Does Microsoft Suffer from Premature Innovation?

From when I joined Microsoft in 1997 until I left in 2004, and even since then (but especially during the heady days of the DOJ lawsuit), accusations flew fast and furious accusing Microsoft of misusing the word “innovation”, that Microsoft couldn’t innovate, or didn’t innovate.

To a large part, I don’t agree with that. Sure, you can say that Windows Phone 7 was a reaction to the iPhone and Android, but it wasn’t a clone, and it attempts to offer a unique usability value proposition versus iOS or Android. Even Windows 8 and the WinRT framework, while in some ways surely inspired by the iPad, it does so in a unique way as well, and will likely be seen in a diverse breadth of devices, versus the iPad’s “one size device fits all” strategy.

However, the past is littered with what I have sometimes referred to as “premature innovation”. Technologies created by Microsoft (or sometimes acquired by Microsoft) way ahead of their time. So far ahead, in fact that to many people, they don’t make sense – and as a result, either die on the vine or fail to sustain themselves in the market after release.

Voice (1995-Present) – TTS and SAPI. Do those mean anything to you? They do to me. Text To Speech and the Microsoft Speech API. Before I joined Microsoft in 1997, I built a PC for the first time. It had a sound card, a horrible microphone, and some speakers I had kept from an old Acer. At Slate, I helped build a custom audio version of Slate, generated using text to speech. It was shelved after I left since it sounded so robotic. Microsoft has been beating the speech and text recognition drums for decades, only to see Siri, a quirky, accent-fussy smart agent (there’s a term you probably haven’t seen in a while) steal any and all thunder that Microsoft had even thought of years before, but not in a mobile context. Speech is complicated. Speech recognition (hearing) and text to speech (speaking) are also two separate problems. Microsoft has moved these so far forward since 1997, it’s truly amazing. But frankly we’re still at a point where text to speech still results in inhuman vocalization (“Robby the Robot”, as Kinsley called it at Slate), and speech recognition is most usable for short form commands (Siri, or Sync), not reciting out a novel. SAPI was also used by the Microsoft Agent technology derived from Microsoft Bob (another conversation for another time), and is still used by Speech Server. The emphasis of Microsoft’s speech technology development appears to be on the server side today, but SAPI still works on client operating systems.

Killer? Not dead.What has held it back? Processing power, and the fact that computers interacting with speech is just a complicated problem.

WebTV (1997-~2002) – In 1997, Microsoft acquired WebTV for $425M. The idea of the Web on television surely appealed to many, and when combined with a DVR, conceptually made sense. But the fact that Web content for WebTV was custom, the amount of the Web actually available, and then how usable it actually was, given the state of televisions in the late 1990’s, led to low adoption. Today, the product is effectively dead. Yet here we are, 15 years later, and Google is still trying to resuscitate the idea, with multimedia content and apps intertwined. If Microsoft had run a different course with WebTV, it could have possibly gone further and might be something today. What’s left of WebTV? Well, in many ways, WebTV was the progenitor of UltimateTV and Mediaroom (the IPTV infrastructure solution licensed from Microsoft for BT, DT, AT&T U-verse, and several other telephony-based broadcast solutions. Not dead, but not mainstream, and most people wouldn’t ever associate Mediaroom with WebTV.

Killer? TV resolution, need for custom content, lack of broadband, lack of ubiquitous streaming video standard.

Mira (2002-2003) – One of my favorites, because I held out so much hope for it. You don’t remember Mira? Mira, also called a “Smart Display” as Mira was a code-name, was an attempt to create a tablet-style computer that you could use anywhere, but that featured the power of your Windows desktop. Using wireless connectivity, you can undock your “Smart Display” and take it with you around the house, using Windows Remote Desktop from the version of Windows CE built into the device. In many ways, Mira was cursed to begin with. Due to the state of technologies at the time (wireless, LCD displays, memory, CPUs), it was far too expensive in 8″, 10″, or 15″ models. It also couldn’t serve as your primary monitor because it was both too small, and when you disconnected it, nobody else could use the PC, since there was no monitor. Even if you had another monitor, because Windows client operating systems have only ever allowed one interactive user at a time (a rule that was bent for Media Center Extenders, which enabled more than one interactive user at a time, but in a very fixed function user interface), it was still limited to a single user at a time. Mira shipped, and you can even find used models around (that still cost too much, frankly). You could build Smart Displays today and, if packaged well, might even sell. But I’m not sure users today would want tablets that require a tether to a Windows PC in their home, as well as the difference between useful tablets (7″-11″) and home displays (generally 19″+).

Killer? Cost of components, lack of ubiquitous wireless home networking, windows single interactive user limitation, and frankly, a confusing use case.

Tablet PCs (2000-Present) – Mira was in many ways an early, and yet late, foray into tablets for Microsoft. Tablet PC Edition was, of course, the big coming-out party for Microsoft. Like Mira, Tablet PC was significantly more than a traditional laptop, yet often required a compromise in display size or performance. It also required significant buy-in from the user in pen-based input. Ink searchability has gotten significantly better over time, but like voice, pen/digital ink still isn’t a way to write voluminous amounts of text. Tablet PC predated the iPad by almost a decade. Yet Tablet PC sold at a trickle, while the iPad has established a distinct market niche. Why did it fail? The exclusive bet on ink, while not evangelizing a framework for tablet-centric apps (the opposite tact of what we see happening with WinRT, where the framework, not the device, is the most important aspect), but again, cost was a huge factor. Consumers wouldn’t choose a Tablet PC over a traditional laptop, and while IT pros might, it was often a hard argument to make in procurement, “I need to pay more for a computer so I can use a pen”. I experienced that one myself. Pen-based Tablet PCs are still around today, bought by users specifically looking for them, and sold into verticals such as the medical industry or into other industries with specific point-of-service needs.

Killer? Not dead, but undergoing fundamental transition with Windows 8). Pen focus wasn’t on target. Pen input simply isn’t for every person, or for every task. Touch+pen may prove to be a strong selling point to Windows 8 tablets. What kept it on the sidelines for 10 years? High cost, pen focus, and confusing use case.

Reader (2000-2011) – The item that triggered this whole post was yesterday’s Apple announcement. Microsoft pre-dates Apple with eBooks. REALLY pre-dates it. I was lucky enough to work with the guys on the Reader team back in 2000, as we created a version of Slate for use with the Microsoft Reader software on Windows CE devices and Windows itself. Reader was eBooks before eBooks were cool.

Killer? You can pin the death on several things, but I believe fundamental problems were content acquisition and a lack of truly compelling form factors for eBooks. Windows itself makes for a reasonable reading format, but for the best display, ClearType in Windows relied upon LCD displays, which were not ubiquitous when it arrived – and reading a book at your desk isn’t always what you want to do. Conversely, Windows CE had clear, color screens, well ahead of Palm. But the devices were very expensive PDAs, in a niche of the market – and the small screens were hardly ideal for voluminous reading. I think the timing of Reader’s death was bad – as the lack of compelling reading content without relying on Amazon or other content partners won’t be a good selling feature for smaller Windows 8 tablets.

.NET My Services (2001-2002) – Known initially as HailStorm, .NET My Services faced the wrath of public opinion. Unlike Normandy, which can in some ways be viewed as it’s predecessor, .NET was more of an SDK and less of an infrastructure. It was also less rejected by ISPs or network providers due to the high cost (as Normandy was), but pundits fueled the fire on concerns about Microsoft being the holder of this much profile data for users across the Web, and resulted in it being shelved in 2002. A few months ago in our office, I ran across an old copy of the Microsoft Press title overviewing the feature specification for .NET My Services. Here’s a few of the Web services that .NET My Services would enable for a user with .NET Passport (stop me when it sounds familiar):
  • presence
  • notifications
  • calendar
  • contacts
  • inbox
  • application information
  • profile information
  • favorite websites
  • lists
  • wallet storage

Yes, that’s right. The 11 year old specification for .NET My Services is like the back-end API if you wanted to build your own version of Facebook. I’m not the only person to think that .NET My Services was ahead of it’s time.

Killer? Paranoia, development complexity (SOAP). Though the paranoia side may have not gone away, I have to wonder if Microsoft were to resuscitate .NET My Services all over again, with a REST-based framework instead of SOAP, if developers would be more interested. Problem is, it’s still a lot of work, and relies on the user base to fuel it. May be too late to try, since Facebook has such a hold on the “personal social networking” market today.

Spot watches (2003-2012) – Smart Personal Objects Technology (SPOT) was more than just a watch, but the watch was what most people knew of. Riding over FM radio frequencies, SPOT was an attempt to get weather, traffic, and other information to users in an era without ubiquitous wireless Internet connectivity. A $59/yr charge enabled watches, coffee makers, GPS units, and a few other devices to receive this information. Born in 2003, SPOT was quietly put down earlier this year.
Killer? Wireless Internet access and smart phones, as well as cost and limited market. The devices were expensive and large, and required the annual subscription just to get a very limited range of information.

There’s a whole other series of “premature innovation” I could speak to in MSN back in the last century and early in the last decade – but to keep your sanity, I’ll save that for now, and perhaps do another blog post in the future on that.

 Microsoft has truly innovated many times. Kinect is one clear example, but it’s hardly alone. Sometimes it has frustrated me to look back at the last 15 years and contemplate how different Microsoft innovations might have done if their timing had been different. If, well, Microsoft had waited until the customer scenario was clear and the economics for hardware partners were on target. Hopefully we’ll continue to see more innovations like Kinect, where the right time, right price, leads to a strong uptake and further innovation as a result.


  1. Doesn’t the Hailstorm/My Services featureset largely exist now in the form of Windows Live Connect, with the difficulty simply being strength of the competition?

  2. In some ways, yes. But it’s much narrower in scope than My Services ever was, and I don’t believe it is nearly as open. But at least it’s considerably easier for developers to use.

  3. Don’t forget, what started as SPOT watches has turned into the .NET Micro Framework (which still uses SPOT in some namespaces), and the .NET Gadgeteer. We have tons of OEM and hobby devices to thank for the innovation that came out of SPOT 🙂


  4. Faisal Ahmed Farooqui

    I would not agree with phrase “Premature Innovation”, but I’d say that MS couldn’t figure out that how to use its innovations. Use cases are important things. the problem with engineers is that they are not like others. they think that everyone is smart like them. But It’s not case. We need to think about the practical approach to the general public.

  5. […] “From when I joined Microsoft in 1997 until I left in 2004, and even since then (but especially during the heady days of the DOJ lawsuit), accusations flew fast and furious accusing Microsoft of misusing the word ‘innovation,’ that Microsoft couldn’t innovate, or didn’t innovate,” Wes Miller, now an analyst at Directions on Microsoft, wrote on his personal blog. […]

  6. A friend on Facebook reiterated the same. I did mention a few points in there about confusing use cases, and I think in many of these, that did play a role in addition to any cost or technological complications.

  7. […] not immediately chime with readers but bear with me. A few days ago I also read an interesting blog post by Wes Miller in which he explored the concept of ‘Premature Innovation’ in the context of […]