webOS on The Verge
I'm not sure I'll ever be ready to tell everything about my time at Palm. Certainly not now. Perhaps in a novel or an 8bit video game, one day. I don't know. I really enjoyed my time there and made wonderful friends. It was also two straight years of frustration. For now I suggest you read The Verge's excellent in-depth article on the 31 months from Palm's 2009 CES debut to the end of the platform (and possible rebirth?). After you read it come back to compare notes. Below are some inaccuracies or clarifications based on my own recollection of events. Think of it as directors commentary if the director was forced to sit in the back and watch through a 3 inch screen.
A note to the reader. Before you continue there are three things you must keep in mind:
- I am now an employee at Nokia. The comments below reflect entirely my own opinion, not of Nokia or my team. Also note that I am not working in the Windows Phone division (or any product division for that matter) and have no insight into product plans.
- The comments below reflect entirely my own opinion. This is based on my recollection of events and personal speculation. Others have different recollections. These are mine.
- I wrote the following comments while reading The Verge's article. The thoughts are spontaneous, passionate, off the cuff, and unedited. Please adjust your expectations accordingly.
Prima (2008). I never got a chance to see Prima. I wish I did. Sounds ugly. I was able to get my hands on a Foleo, though. Insert another epic yarn about a brilliant but doomed product. I do know that Prima was Java based, and closer to JavaME than JavaSE. It had major performance issues because it was built on some wonky embedded JVM they got from a small company (I don't know who) rather than a proper JITed VM like HotSpot and JavaSE Embedded (though that wasn't a shipping product from Sun at the time these decisions were being made).
Prima's JVM at CES 2009. Yes, it was still there running some system services. Those lasted until around the time of WebOS 2.0, towards the end of 2010 as I recall. There were always references to it in the open source code dumps which is why people kept asking me if webOS supported Java. It never did in any way that a developer could actually use. That JVM also sucked up way too much memory, making a lot of system services dog slow.
CES 2009 This is the event which sold me on Palm. When I met Ben and Dion in Sweden to discuss working for Palm later that year this was the event I watched and analyzed to see if I truly wanted to work for Palm. The polish and practice was really amazing. If you haven't seen it you should watch it. It even made up for the truly horrible TV ads Palm later aired.
The de-Mercerization to create Blowfish: I don't know the details of exactly how Blowfish was built. Engineering was very closed off to us (Developer Relations). It was like pulling teeth to get information about them. I learned a whole lot by searching JIRA bugs and watching commit logs, however. Blowfish turned into a classic Fred Brooks "Second System Effect" death march. The release stretched on and on. Things which worked previously were broken and never fixed. Testing started to fall by the way side. Quality declined and the release kept slipping. In the end we missed an entire product cycle. This probably doomed us more than anything else.
2010 This when I joined. The Droid campaign had already screwed Palm (or Palm let itself be screwed by placing so much fait in a a carrier). Verizon promised to support the Pre Plus but it turns out that support didn't extend to the actual employees in the retail store. In the US, at least, a phone lives or dies by the retail staff in the carrier stores. Nothing else matters. Not price. Not features. Not apps. If the retail staff doesn't like you... you die.
Anyway. I'd been at Palm for only a few months when the HP deal was announced. I still feel it was the best decision. The smart phone game of the early 2000s changed forever after the iPhone. It was now a rich man's game. Palm was against companies literally 10 times it's size. To create a viable smart phone platform today you have to be prepared to spend about a billion dollars a year. Palm was simply too small. HP had the size and the cash, and a desire to not just be a Windows OEM forever. It was by far the best option. (All other options would simply have liquidated us). But the best laid plans of mice and men…
Leo: Well…. We will never really know what happened inside Mr. Apotheker's mind when he decided to cancel all webOS hardware. Did he not have the stomach for a billion dollar a year run rate? Did he never really want to be in the consumer business? Was his talk of "doubling down" a lie all along? We will never know. It was clear then and now that he never believed in the webOS vision. Perhaps if Hurd had not left history would have gone differently. At the very least the TouchPad would have had the chance to develop into the strong product it could have been. We were, for a brief time, the #2 tablet in the US.
Greg Simon: It really hit me hard when Greg Simon left. He was the only one who was getting my graphics bugs fixed! We had no hard specs for what our Canvas and CSS graphics output should be so I made it my personal mission to write tests for every missing feature of the spec and then push to have them implemented. For a long time Greg was the only guy who could help me. He's the one who implemented CSS gradients for me.
TouchPad hardware: One thing I don't see mentioned is that the hardware design for the touchpad came from HP. It was designed before the acquisition and originally ran Android. That's why you'll occasionally hear rumors of a touchpad the shipped with Android on it. That's also why the specs were a bit anemic when it shipped over a year later.
So ultimately the question is why did the TouchPad fail in the market place? The software was definitely buggy, but we made pretty rapid improvements in the update releases. The iPhone was buggy when it launched but got better very quickly. The hardware specs were considered anemic, but really it should have been flying with a dual core 1.2ghz proc. The immature software held it back. My only real complaint about the hardware was the weight and the lack of a camera. The 7" Opal / TouchPad Go would have solved the camera issue but it was still too heavy. The other big issue was the price. It should have gone out the door at 399 or lower, even if we would have lost money for a while. We didn't have a comparable product to the iPad, so we shouldn't have priced it as such.
So why did it fail in the market place? Actually, I question that assertion. Clearly we weren't selling as well as the iPad but based on anecdotal evidence we were selling more than any Android tablet. One Best Buy employee told me the iPad sold about 10 units a day a his store while the TouchPad sold two a day. The other tablets (Android and BlackBerry) were lucky to sell one a week each. While Samsung may have shipped millions of tablets they later admitted that most of them were never actually sold to end customers. We were the #2 tablet even before the fire sale. A weak #2, but still #2. Not bad for a 1.0 product. If we continued working on the software, pushing out consistent updates, and shipped the Opal then we would have gone into the Christmas season with a strong momentum.
I will say this for both Apple, Google, and Microsoft. They go in big and they don't give up. They take the long view; steadily improving their products over months and years. Spending heavily on advertising then giving a product only six weeks in the market is *not* long term thinking. If HP was only going for a quick buck then webOS would never have survived, whether or not Leo was at the helm. This is a game for the rich and patient man.
So then why was the TouchPad canceled? As I said, we will never know. Perhaps the sales projections weren't realistic. But we do know that the *way* it was killed was designed to be irreversible. The press release announcing cancelation took everyone by surprise, including our partners. AT&T refused to carry the Pre 3 just before it was scheduled to go on sale. The Opal was canceled just days before it would have gone into production for a release in late September. Once you scupper a supply chain that way it takes at least a year to rebuild. The damage was irreversible. Even if Leo had been fired a week later and Meg reversed the decision, it still was too late. And it was that supply chain which cost most of the two billion dollar write down HP had to take later. I sometimes wonder if it would have been cheaper to build the TouchPad Go anyway and still sell it for 99$ like the TouchPad. In any case, the decision was made and it was immediately final.
Of note: when Meg Whitman was asked in an all-hands meeting how the board could have voted for such a decision, she said the board was "informed" of the plan, not consulted. No vote was required or taken.
Post cancelation stasis: my biggest memory of that period is when we were scheduled to go on a three stop tour of Asia (Australia, Singapore, and Bejing) to excite developers about products that now would never ship. In the end we made the trip because Richard Kerris (head of devrel) felt it was important to meet our developers on their own terms and be honest with them. It was both an interesting and awkward experience. I'm glad we did it.
Meg Whitman. During my final months at HP, Meg was refreshingly honest. She took her time to make the decision, but honestly we had plenty of time after the hardware cancellation. A few months either way would make no difference, and she had bigger fish to fry (like the PC division). In the end I think she made the best decisions possible under the circumstances. I wish her and the rest of HP well. I do not envy the work ahead of her.
Sam Greenblat & Martin Risau: I ramped down my travel after my son was born so I only came into the office twice after the cancelation. I only met each briefly. They both seemed smart and competent but had different visions of what Open webOS would be. It doesn't surprise me that there were conflicts.
- There was other hardware in the works beyond the TouchPad Go. I saw prototypes of many products including a transforming tablet / netbook combo with an ingenious sliding hinge and super thin keyboard. I really hope someone makes it one day.
- On "competing for #3" Yes it would have cost us billions of dollars over a bunch of years, but I think we would have been #2, not #3. It has been nearly a year since the TouchPad launched and Android tablets are still failures. I think there is a lesson in here for us: without a carrier and contract pushing them, Android devices don't sell themselves.
- WebKit: Using WebKit for the GUI was pure genius. Forking Webkit so you couldn't take advantage of community improvements was pure stupidity.
Oh. and the Palm V is still the best PalmOS device ever. It was the perfect size to fit in a shirt pocket, had a beautiful and unique shape, and was made of good metal materials. And the battery would last forever. It was an iPod Touch ten years early.
I am still very proud of what we created, especially my teammates in Developer Relations. The TouchPad launched essentially on schedule and had a solid over the air update. We opened the catalog with 10k webOS apps and 600 specifically for the TouchPad. That put us ahead of the number of Android Tablet apps at the same time, even though we shipped 6 months later. We continued to increase the number of apps in the catalog even after the hardware was canceled. We ran fun developer contests and promotions. Our developer community was always been passionate and appreciative, even in the times when we weren't allowed to say much. I'm glad we were able to open source Enyo in a way that lets those developers take their apps to other platforms.
Post Post Mortem
It has been nearly a year since the webOS hardware was canceled. A year that has afforded me some perspective. I left HP not because I didn't enjoy my time there. My reasons were more personal. After seven years of being a technical evangelist for Java, JavaFX, the JavaStore, and then webOS; I was simply tired. Tired of putting emotion and energy into platforms. Tired of the travel and speaking engagements. Tired of the constant product deadlines. Combined with having a (now) one year old child I realized I needed a break.
My current position requires little travel, has a flexible schedule, and lets me work on interesting things not tied to any particular platform. Best of all a specifically *can't* talk about what I'm working on. Perhaps one day I will return to platform evangelism. I still enjoy it. But for now I'm in a nice peaceful place.
Thank you for taking the time to read my tiny part of this epic tale. One day it will truly be the stuff of valley legend.
posted Wed Jun 06 2012