Remembering GW Micro

Introduction

Over the past few weeks, those of us who follow such things have heard two major announcements from GW Micro, the first announcing that they had decided to start selling consulting services and the second announcing that they had been acquired by AI Squared, the Vermont based publishers of the market dominant low vision AT ZoomText.

Over the years, in this blog and on BlindConfidential, I had often mused about GW Micro and its Window-Eyes screen reader. After learning of their merger with AI, I found myself reminiscing about Window-Eyes, especially during the years when I worked at Henter-Joyce and Freedom Scientific.

Before I start, though, I want all readers to understand that this article is a personal essay written entirely from my own memory of events and based entirely in my own opinion of the history I relate herein. This article will not cite references as I’m not using any. I will try to tell the story as best as I can remember but readers should keep in mind that human memory is highly fallible so be assured that some facts and dates and such in this article are probably incorrect. Also, readers should remember that I’m an engineering manager by profession and a writer by avocation and not, therefore, a credible business analyst. The opinions regarding how companies behave and how businesses ebb and flow are given from my personal perspective. I have worked professionally in the software business since 1979 and my opinions are based in having worked in the industry for most of my life and are not definitive by any known definition of the word.

Thus, if I get some facts wildly wrong, I’m happy to (as always on this blog) correct them. Minor factual issues, though, will be ignored if they do not reflect on the story I’m telling.

Window-Eyes: My First “Real” Screen Reader

In 1997, I found myself in a tremendously unfamiliar situation, I had no job and my vision had, due to retinitis pigmentosa, deteriorated to a point in which I could no longer see well enough to use a computer. As I had been programming computers professionally since 1979 and as a hobbyist since 1971, I fell into an incredible state of despair. “Computing is all I know,” I thought, “My career is over.” Then, as I had done so many times before, I decided to solve my problem by making a program that would read aloud the contents of the screen. I had a Macintosh back then and using Apple’s development tools and a combination of C++ and AppleScript, I wrote myself a little utility that would take the selection from the screen, copy it to the clipboard and then push the contents of the clipboard to the old AppleTalk speech synthesizer. Thus, my first screen reader was a home-brewed hack I made for myself.

With my old Macintosh desktop loaded with my personal screen reader, I could navigate menus using CloseView, a really terrible screen magnifier that Apple once included with Macintosh products, at 16X magnification in reverse video. With that, I enrolled in a creative writing program at Harvard and never expected to work in software again. A few months later, though, my father told me had had a conversation with an old friend who, coincidentally, was also blind from RP who had told him of a program called Window-Eyes. Being a terrific dad, mine bought me a Gateway laptop and a copy of Window-Eyes.

When the packages arrived at our home, I had already done some research into Window-Eyes and was really looking forward to giving it a try. After my wife spent a few hours on the phone with GW Micro technical support getting it installed, I sat down and gave it a whirl. “Holy shit!” I exclaimed. “This is the coolest thing I’ve ever seen, er… heard!” I would use Window-Eyes for the rest of that year enjoying learning about writing from some really outstanding instructors.

The following summer, though, I started thinking more vocationally and wanted to get back into the work force. Clearly, with a screen reader, I could probably work making software again. I sent my resume to GW Micro and never heard a response. I sent my resume to a friend at Microsoft and he suggested I apply for a job at a Florida based company called henter-Joyce. I went to the HJ web site and saw that, indeed, they had an opening for an engineering manager, I sent them my resume, had a few phone calls with Ted, Glen and some others and the rest is history.

Switching To JAWS

During the year in which I used Window-Eyes, I really didn’t become anything one would describe as a “power user.” Instead, I did pretty well in Word, Internet Explorer 3.x and the long forgotten email client, Eudora. I studied creative writing so all I really needed was a word processor, a browser and a mail program. While I interviewed with HJ, though, Ted Henter felt it essential that I at least try JAWS and he sent me an evaluation copy while they decided whether or not to hire me. After getting JAWS installed on the same Gateway laptop, I started using it for more or less the same tasks. The one major difference I noticed was that, at that point in history, the Window-Eyes “MSAA Mode” for reading web content did a much nicer job than did the JAWS 3.20 “page reformatting” techniques. Otherwise, once I learned the JAWS keymaps, the switch was simple and, as HJ offered me the job leading the JAWS development team, I was sold on it.

The Collegial Competitors

Prior to joining HJ in 1998, my career in software had been entirely mainstream. I was, therefore, accustomed to the sort of competition one observed in the mainstream software industry back in those days. In access technology, though, I was introduced to a whole new way of viewing things. When, after we released JAWS 3.31, the first with the virtual buffer approach to the Internet in September 1999, I mentioned on a public mailing list discussing screen readers that “JAWS can load Jamal Mazrui’s www.empowermentzone.com site in under 30 seconds while it takes Window-Eyes more than 25 minutes to do the same,” the reaction I got puzzled me. Another HJ executive pulled me aside and told me, “the GW guys are upset about your email.”

“Sure,” I said. “our performance kicks their ass, I’d expect them to be upset.”

“It’s not that,” continued my colleague. “They’re upset that you criticized them so publicly.”

“How else would I criticize them, they’re the competition after all,” I added.

I couldn’t believe that the AT biz took such a “candy assed” approach to competition. In my mind, if we said something that was absolutely true, something anyone who owned the two screen readers could test side by side, that we should announce where we did better as loudly as possible. I also felt that, when we heard that Window-Eyes bested JAWS in some areas, that we would do our damnedest to ensure that our users would soon have something equal to or better in an upcoming JAWS release.

I thought this was how competition worked in the software industry: if a competitor beats you at something it’s incumbent upon you to catch up and try to beat them; if you do something better than a competitor, then you should expect them to retaliate with something cool in their next release as well. What I would learn about the AT biz, though, is that it follows a very different set of rules.

The Race To Windows NT

In the years prior to my joining HJ, they acquired a screen magnification program called MAGic from its Massachusetts based developer. This deal happened before I joined the team so I wasn’t present and have no details about the acquisition. MAGic would never become a market success, losing in every marketshare report I’ve ever seen regarding screen magnifiers to the AI^2 ZoomText which still maintains a monopoly like share of the market. What MAGic did, though, was give JAWS the video hooking technology necessary to get into the Windows NT market. Ted and the gang at HJ realized that the US federal government, primarily for security purposes, were upgrading all of its computers from DOS to Windows NT 4 and JAWS and MAGic became the only screen reader and magnifier combination that the feds could buy.

Having a Windows NT solution allowed HJ to sign a long term contract with the US Social Security Administration (SSA) which would be followed by large and long term contracts with the Department of Education and a number of other US federal government agencies. These big time deals followed by the implementation of Section 508 in the federal space gave HJ and later Freedom Scientific a massive infusion of cash dollars, something that neither AI^2 or GW could then boast.

When I joined HJ, the estimated marketshare figures for JAWS and Window-Eyes would have them tied with around 40% each with Dolphin and all others holding onto the remaining 20%. On screen magnifiers, MAGic was third behind both ZoomText and the Dolphin products.

Investment In JAWS

When HJ got the first contract with SSA, as I show above, JAWS and Window-Eyes were effectively tied in the marketshare competition. Ted henter could have chosen to take a whole lot of those dollars home as a windfall but, instead, chose to invest most of your tax dollars back into the business. Along with other executives, Jerry Bowman, Eric Damery and Glen Gordon, Ted worked to bring in professionals like Sharon Spenser and me to help build a management team who could successfully bring the business to the next level. From my first months on the job, I worked well on the team with Glen and Eric and, together, we set out to win the screen reader wars. We were done with “speak no evil” marketing, we dropped the gloves and got down to the business of building a market giant.

What none of us expected, though, was just how easy GW Micro would make our quest for the dominant position.

The GW Micro MSAA Fetish

As I wrote in my recent article on the importance of standards, MSAA was the first accessibility API available on a major operating system. Unfortunately, as I also describe in that article, the early versions of MSAA were not up to the task of providing a truly accessible experience in all but the most simple of applications. For that reason, we at HJ/FS chose to use different techniques to gather information which we would then provide to our users in speech and braille. Although, with JAWS 3.31, we demonstrated how the GW MSAA solution had profound performance problems and would, along with IBM’s Home Page Reader, over the following few years also demonstrate how we could, using other techniques, approach 100% of the web user agent guidelines, GW Micro steadfastly adhered to their MSAA only strategy, a strategy that would help spell their market demise. Year after year, review after review, blind users, government purchasing agents and everyone else who paid any attention would read how JAWS had gotten “even better” on the Internet while Window-Eyes stood more or less in place.

To a guy like me with a mainstream software background, this was incredibly confusing. Both FS and IBM demonstrated solid progress year in and year out and GW Micro, ostensibly the competition to JAWS did virtually nothing to catch up.

One might think that the income HJ/FS derived from its big time government contracts might have fueled the rapid improvements in JAWS but, while this is true for other areas, virtually all of the code written to support the Internet in JAWS 3.31 until at least JAWS 5 was all done by one guy, Glen Gordon, who was in the employ of HJ before the big contracts started coming in. If one programmer, granted one very smart programmer, at HJ/FS could do all of the work, certainly GW Micro could have afforded to do the same. Of course, that’s how a “normal” company in a “normal” industry would behave and, as I said at the top, the AT industry plays by a weird set of rules.

The Great GW Micro Contribution

Shortly after Congress passed Section 508 of the Rehabilitation Act and the Bush administration published its rules for implementing such, a whole lot of mainstream software companies realized that, in order to keep their big time government sales, they had to at least pretend to pay attention to accessibility. AT FS, we would get inquiries from multi-billion dollar companies like Adobe, MacroMedia, PeopleSoft, Oracle and others. These mainstream giants needed an accessibility solution. To work with these businesses I proposed and FS launched its software consulting team, today, roughly 15 years later, led by accessibility rockstar Matt Ater. We viewed the accessibility of mainstream software as the problem of its publishers and, while such work was important, we were certainly not going to work for software monsters like these without compensation. Typically, in one of these situations, the big time software company would approach us, we’d say that we’d be happy to work with them for our hourly rate ($125 per hour in those days) and, soon, we’d get a “thanks but no thanks” note from the prospective client and would learn that GW Micro was willing to take on the work at no charge.

All of that was a long time ago so I do not recall the order of the inquiries but, after the first one approached FS but chose to go with GW instead, we heard back from the mainstream company with a, “the federal government doesn’t want it if it doesn’t work with JAWS,” which, of course, we knew already and had already told the potential client in our original proposal. At this point, though, the mainstream vendor had an MSAA solution delivered to them at no cost by GW Micro so would only need to pay FS to do the JAWS side of the effort, saving the billion dollar corporation tens to hundreds of thousands of dollars. At FS, we would privately say “thank you” to GW for doing the work as, typically within a month of their doing the heavy lifting, JAWS would use the MSAA implementation they had done pro bono.

While this work was disastrous for the GW Micro business and, until today, I cannot understand why they would repeatedly make this same mistake, it served the community of access technology users very well. A whole lot of mainstream software became more accessible because GW Micro worked as volunteers on their MSAA solutions. This, in turn, helped set a precedent that, indeed, using documented accessibility API was a good strategy and, today, with excellent API on virtually all OS, the model GW professed is the industry standard.

We should all be grateful to GW Micro for doing this work as it established a solid foothold for a standards based approach to accessibility if, indeed, it hurt their business badly. It may have made sense if an independent advocacy group like NFB or ACB helped fund this kind of development work but for a small company like GW Micro, it seemed suicidal.

Microsoft Office

The team at Microsoft that wrote Office back then ignored MSAA almost entirely. JAWS users needed to use Office in their jobs and would need access to its more advanced features in order to get a promotion or do well in a university. At FS, we used a different standard to gain access to the information in these programs. Our decision was to use Visual Basic for Applications (VBA), a programming tool used by corporations and others who want to use Office features in their proprietary applications. By simply adding the ability to query applications via the VBA interface, the JAWS scripting language would have a well documented manner of providing our users with access to information otherwise impossible to a screen reader. This is one area where the greater HJ/FS income does come into play, once we added the VBA feature to the scripting language, we could hire a staff of full time JAWS scripters and let them invent the future.

GW Micro held to the position that a scripting language was too hard for the average user to learn, a straw man argument as few JAWS users would ever attempt to write a script as few ever have the need to do so. As we’d add dozens of new features to the JAWS Office support with each release, including advanced features in PowerPoint, MS Project and even charts and graphs in Excel, Window-Eyes seemed like it was simply waiting for the day that MS would improve MSAA enough to do all of the cool things we could with JAWS. Ultimately, GW Micro would add a VBA enabled scripting facility to Window-Eyes but it would come years after JAWS and NVDA had crushed them in the marketshare battles.

Why GW Micro refused to add the functionality that JAWS had for years in Office still confuses me today. Did they really think they were competing by simply doing what appeared to me to be as little as possible?

The No Cost Window-Eyes

Earlier this year, Microsoft announced that any user of a licensed version of MS Office 2010 or newer could download a copy of Window-Eyes at no charge. This announcement made quite a splash but I chose not to write about it as our friends at Serotek had written a good piece on the matter and I had nothing to add. Like many others who hadn’t given Window-Eyes a look in a long time, I downloaded a copy and gave it a try. Right off, I noticed that it didn’t work with my favorite applications as well as does NVDA and, rather than spending a lot more time digging into it, I left it installed on my Windows tablet but haven’t launched it in months.

My impressions, though, from talking to a bunch of other blind people who’ve also tried the no cost Window-Eyes is a solid sense of “who cares.” Most of my screen reader using friends gave it a whirl and, for the most part, returned to NVDA with a few still using JAWS.

Any regular reader of this blog would know that I’m something of a standards freak. A decade ago, Window-Eyes refused to compete in the standards battle. Today, compared to NVDA, a free screen reader written mostly by two very underpaid guys in Australia, (I cannot comment on the current state of JAWS as I don’t have a copy installed anywhere), Window-Eyes remains far behind. As far as I can tell, Window-Eyes ignores most of the Aria specification so would be useless even in a properly standard accessible web app. So, if you care about web accessibility through standards, best practices and objective measures, support NVDA as they’re doing it right and they need all of the help they can get.

Conclusions

To be honest, I can’t conclude anything about GW Micro other than they were the most puzzling competitor I’ve ever come up against. Their willingness to give away their consulting time for free and refusal to even try to match JAWS functionality in MS Office or on the Internet still feels like corporate suicide to me. That GW Micro lasted as long as it did as an independent endeavor also confuses me, who are their customers? Obviously, given the most recent WebAIM statistics that show Window-Eyes with a share below virtually all other screen readers, the answer is “very few people.”

I predict that when MS releases Windows 9, it will contain a new Narrator that will be competitive with Apple’s VoiceOver on OSX. At that point, the no cost Window-Eyes will be obviated and it will become a forgotten product overnight. People who need more performance and more features than the Narrator I imagine will continue to use either NVDA or, especially those who use applications that require a lot of customization, JAWS. My predictions are based on nothing more than an idea I’ve pulled out of my butt so, if it comes true, remember, you heard it hear first.