an excellent place to live. I had a bunch of friends at Microsoft from my alma mater (University of Waterloo, in Waterloo, Ontario, Canada), and on hearing I was looking for a job, they hooked me up with the Excel team, which needed a Japanese-aware person in the Redmond office. The job I was offered had everything I wanted (Japanese content, customer-focus, design, technical content, good employee benefits, location, etc), except it was for the wrong company. I wanted to work at Apple - but they turned me down - quite rudely I felt given I was such a fan.



I worked a little on feature design in Asian versions of Excel but mostly on the process of actually getting the product "out the door" so to speak. This process is so unbelievably complex and difficult I think you would all be stunned (by boredom perhaps) to hear the details, but that's a discussion for another time. It has also changed a lot in the intervening decade to be unrecognizable except in its complexity - which has actually increased. So I'm no longer an expert.



For the curious, here is what "drinking the kool-aid" is really like: After about a year at Microsoft, I got tired of these endless Type 11 errors every 20min running Netscape on my Mac at home. These required me to restart the machine each time - losing everything I had been doing. The flawless and imperturbable performance of Windows NT at work won me over on purely pragmatic grounds - so I finally got a PC at home. After a year of distrusting the company somewhat, I began to gain an appreciation of how Microsoft worked, and to see it for what it was - a machine that was focused on building products that people wanted, as quickly and as well as they could. Note the "quickly" - this was what distinguished MS from Apple in the end - a focus on moving quickly, and beating the competition. Details like great design were not critical to most customers, so that didn’t really make it into the products, except where it mattered to the customer. It's hard to fault this logic really - it is pure efficiency from a business perspective, and in the mid-90's, Microsoft could do nothing wrong - it was the business world's darling. (see an earlier post about methods of development)



I joined Word in April 1995. Again I was focused on the Japanese product primarily, with Chinese and Korean as part of my job as well. Did I mention I speak and read Japanese fairly well, and my (now) wife Seiko is Japanese?



At the time that I joined, Word was hurting in Japan. We had a stagnant (about 5-10%) market share, while the US was well over 50% and climbing. My job was to focus a subset of the Word team on the Japanese market and try to fix the situation.



Before I tell you about that though, you need some context. So here's what was going on in the US. In the US in 1995, Word was in the process of winning its battle with the previous leader in word processing: WordPerfect (WP). WP had had a tough time making the transition to Windows 3.0/3.1. To understand that, let's go back even farther. If you remember back in the early 80s, there had been another dominator of the word processing world: WordStar. This was a DOS-based word processor, and everyone knew its arcane set of keystrokes, since it was "the" word processor you had to be familiar with - mainly because WordStar was the first one to emerge from the early years of DOS word processing as a serious company when it was anyone's guess what was going to happen. Interestingly for those who think Microsoft always tilts the tables, PC-Word for DOS (the MS product) never really went anywhere, despite lots of trying for almost 10 years, and for the conspiracy theorists: the same company was making the OS - not that it mattered.



WordStar made a few releases, each time preserving their set of keystrokes and operation, as well as being able to work with old files. Then they made a huge mistake. They created an app called WordStar 2000 (WS2000). This was completely different in its interface, and in its file format (backwards and forwards). It was essentially an entirely new and different application, designed from the ground up. I would love to hear from someone who worked on that version of WordStar about what the thinking was behind that release. Maybe they thought they had the market so sewn up that they could make what they probably felt was a radically better product by doing a rewrite and not lose customers. But what happened instead was that they leveled the playing field.



Since the new product was no more similar to the old one that any of its competitors, what customers did (other than yell and scream) was evaluate all the options. And WordPerfect (for DOS) won out. It was winning reviews in magazines that year. (remember this was back when PC-focused magazines existed in large numbers, and actually reviewed products and compared them). The thing about these comparative reviews is that they tried to evaluate the products each release as if you were a Martian - that is "objectively", as if you had no previous experience in any tool - which of course did not reflect reality. I also put "objectively" in quotes because, well, these reviews were really totally subjective and reflected the bias of the reviewers pretty strongly. They rarely connected with real customers to see what mattered - instead they prioritized what they thought was important (why on earth was "word count" such a big deal? It is rarely used among the real user base outside of students and professional writers - we have quantitative proof of that. But it turns out reviewers need it all the time, so it became one of the "critical features" of a word processor according to these reviewers).



Anyway, WordPerfect had a great release of its product in 1985, and people started to switch, since WS2000 had no user base familiar with it, and WP had at least some, and it was a better product according to most reviewers. BTW, MS-Word for DOS (PC-Word) was still an also-ran at this point, although Microsoft's MacWord was doing very well. So WP took on the new mantle of "must-know" word processor. If you applied for a job as an admin assistant (those who did a lot of the typing back then), this was the product you had to know.



In case you're too young to remember, Windows development started back in 1983, and it was a joke in the industry. Windows 1.0 (released in 1984 I think) was sort of a demo. Windows 2.0 (1987 or so) was much better, but it was limited in memory (286 processor had a max of 1MB addressable RAM), and ran too slowly for practical usage. It is also hard to believe now, but all the pundits in the industry thought GUI interfaces with windows and dialog boxes and menus and mice (the Mac, Windows 2.0, etc.) were for novices and were basically toys, since they lacked the power of a command line interface. Lotus 1-2-3 and WordPerfect ruled the desktop, with arcane command sequences that a professional user could work magic with, but which new users found impenetrable. Especially interesting was the discussion that came up around the impending release of Windows 3.0 around 1990. In 1989, all the editorials talked about whether application makers should bother with a Windows-version of their DOS apps. WordPerfect was pretty clear - they saw Microsoft as a competitor, Windows as a lame horse, and they felt pretty strongly that they would best serve their customers by sticking with DOS. Their customers knew the WP-DOS interface, it was faster and more professional than the goofy toy-like Windows interface. It became a point of pride that WP would not do a Windows version.



PC-Word, on the other hand, tired of losing reviews and not being able to shake the stranglehold that WP had on the DOS word processor market, had nothing to lose by making a Windows version. Fortunately, that also coincided with the direction that Microsoft was taking: bet the company on Windows. In retrospect, this seems like a no-brainer, but remember that at the time Windows was still considered a joke. Betting the company on it was a big, big bet.



Windows 3.0 came out and it wasn't a toy. It wasn't great, but it actually worked well enough that people found they could be productive using it. Windows 3.1 (and then Windows for Workgroups 3.11) came out, fixed up a lot of rough edges, and made a workable GUI system that ran on top of DOS. (BTW, one of the two guys who figured out how to get Windows to run in virtual 386 memory to break the 1MB barrier works down the hall from me)



Word and Excel (which until then had been called Multiplan on DOS, and "Excel" only on the Mac) had been rewritten from scratch as all-out Windows applications. And people liked them. Word 1.0/1.1/1.2 actually won some reviews against DOS WordPerfect, especially in things like ease of use and WYSIWYG editing. The Word team knew they had something, and put a laser focus on WordPerfect customers, asking them what they hated about WordPerfect, and making it a product goal for Word 2.0 and later to deliver features that made the most annoying things in WP trivial in Word.



Other moves were tactical. The Word planning team discovered that the WordPerfect sales force was going around to customers and showing Word opening a complex WordPerfect file (printer.tst) to show how bad the conversion was, and therefore how pointless it would be to try to switch to Word. So the Word team organized a special dev team that focused entirely on WordPerfect document import, "reverse-engineering" the WordPerfect file format (documentation for which was jealously guarded, as was the norm back then). Their goal was to make any WordPerfect doc open flawlessly in Word, but in particular their goal was to have no errors at all on printer.tst. Later the Word sales force used that same file when talking to customers as proof that Word 6.0 could open WordPerfect files flawlessly.



For the release after Word 2.0, the team merged with the MacWord team (then on release 5.1), and built a shared product called Word 6.0 (released in late 1993). That’s why on Windows the Word version numbering seemed to jump from 2 to 6 - because the Mac was already on 5.x. WinWord 6.0 was a monumental release in that it focused on ease of use, power and performance and really delivered a quantum leap over 2.0. It was devastating for WordPerfect, which until about 1992 had publicly said that they would not bother with a Windows release. Eventually it became obvious that Windows 3.0/3.1 was no longer a toy, and people were expecting GUI apps because they were so much easier to learn than DOS apps. But the WP designers were stuck - they owned the market, and had fiercely loyal customers. These customers had told them: "do not move to Windows and destroy our beloved DOS app with its arcane but powerful key commands". So the WP team compromised, and produced WP 5.1 for Windows, which was sort of a DOS app that ran on Windows. The idea seems to have been to allow DOS users to move to Windows and keep the UI they knew. The critics trashed it, saying that WP "didn't get" the GUI world. WP5.1 was unstable and was quickly replaced with WP 5.2, which was better but still the same approach.



In the period 1992-1994, Word wiped the floor with WordPerfect in reviews, winning just about all of them. Then WP 6.0 came out, which was a rewrite of WordPerfect to make it a real Windows app. So at last WP was at least a reasonable Windows application, but it had missed its "window" of opportunity (excuse the pun :-)). Word 6.0 (the third version on Windows, which had had a chance to respond to Windows users' feedback) still beat it in reviews, and in any case by then the momentum had shifted, and more importantly, the market was doubling in size every year (just about) thanks in part to the ease of use that Windows brought to what had previously been a DOS-only world for PC users. All those new users asked their friends or read the reviews to find out which word processor to use on their new Windows machines, and so Word was overwhelming selected by these new computer users who had only ever used Windows. WordPerfect began a slow decline that is still going on (they kept the loyal users, who numbered in the few millions, but almost all the rest of the market went to Word - now in the hundreds of millions). As an anecdote, the move to combine MacWord and WinWord was done to conserve development resources, but the result was that MacWord 6 was not nearly as great a product for the Mac as Word 6.0 was for Windows. That was a big mistake and got rectified later by creating a special Macintosh Business Unit to focus on the Mac business - so yes, even Microsoft makes mistakes. You can read more about MacWord on Rick Schaut's blog.



So let's get back to where I come in. In Japan, where computers had a little different history than they did in the US, word processing was mainly the realm of purpose-built devices called "Wa-Puro" (for "wa-do purosessa-", the Japanese pronunciation of word processor). PCs had only a fraction of the market, and on DOS, a product called "IchiTaro" was king. The Windows version of Word in Japan was basically a port of the English version, and unlike Excel which competed against another US import (Lotus 1-2-3), IchiTaro was a home-grown Japanese product. It sort of defined what "word processing" meant in Japan, and it was pretty different from Word - so Word was only popular with a few people who had to make a lot of English documents.



To give you an idea of how the Word team was successful at what it did in general, I'll give you a rundown of what we did to "win" in Japan. We had a team there already, but they were mainly a dev team working on porting the English product. We sent planners (and myself) to Japan to visit a lot of customers to find out what they hated about Word. It turned out that they hated Word for 5 major reasons - not because it was a bad product, but some common tasks that they did every day in IchiTaro could not be done in Word. We collected hundreds of sample documents and interviewed many users. We also set up a temporary "usability lab" in our Tokyo office and did side by side tests of Word and IchiTaro to see where we were going wrong. We used typical sample documents we had collected and asked users to create them in each application. What we found was that many of the documents simply could not be created in Word, and those that could took on average 5 times longer than in Ichitaro, even accounting for familiarity with the products.



So, we developed a prioritized list of things we had to fix in Word. Word 6.0 for Japanese was already in the bag, so our main focus was on Word95 (Word 7). We decided to work on the biggest problem, which was that Japanese documents used a lot of really complex tables - in effect their documents were more like forms than memos. So we built the Table Drawing tool (you can see this in Word today in all languages). We did a few other things that Japanese users expected, and released the product.



From a marketing perspective, we knew it was critical to "sim-ship" with (release on the same day as) Windows95, since that was a big deal worldwide and unlike Office, Windows had a huge marketing budget we could draft off of. We made our goal, and having Word95 as the only 32-bit application in the Japanese market just as Win95 hit really helped us too. Now Just Systems, makers of Ichitaro, also knew that Win95 was a big deal (they knew the WP story), and they also tried to hit the same date (which was widely known for more than a year), but they couldn’t quite get it together, and shipped several months later. For those not familiar with Japanese Windows at the time, Win95 was an even bigger deal in Japan than in the US, since Win3.0/3.1 for Japan was a pretty weak product, and the market there really needed a big advance like what Win95 offered to get it expanding as Win3.1 had done in the US. So Win95 was for Japanese Word what Win3.0/3.1 was for English Word. We hit 40% market share of new sales in the year after launch of Word95 for Japan.



We plowed on with Word97, adding features that Japanese users (and reviewers - not always the same!) expected, such as support for vertical writing, better Japanese input, etc. We simply went down the list of features that people could either tell us they needed, or that we deduced were necessary from our customer research. Word97 added another 20% to our share, and we even did a special Word98 for Japan, with only a couple of new features but a radically improved Japanese input method, and our share went even higher. About this time, Ichitaro underwent a total rewrite. For some reason Just Systems decided they needed to redo their whole application as "component software" (a fad at the time). You may remember this idea that you would be able to "buy" individual features of a product and plug them together to get the set you wanted. Hmm… Anyway, the rewrite took too long, and in desperation JustSystems put out a buggy, slow (over 30sec to boot - and Word was 8sec on the same machine) Ichitaro that actually had less features than the product it replaced. It also required more memory than any machines in the market had at that time (32MB, vs. the 8-16MB that Word required), so to use it you also had to buy extra memory. And their big tagline was "Now, componentized!". They were clearly pretty out of touch with their customers. Meanwhile, we kept going with Word2000, which really polished off the remaining things that people needed in Japan. I went to Japan every six months to meet with customers and understand their concerns. By the time we were working on Office XP (about the year 2000), the customers in Japan had largely dropped their resistance to Word. As they told me - "We don’t see anything wrong with it. It used to drive us nuts but it's pretty good now."



So, that in a nutshell is the Microsoft method. Understand the market, and the customers, and then go pedal to the metal, with release after release focused on what the customers need, incorporating their feedback. That puts the competition into reaction mode. And of course it helps if they also make a strategic error because they are under so much pressure.



After Word97 shipped I became a lead program manager in Word, so in addition to my Asian version responsibilities I took on all the international work, as well as "basic use" in Word. For 2000, my main goal in addition to the Japanese work was to solve a problem our best customers had but could rarely articulate. The whole world seemed to assume that if you had offices around the world, you had to deploy specialized local versions of software in each office, and of course that was a hassle for large companies, since they had to test all their add-ins, develop special set-up scripts, and try to understand the product differences for support reasons, not to mention the huge number of patches that had to be uniquely made and applied to each localized version (and there were over 30 of these localized versions, each slightly different). I heard this feedback in the form of all the questions I kept getting about international this and that. So my personal mission for Office2000 was to make all of Office into a single-binary, worldwide enabled, Unicode-based set of applications, with the parts that had to be different (the UI and help) being swappable components. I learned a lot about taking initiative during this time, since I just went ahead and told people they had to do this, and no one ever said "no" (well, some did, but I got everyone else to say yes first, so they couldn't really argue). The MUI ("multilingual user interface" as it came to be known) of Office and later Windows was a big hit, and is now the most common flavor of Office deployed internationally.



After Word2000 (really Office2000 for me), I worked on Word2002 (OfficeXP), as a lead for awhile before becoming the group program manager (GPM) near the end of the project. The GPM is the person in charge of all the application designers and spec writers. Word2003 was my first version where I was "running the show" so to speak. The 2003 XML work, research tools, collab tools, and so on were all done under my watch. I also oversaw the creation of the OneNote team at this time, as I have written about earlier.



I've shared all this with you to try to give you a sense of how we (I) see the world, and how we work on products. So, let's talk. Reasonable comments will be replied to.

posted on Tuesday, April 27, 2004 3:36 AM