The New York Times has an interesting article this week explaining why HP's adventure with Palm failed. The latest explanation is that Web OS just wasn't ready for prime time, according to Paul Mercer, who was senior director of software at Palm (link).
Paul's an extremely bright software guy. It's unusual for someone with his seniority to go on the record with criticisms of his former product, and I applaud him for it because it helps us all learn. If Paul says Web OS was unready, I'm sure it was. But respectfully, I don't think that's why Web OS failed. I think the company's business strategy was fundamentally flawed, in ways that would have almost certainly doomed Web OS no matter how it was built.
The point is important because other companies planning similar products might take away the wrong lesson from Palm's demise. (For example, Information Week concludes that it's too hard for any startup to play in the mobile device market [link]; MIT Technology Review says the lesson is that you have to retain key employees [link].) To explain what the right lesson is, I need to give you a little background on the dynamics of creating a new operating system.
New operating systems always suck
Sorry for my language, but sometimes it's best to be blunt. An operating system is an incredibly complex piece of software, just about the most complex software you can write. In the first version of an OS, the list of features you want to add is always much longer than what you can implement, there are always bugs you can't find, and performance is always a problem. What's worse, there is a built-in tension between those three problems -- the more features you add, the more bugs you create. The more time you spend fixing bugs, the less time you have to improve performance. And so on. As a result, every new operating system, without exception, is an embarrassing set of compromises that frustrates its creators and does not deliver on the full promise of its vision.
Remember these beauties?
--The original Macintosh can't create a word processing document longer than 10 pages.
--The original version of Windows can't display overlapping windows.
--The original iPhone doesn't allow third-party native apps, and lacks 3G and MMS support.
The operating systems that succeed are the ones that survive long enough for their big flaws to be fixed. That happens if the OS's supporter has a deep, multi-version commitment to it (Windows) or if the OS does something else so compelling that customers are willing to buy it despite its flaws (graphics on the Mac). Your chances are best if you have both patience and differentiation.
Palm's problem: Lack of a compelling advantage
The Palm Pre and HP TouchPad had neither advantage. Palm was not rich enough and HP was not patient enough to keep investing after the first versions showed a lot of flaws. And more importantly, there was nothing compelling enough about either product to make people buy it despite those flaws.
Think about it, what was the one special thing Web OS devices could do that absolutely compelled you to go out and buy them? And don't say "multitasking;" I'm talking about a genuine, easily explained benefit that would appeal to normal people, not technophiles.
I wrote about this problem back in 2010 when the Palm put itself up for sale (link). To recap: you don't run TV ads featuring a Borg hive queen if you have something compelling to say about your product (link).
Contrast those ads to Apple's current iPhone ads in the US, which are basically a 30-second demo of Siri (link).
The original Palm OS succeeded because it made a great appliance for managing your calendar and address book. That jump-started the market, and all the additional stuff empowered by the OS came later.
iPhone succeeded, in my opinion, because it was the first device to make PC-style browsing work well on a smartphone. That killer feature bought Apple the time and market credibility it needed to enable native apps, fix the phone's problems, and add a raft of additional features that fleshed out the product vision.
Android succeeded (in part) because Apple stupidly left a void in the marketplace that Google could fill. In the wake of Steve Jobs' death, there has been a lot of well-deserved praise online for the brilliant decisions he made. But I think one of Steve's biggest mistakes ever was the decision to wed Apple exclusively to AT&T in the US for multiple years. That forced Verizon to find an iPhone competitor and market it aggressively. Verizon's choices were Windows Mobile (unpopular with customers, and a vendor with a history of shafting its partners), Nokia/Symbian (unpopular in the US, and a vendor with a history of shafting operators), or Google (sexy web brand, believed at the time to be open and non-controlling). People outside the US don't realize this, but in the US Verizon was the main marketing muscle behind the success of Android. It forced the product into the market and kept pushing for a long time, giving Google the time it needed to improve Android and get it past the crucial first release.
The Pre and TouchPad had no patient sugar daddy. And they had no breakthrough feature that would compel people to buy the first versions despite their inevitable flaws. I think Palm's product strategy was broken, and so Web OS was probably doomed no matter how well it was implemented.
The lesson: Who's your daddy, and what's your killer feature?
Two companies are working on new mobile platforms scheduled to ship in 2012: Nokia's next-generation Windows phones, and RIM's BlackBerry 10. In both cases, the press has been focusing on their development schedules. The schedules are very important, of course. But the real questions to ask are:
1. Do they have the financial backing to complete versions 2 and 3, which will be needed to fix the inevitable flaws in version 1? and
2. Will the products do anything unique and compelling that will cause at least some customers to prefer them even if they have other drawbacks?
I think Nokia can probably say yes to question 1; RIM is in doubt. And as far as I can tell, neither vendor has even started to address question 2. If they don't, in a year or two we'll probably be doing more post-mortems.