Software Everywhere — David Beers on Mobile Computing

I wasn't surprised to see a few shrill articles about the death of the Palm OS this week. The ten-year-old mobile operating system has been "dying" for most of its life in the assessment of the technical media—at least dating back to when the first Pocket PC devices hit the market in 2000. But the original PDA OS still dominates the handheld device market and has in recent years occupied a small but high-profile beachhead on the global smartphone market, thanks to the popularity of the Palm Treo. Still, when its owner ACCESS announced it was renaming the venerable Palm OS Garnet to "Garnet OS" it seemed even to many fair-minded observers to have suffered a humiliating blow.

At the risk of sounding absurdly optimistic in the face of the declining PDA market and the Palm OS's beleaguered status in the tough smartphone market, I am going on record saying that its renaming is actually something of a milestone in the rebirth of the "Palm" brand. To understand why this could be so you need to know a bit of history.

To start with, while ACCESS has been decried for the renaming, it wasn't actually something they had a choice about. In May of 2005 PalmSource sold the right to the previously shared Palm brand to then-PalmOne, and agreed to phase out its use of the name in all its products within four years. PalmOne immediately changed its name back to Palm Inc. and it was only a matter of time before ACCESS—party to the agreement by dint of its acquisition of PalmSource—would begin the process of removing the Palm name from all of its products. When PalmSource was renamed ACCESS last October this was symbolic of the integration of parent and subsidiary, but it was also the first step toward compliance with that two-year old agreement. Changing "Palm OS Garnet" to "Garnet OS" is just the next step. Eventually ACCESS products like Palm Desktop will have to be renamed or replaced as well.

What's more interesting is what the change means for Palm. As has been widely publicized, Palm reacquired from ACCESS the right to use and develop the Garnet OS source code in pretty much any way it choses as long as backward compatibility is maintained, as established by a mutually agreed upon test suite. Less publicized is the fact that Palm is actually exempt from having to use the name Garnet OS in any products it releases with this OS. In explaining this deal with Palm, ACCESS stated:

Under terms of the agreement, we have given Palm the right to use either "Palm OS by ACCESS" or the product's new name... when referring to the operating system licensed with this new agreement.

So from Palm's perspective this renaming exercise by ACCESS would appear to mean next to nothing.

But appearances can be deceiving. As I've mentioned several times before, Palm's reacquistion of the Garnet source code was not just so Palm didn't have to depend on ACCESS to make minor tweaks and bug fixes. The $44M agreement included concessions by ACCESS that give Palm "the right to use Palm OS Garnet in whole, or in part, in any product from Palm and together with any other system technologies." Palm's press release about the agreement was a bit more definite about what this meant:

Palm has secured an expansion of its existing patent license from ACCESS to cover all current and future Palm products, regardless of the underlying operating system. [emphasis mine]

I've gone into some detail in several other posts about the evidence that Palm needed this agreement to enable it to develop it's own Linux-based successor to the Palm OS, apart from that being developed by ACCESS. If I'm right about this, what makes the renaming of Palm OS Garnet so interesting, is that it actually clears the way for Palm to give the name "Palm OS" to its next-generation operating system. Palm owns the brand, it owns its OS, and as of this week no other company will be able to use the name Palm OS.

The Palm brand had become quite a mess in the last few years, what with the ill-fated PalmOne/PalmSource split and the subsequent buyout of PalmSource, but since Ed Colligan took the helm at Palm he's done a great job in putting the brand back together again. More than just the brand, he has hired back many of the original developers of the Palm OS and regained the right to build the next generation of the Palm OS. From where I stand it looks like this project has been under way for at least a couple of years, and I expect we'll be hearing an announcement about a modernized but Garnet-compatible "Palm System 7" (or some such) within the next few months.

Yesterday, ACCESS issued a press release stating that they had given Palm a perpetual license to the Palm OS with expanded rights to modify it and—critically—the right to "use Palm OS Garnet in whole, or in part, in any product from Palm and together with any other system technologies." Palm waited until today to issue their own release. Here are the salient bits:

• "Palm has a perpetual license to use as well as to innovate on the Palm OS Garnet code base. Palm will retain ownership rights in its innovations."
• "The new agreement also provides Palm flexibility to use Palm OS Garnet in whole or in part in any Palm product, and together with any other system technologies. The company plans to ensure that applications now compatible with Palm OS Garnet will operate with little or no modification in future Palm products that employ Palm OS Garnet as the company evolves it over time to support Palm's product differentiation strategy."
• "Palm has secured an expansion of its existing patent license from ACCESS to cover all current and future Palm products, regardless of the underlying operating system." (my emphasis)

I like that Palm is framing this agreement in terms of supporting its product differentiation strategy. I know I've been making the argument that these negotiations with ACCESS were critical to its ability to "develop and distribute new products based on a next-generation version of the Palm OS" (Palm's own words) and that that next-gen Palm OS would run on Linux. The emphasized phrase in the last bullet point underscores once again that this is a part of what Palm plans to do and something they sorely wanted from this agreement. But I'm of the mind that there are still lots of interesting and profitable new products that could be developed on a modestly refreshed Palm OS Garnet kernel and framework. I believe we'll see Garnet on future Palm devices in two forms: as a compatibility layer in an advanced multi-tasking OS that Palm themselves are developing, and as a standalone smart device OS similar to what we have today.

Garnet's roots in the original Palm Pilot, a device with miniscule RAM and processor resources, have made it the most performant major mobile operating system on the market today and a paragon of efficiency. The experience using Palm OS on the new Treo 680, for example, is darned good compared to the competition. A lot of the credit goes to Palm for innovating within the limited rights it had on top of the base OS. Not all mobile devices that people will want to use in the coming years will require the advanced OS features that Linux offers. In the short-to-mid term, while mobile Linux is getting the kinks worked out, those features may come with significant overhead that demands faster processors, more memory, and bigger batteries, impacting costs and important design criteria. So there is reason to believe that during that period Palm may be able to better satisfy certain classes of customers with the traditional Palm OS, or an improved version that preserves the current kernel and architecture.

Palm is not a big company and can't risk losing its connected device focus, but I hope we'll see them continue to roll out a nice line-up of handheld computers that stick with the tried and true Palm OS architecture. Palm has regained its dominant position in the still-sizable handheld organizer market and there's still a billion dollar paper planner industry out there for well-designed entry level Palm OS PDAs to whittle into. I think Palm's emphasis on supporting their product differentiation strategy—and the sizable price tag paid for these expanded rights to the Palm OS Garnet source code—are signs that they have much broader plans for the coming years than just being the maker of the Treo.

Unanswered questions about this agreement:

• What about the pieces of the Palm OS that aren't on the device: HotSync Manager and Palm Desktop? I'm not sure, but it doesn't sound like these were included in the agreement.
• Is this agreement a one-time payment that relieves Palm of future royalty payments to ACCESS for its use of Palm OS Garnet? Seems like an obvious question that I'm surprised isn't discussed in either of the press releases or the FAQ published by ACCESS.
• If ACCESS is now OK with the idea of Garnet being used as an execution environment that could potentially run on top of any number of different operating systems, will they consider this as a new business strategy rather than just a one-off agreement with Palm? Might they try to position Palm OS Garnet the way they do the Java environments they sell: a kind of virtual machine that licensees can add to almost any phone platform they might be using? It would be interesting if they did this and tried to turn Garnet into a de facto industry standard execution environment like Sun, QualComm, and Adobe have done with Java, Brew, and Flash. But my guess is this deal is a one-off. ACCESS wants ALP to become an industry standard, not Palm OS. Those are not mutually exclusive options, but there's a difference in emphasis that's important for their marketing.

You didn't have to read too much between the lines to know that Palm and ACCESS have been involved in a big stand-off ever since ACCESS acquired Palm's OS supplier, PalmSource. When ACCESS announced their new Linux platform in February there were official words of enthusiasm from the Linux community, various mobile operators, several hardware companies and software vendors, but ACCESS's biggest Palm OS customer, Palm, was silent as a stone. Palm has kept up this silent treatment for a full year now, with the only sign of what was going on behind boardroom doors being a statement buried in a July SEC filing:

Contemporaneously with the [Palm OS] license agreement, we entered into a co-development agreement with PalmSource to develop a next-generation Palm OS for use in future Palm products. PalmSource did not timely meet certain of the milestones under the co-development agreement, relieving us of our obligation to make minimum royalty payments under the license agreement after calendar year 2006. We are presently in negotiations with PalmSource to expand our development and distribution rights to the current version of the Palm OS. If we are unable to successfully conclude these negotiations, it may adversely affect our ability to develop and distribute new products based on a next-generation version of the Palm OS.

Well, tonight we learn that Palm finally got what they wanted from ACCESS. As stated in the ACCESS press release (we don't have Palm's statement at this time) the two companies have entered into an agreement that gives Palm a perpetual license for the Palm OS Garnet source code. Most importantly:

Under terms of the agreement, ACCESS has granted Palm specific rights to modify the code base of Palm OS Garnet for use in its devices such as the Palm Treo smartphone family and the company’s other handheld computers. The agreement also grants Palm the right to use Palm OS Garnet in whole, or in part, in any product from Palm and together with any other system technologies. To ensure forward-compatibility of Palm OS Garnet applications, ACCESS and Palm have agreed to continue to measure compatibility against the compatibility test harness in use between the two companies.

Does this mean that Palm wants to keep cranking out products on the old Palm OS for as long as they can get away with it?

No. What I believe it means is Palm now has the right to use Palm OS Garnet "in whole, or in part... and together with any other system technologies" to develop Palm's own next-generation Palm Operating System. The terms of this agreement validate my interpretation of the July SEC filing that Palm has been planning to regain control of its OS destiny by negotiating for the right to include Garnet as a legacy execution environment within an advanced, probably Linux-based OS successor of Palm's own devising.

You can see why ACCESS has been slow to accept these terms. I'm sure they've been showing Palm the progress on the ACCESS Linux Platform and trying to woo them with ALP's Garnet compatibility and multitasking MAX application framework, hoping that Palm would be among the first licensees. We don't know whether Palm had concerns about ALP itself, or if they simply didn't think their business vision was sufficiently aligned with ACCESS's to risk dependence on ACCESS for their critical system software. But for whatever reason, Palm wasn't buying ALP. And apparently ACCESS became sufficiently concerned that they would lose Palm entirely as a licensee that they decided the best thing was to give up on the ALP hard sell and take Palm's money ($44M) in exchange for Palm's freedom to do what they pleased with the current version of Palm OS.

This is a good thing for all concerned, in my opinion, and I wonder if the time it took for the companies to strike a deal wasn't simply spent with attorneys wrangling over the terms of an agreement they all knew they would reach. If both ALP and Palm's next version of Palm OS support Palm OS Garnet applications, Garnet essentially becomes a cross-platform execution environment like Java or Brew. And not necessarily just across two platforms. If Palm creates a new product that isn't really a PDA or smartphone and has a very different kind of OS, the agreement enables Palm to include a Garnet emulator or compatibility layer for legacy Palm OS apps to run on that OS as well. All this gives credibility to the longevity of Garnet and all the thousands of applications written against the Garnet APIs. This in turn helps give ALP some traction right out of the gate as one—even if not the only—successor to a Palm OS that has a strong future and a viable ecosystem. Developers can have confidence that their applications will have a market not just with customers of ALP devices, but with their usual Palm customers as well, which should inject a little life back into the flagging Palm OS economy.

Palm gains from this deal not just because of the freedom to use Garnet in a future Palm OS that they control, but also because they may need to sell some more products on a spruced up version of Garnet before their next-gen Palm OS is ready. I know I wrote here that someone familiar with the matter told me Palm's first Linux products will be released next year. And Ed Colligan's recent remarks may give some credence to that report. But that product is likely to be for early adopters who are willing to suffer through some of the birthing pains of a new operating system (and maybe a bold new device category) to be in on the ground floor of Palm's Next Big Thing. If it does run Linux it may initially face some of the challenges we've seen on devices like the Nokia 770 Internet Tablet: subpar performance and battery life, for example. Linux systems are powerful and sophisticated, bringing many benefits to vendors and end users of mobile devices, but they may require another year or so of Moore's Law and software optimization before one can deliver the kind of snappy, run-for-days user experience that Palm users have come to expect. During that time the venerable Palm OS Garnet—with renewed ministrations from Palm's engineering staff—should keep the Palm OS torch burning.

It will be very interesting to hear how Palm reports on the new agreement. This could be the time when Palm opens up about some of its plans and verifies publicly some of the things I've speculated about on these pages.

Update: See also the FAQ about the new agreement that ACCESS posts on their site. It reads in part: "The agreement also grants Palm the right to integrate Palm OS Garnet on top of other operating systems." I believe that this right alone is what made the $44M deal worthwhile to Palm and what prompted the long negotiation with ACCESS.

Update: Palm has issued their own press release now, which I analyze here.

Palm announced the latest addition to their Treo smartphone line last week: the Palm OS-powered Treo 680. Perhaps I should have said "ACCESS-powered" since that's the brand that's stamped on all the new Palm OS devices now and since OS maker PalmSource has itself adopted the ACCESS name as the next step in the integration process with its parent company. (Check out their snazzy new merged web site, by the way, a complete replacement of the old ACCESS site.)

The entry-level Treo 680 wasn't a big surprise to anyone who has closely followed Palm over the last year, but it raises some interesting questions. If you were among the folks who was hoping this would be the announcement where Palm unveils a Treo with a next-generation version of the Palm OS, sorry, no dice. Not yet. While Palm remains tight-lipped about what the next step for the Palm OS will be (they have yet to approve the presumed successor, the ACCESS Linux Platform) they remain adamant that they are sticking with their dual-OS strategy. So one question raised by the 680--the first new GSM Treo running Palm OS in two years--is how long will Palm ride the new horse before they decide it's time to rev its old OS?

While I'm betting this Treo will be the last Palm OS Garnet smartphone to come out of Palm, I wouldn't be surprised if they plan to let the 680 run for a good year before they announce a Linux-powered successor. The ACCESS Linux Platform is not going to be delivered to licensees for a few months and will take another 12-18 months to ship on any handsets, due to integration, carrier testing, and regulatory hurdles. Even if Palm has secretly partnered over the last couple of years with someone other than ACCESS to develop their own Linux variant of the Palm OS, as I've suggested is likely, that's a very big project that's unlikely to give fruit for a while yet.

There are other reasons for Palm to hold off in bringing out a Treo with a next-generation Palm OS. I expect the 680 to be a substantial success for Palm, one they won't want to squelch early by announcing the obsolescence of the Palm OS. There's no question in my mind that the new sleeker, slimmer form-factor and the distinctive color options will breathe some new life into the old platform. Yes, the technical changes are incremental (more memory, tweaked software, supposedly improved reception) but I expect just the cosmetic improvements together with the aggressive pricing will substantially drive sales and expand the demographics of the user base. The volume of Treos that Palm is shipping now should be giving them some leverage to keep the bill of materials down on the hardware, so I'm guessing the Treo 680 introduced on Cingular for $199 will do very well for Palm, despite the new competition from Nokia, Motorola and BlackBerry. None of these contenders have the Treo's touchscreen, screen resolution, usability and pedigree.

While the 680 is clearly designed to appeal to a consumer market as compared to the Treo's prosumer and enterprise focus of the past, I believe it could have good legs in the enterprise as well. It's got everything that the popular Treo 650 has (including, I expect, BlackBerry Connect) but the low cost of the handset really starts to make some large smartphone deployments feasible, like the one being planned by my current client. The absence of a 3G radio may be more of a boon than a liability for many companies. 3G doesn't add a lot of value to most enterprise applications--mainly bigger wireless bills. All you can eat data on Verizon's EvDO network is $80/mo compared to $20 on Cingular's EDGE network, which is plenty fast for email and business apps.

Travelling through airports four or five times a month I meet an amazing number of business people with Treo 650s who almost universally favor them over the BlackBerry. (Pull out a smartphone next to me on a plane and I become quite a pest with my questions!) By comparison I see very few of the Windows Mobile Treo 700w. These business users have no concept that the Palm OS is "outdated"--only that they find their Treo 650s to work better than anything else they've tried. Palm will need to update the OS before very long, but I think with the right hardware Palm OS Garnet has a couple of good years left. The devil is in the details, of course, but from what I can see so far the Treo 680 is looking like it could be a winner for Palm.

Holy crap! The Palm community has become completely unhinged over Friday's Palm Inc. annual report. (See my previous analysis here). There is very little accurate reporting going on. Let me try to straighten out the mess. The first thing to understand is that the statements from the annual report that are being repeated (more often badly paraphrased) are made under SEC rules requiring Palm to disclose every conceivable business risk that investors should be aware of (even the possible effect of a terrorist attack). This isn't a press release, it's a legal document. Keep this in mind as you read quotes from the report.

Here are some of the more eggregious misstatements of the facts from various blogs and mainstream tech news outlets:

Misstatement #1:

The future for the Palm faces a bit of turmoil, with the recent decision by them to stop using PalmOS. (TechSpot)

Correction:
The annual report is arguably one of the strongest statements of Palm's continued commitment to the Palm OS that we've had all year. It makes very clear that Palm wants to create products on a "next-generation version of the Palm OS" (even if they have to substantially make it themselves, they hint) and that there will be new products released on the existing version of Palm OS until that happens.

---

Misstatement #2:

Palm's License to use Palm OS Expires in December
(DailyTech)

Correction:
Palm's license to use the Palm OS is totally intact through December 2, 2009. Palm even has an option to extend its license for two years beyond that at a cost of $10M per year. Nothing stated in the annual report has changed any of this.

---

Misstatement #3:

Due to a disagreement with PalmSource -- the developer and supplier of Palm OS -- Palm has decided to stop paying royalties to PalmSource. (DailyTech)

Correction:
The annual report says nothing about a disagreement with PalmSource. PalmSource was late in meeting some project milestones for its Linux-based successor to the Palm OS, which (for now) relieved Palm from having to pay royalties after December of this year. But the report states that Palm and Palmsource are negotiating a new agreement, which would involve Palm paying royalties to "expand [Palm's] development and distribution rights to the current version of the Palm OS [Garnet]."

---

Misstatement #4:

Palm on Friday warned investors that development delays by PalmSource on Palm's next operating system have hurt the hardware maker's ability to compete in the smart phone and PDA markets. (CNET)

Correction:
Palm made no such warning. They did state that their "business could be harmed if PalmSource and Access do not continuously upgrade the Palm OS and otherwise maintain the competitiveness of the Palm OS platform." Actually, this is a correction to a statement I myself made Friday to the effect that Palm didn't mention PalmSource's delivery of ALP as a business risk. Clearly PalmSource's performance and ACCESS's business plans do make a difference to Palm and it's safe to say that these aren't idle concerns. They are reasons why Palm is seeking a new agreement with PalmSource that gives Palm more control over the future of the Palm OS.

---

Misstatement #5:

Despite those potential hurdles, Palm says it is continuing to work with PalmSource to develop a new operating system featuring a Linux kernel, though no timetable has been set. (CNET)

Correction:
Actually, we don't know this. Palm said that it had a co-development agreement with PalmSource, but did not say anything about whether the new agreement it is seeking would involve further development work by PalmSource or not. In fact, the wording of the annual report reads fairly strongly in the direction of Palm not being interested in PalmSource's future work on ALP. The negotiations that are under way are about Palm's development and distribution rights over Palm OS Garnet--something PalmSource has shown no interest in developing other than to create the GHost emulation layer that enables legacy Palm applications to run in ALP. It stands to reason that Palm either wants to enhance Palm OS Garnet to create a next-generation Palm OS or to deliver a new platform of its own that, like ALP, uses a Garnet emulator to provide backward compatibility. If Palm wanted to continue working side-by-side with PalmSource on the next version of Palm OS, it seems to me the negotiations would be about how Palm could contribute to ALP, not Palm OS Garnet. This is an interpretation on my part, certainly, but at the very least we can say that it's not clear that a new co-development agreement is on the table.

---

Misstatement #6:

The failure of PalmSource and Access, Co, Ltd. to meet contract obligations may result in the termination of Palm Inc's involvement in the development of a future Palm OS successor. (Brighthand)

Correction:
If anything, it seems the other way around: Palm looks determined to regain control over the Palm OS even if this means cancelling its co-development agreeement with PalmSource and anteing up for the rights it needs to rev the OS without PalmSource. How this will actually shake out is in question (will there be two pretenders to the Palm OS throne or will a new co-development partnership be reached?) but I certainly don't see it as a case of Palm becoming less involved in the development of Palm OS.

---

Misstatement #7:

Unless PalmSource can get back on track, Palm will get to use Palm OS Garnet for free until 2009. (Brighthand)

Correction:
Thanks to Surer at TreoCentral for pointing out that Palm can only get out of its minimum royalty obligations, which is the flat amount it must pay if Palm OS device shipments fall before a certain threshold. This doesn't get them out of paying royalties for Palm OS Garnet on a per-device basis. In Surer's words: "The way it works, there is a royalty cost for each device with PalmOS, and if its less than the minimum royalty they have to pay the minimum e.g. Sony still had to pay $20 million until recently. If they ship MORE, they have to pay the difference too. This is all in the 8-K filing."

---

Conclusion
So where does this leave the Palm OS? It's late and I'll say more about this later, but while none of this reflects well on the decision to split Palm into two companies four years ago, my feeling is that it's not going to be bad news for the Palm OS over the long haul. It could even be good for ACCESS's licensing of ALP and for Palm's caché if ALP is differentiated from Palm's platform. We will have to wait to see what Palm and PalmSource hammer out as their new relationship, but even if it becomes more of an arms-length transaction and less working side-by-side I'm bullish about what both of these companies can do in the mobile Linux space. Linux is growing very rapidly now and will grow even faster in the coming years. There's great talent at both Palm and PalmSource to capitalize on that growth with compelling, easy to use smart device platforms. It may not ultimately make much of a difference which platform is considered the "true" successor to Palm OS. If Palm OS Garnet support is available on more than one operating system and the differentiation leads to wider deployment this could actually attract developers to these platforms. While Garnet emulation will be less powerful than a new Linux-based system it is a vast improvement in power and responsiveness over other popular cross-platform virtual machines like Java and Flash Lite. If ALP and Palm's Linux OS end up coexisting, ACCESS should consider marketing its Garnet emulation as a powerful cross-platform environment that can run on any Linux touchscreen device. (Remember, StyleTap has already extended the Garnet API to Pocket PC devices.)

Admittedly, Garnet-as-a-better-Java-ME is a "happy day" scenario. The sadder fact is that ACCESS and Palm seem to be moving from a position in the market where Palm OS is a powerful brand to one where there is great uncertainty and confusion over what Palm OS means. It will take some clever, proactive marketing and--most important--a series of design wins by Palm and other ALP licensees to overcome this obstacle. But we knew that already, didn't we?

Palm's annual report was released today. It's pretty revealing about where they stand with regard to PalmSource/ACCESS and the once presumed successor to the Palm OS, the ACCESS Linux Platform. For those who don't follow the Palm OS closely, Palm's OS division was spun off as PalmSource in 2003 and subsequently acquired last year by Japanese mobile software company ACCESS. PalmSource announced the Palm OS-compatible ACCESS Linux Platform in February and promised a release before the end of this year.

From the section of the annual report where Palm must spell out all its business risks it reads at one point:

Contemporaneously with the license agreement, we entered into a co-development agreement with PalmSource to develop a next-generation Palm OS for use in future Palm products. PalmSource did not timely meet certain of the milestones under the co-development agreement, relieving us of our obligation to make minimum royalty payments under the license agreement after calendar year 2006. We are presently in negotiations with PalmSource to expand our development and distribution rights to the current version of the Palm OS. If we are unable to successfully conclude these negotiations, it may adversely affect our ability to develop and distribute new products based on a next-generation version of the Palm OS.

Ok, so Palm and PalmSource were working together on the new Palm OS, PalmSource missed some project milestones, and Palm told them "we need a new agreement." And the agreement they are seeking is... what now? For Palm to get expanded development rights to the current version of Palm OS--not the Linux version that was being developed under the previous agreement. Why would they want this, and why do they say that if these negotiations fail it will affect their ability to ship products with a next-generation Palm OS? The clearest interpretation is that Palm plans to take charge of developing its new operating system instead of PalmSource/ACCESS.

As I've discussed elsewhere (and had confirmed by an insider) Palm has indeed been separately working on their own Linux successor to Palm OS, but the annual report provides the first official information that suggests this development. We can surmise that the need to negotiate for expanded development rights on the existing Palm OS code base is due to a desire to provide backward compatibility with existing Palm OS applications, perhaps to get the rights to distribute a Palm OS Garnet emulator and HotSync technology within their new platform.

Elsewhere in the annual report we have it that Palm hired 130 people for research and development last year, increasing R&D expenditures 51% over FY2005. That's a lot of new people doing research in a company the size of Palm--another hint of Palm's new ambitions.

I can think of some good reasons why Palm has apparently decided not to license PalmSource's ACCESS Linux Platform. Given the language above as well as promises elsewhere in the report to release new Palm OS products, it's obviously not because Palm is planning to go completely over to Windows Mobile (although that could be an eventual fallback position if the negotiations described fail). A more likely reason is this: since PalmSource was acquired by ACCESS, Palm entered a very risky situation of depending on a new and foreign company's business plans--a company with aspirations to license the PalmSource operating system much more broadly than PalmSource was ever able to do. If ACCESS is successful in this ambition it would dilute Palm's uniqueness in a market that is already getting very competitive. Even if the number of vendors allowed to license PalmSource's operating system were limited to avoid this, there would still be no guarantee that ACCESS would share the same priorities as Palm. For example, one of the primary focuses of ALP is on delivering a better user experience for handsets that lack touchscreens--a device class in which Palm has yet to show any interest.

It's no surprise that Palm bid fiercely to acquire PalmSource itself when ACCESS and Motorola started their bidding war last year. Let's hope for their sake that they are successful this time in negotiating the rights they need to bring the next version of Palm OS to market.

Related article: Seven misconceptions about Palm's new annual report

There hasn't been an announcement yet, but Palm has filed with the SEC to add a new position to its board. The interesting thing is that the new board member is none other than Bill Coleman, co-founder of BEA. Last I checked, Coleman decided his retirement from BEA didn't suit him and had created a new web services start-up called Cassatt Corp.

So why would Palm create a new seat on its board for a guy who's entire career has been about how "the network is the computer"? Perhaps I wasn't far off when I speculated earlier that Palm, too, is coming to see the power of the server side.

I reported back in April that the 3G GSM Windows Mobile Treo people have been calling "Hollywood" would be released on Vodafone this month. Well, it sounds like the release date has been pushed back to "later this year" for this much-anticipated new Treo, but we do have an official announcement from Vodafone now.

Getting a stronger foothold in Europe is a big deal for Palm since smartphone adoption and mobile email usage is much higher there. They will campaign together with Microsoft with push email and superior user experience as the driving message to European business customers. I'm not too surprised about the delay. A Cingular sales rep I spoke with last week said his company had plans to release both of the new Treos that Palm had promised for later this year (the rumored codenames are now "Lennon" and "Nitro") but that Palm had just set the release date back for some indefinite period of time.

As I've discussed here before, Palm has an interesting business problem when it comes to releasing another Palm OS Treo this year. The 700P is very fresh and designed to run on EvDO networks, so logically you would suppose that Palm would want to deliver a comparable Palm OS Treo to run on 3G GSM networks. The problem is that Palm OS Garnet is going to have great difficulty, in my judgment, complying with the stricter UMTS standards that are applied to 3G GSM networks. Specifically, the requirement to handle simultaneous concurrent voice and data streams without pausing the data stream during a call is a tricky problem for the Palm OS, which at the application layer is fundamentally a single-tasking system. Hopefully Palm has figured out a way to make this work. I've speculated that Palm would try to ship its next Palm OS Treo with it's semi-secret skunkworks Linux system to solve this problem, but the intelligence I've been able glean leaking out of Palm indicates that that won't be ready until some time in 2007. If Palm hasn't managed to retrofit Palm OS with some multitasking capability (at least for the phone application) you have to ask yourself what the selling point for that next Treo is going to be. How will Palm differentiate it from the Treo 650 if they can't tout high-speed data?

Final Update: On further examination, this rumor looks sufficiently baseless that I can only recommend reading on as an exercise in pure speculation. I leave it as a warning sign to other bloggers in hopes that they will be less gullible than I.

We interrupt our Eclipse Week programming to bring you this: A CNet reporter says PALM and RIMM may be announcing a merger this week. Apparently the two companies have rescheduled their quarterly conference calls to coincide with each other. It's still just a rumor at this point, but with PALM being cheap and under pressure to entertain a buyout and RIMM watching the Treo bear down on the BlackBerry at high speed, it sounds pretty plausible.

A blog called Covert Stock Operations blog has some detailed analysis of the merger prospects and outcomes from Palm's and RIM's perspectives. Here's the juice of it:

1. PALM will insist on a merger of equals. (This would neccesitate a 2 PALM shares for 1 RIMM share swap + cash)
2. PALM will take the CEO stop.
3. Jim Balsillie has to go. He will be given a very nice package. The legal department isn't responsible for the NTP fiasco. Jim Balsillie is. He ignored good legal advice and almost got the company into bankruptcy. (RIMM's and PALM's boards are calling the shots now.)
4. The new company will be trading on the NASDAQ, under the symbol (PALM).

The 80+% of PALM that is held by instituations have been pressuring PALM to take a buyout. A buyout is in the works. RIMM is the most likely candidate, as they need PALM to survive.

I need to get with some of my contacts who are close to PALM and get the skinny on this. I'll report back shortly with more details and some of my own analysis.

Update: I spoke with one of the institutional investors that owns a good chunk of PALM and they reported that they thought this was probably just a rumor. They do think PALM would be a good buy, though. Gee, what a surprise there! :-D

Another update: I was duped. The CNet "reporter" that Yahoo referred to wasn't a reporter at all. It was a blogger posting a rumor that traces indirectly back to the same Covert Stock Operations blog that I quoted above, which in turn is just publishing a bunch of guesswork. Embarrassing. No more mergers and acquisitions rumor posts for this gullible blogger!

Thanks to Ryan at PalmInfocenter for publishing a piece of mine that riffs on some themes I've hit on recently here concerning Palm's upcoming product plans. It's mostly based on analysis of the rather descriptive job postings that Palm puts up on their web site, something I've had a lot of fun with lately.

One thing I mention in the article that I haven't discussed here is that Palm seems to be working on software to enable remote device management including firmware updates over the air (FOTA). This would be a great innovation which could potentially leapfrog Microsoft's modest foray into device management in its recent Messaging and Security Feature Pack for Windows Mobile 5.0. In any case, it's nice to see evidence that Palm is once again taking seriously the idea that they could develop products that would compete in the enterprise mobility market segment.

Motorola's "Treo Killer," the Windows Mobile-powered Q has suffered from repeated delays, but we're assured that at long last it will be released on Verizon's network any day now. Interesting that just last week we had images surfacing under suspicious circumstances of a new Treo 700p with Verizon badging on it.


Is the timing a coincidence? You be the judge!

Me? I figured this was probably a hoax perpetrated by an enthusiast, like so many we see showing up on TreoCentral. But I decided to check into it all the same. I contacted my analyst friend with the connection at Palm and, lo and behold, he said he'd been informed that the Treo 700p would be released on both Sprint and Verizon. He sure has been a fount of information about upcoming Treo releases this month, hasn't he? Naturally, I inquired about the timing of the 700p releases. He didn't know exactly, but he did say that Sprint "might" not get the same kind of exclusivity period that VZW did for the 700w "because they were offered the 700w first and passed on it." So not only will the 700p be released on Verizon, but it might be released pretty soon? Hmmmm.

I really appreciate having a ready source of inside information about Palm, but I'm sorry: my FUD-meter is pegging way into the red zone at this point. Since my source has been quite reliable in the past I was initially inclined to take this information at face value, but now I'm wondering if Palm's cozy new relationship with Microsoft has taught someone there a few underhanded marketing tricks. Let's just say I'm going to be a little more cautious in how I interpret information that comes through this channel until I see that the track record continues to be good.

So, will there be a new Palm OS Treo coming out on Verizon this year? Sadly, my final answer is: I don't know any more than the rest of you.

Things are getting interesting at Palm Inc., the makers of the Treo smartphones but not (to date) of the operating systems they run on. First, back in November of 2004, there was a source that reported Palm was "exploring partnerships that could let it use a tailored version of the Linux OS to run on its devices". There was some credibility because this was the same source that first leaked Palm's plans to release a Windows Mobile Treo--correctly as we now know. Other rumors of Linux devices from Palm flew about for a while, but there was nothing tangible. Then Palm went through a spate of hiring Linux engineers that seemed to start around the time they lost their bidding war to acquire PalmSource. Most recently there has been a job description posted on Palm's website for a Linux handset engineer that specifically mentions development of "a new software platform." Now I've received confirmation from someone who actually knows what's going on.

The analyst I communicated with by email has been privy to Palm's plans in the past, so I asked if he could confirm my conclusion that Palm is writing their own Linux OS. He did, and they are. He asked not to be named, but here's what he said when I asked what he knew:

We know about it. Palm has stated on several occasions than 80% of their engineers are software engineers, they are perfectly able to design their own operating system and are working on it. I know they had a prototype of a Linux Treo 650. From my discussions with management my impression is that Palm is hedging their OS risk by developing in-house systems as well as keeping relationships with Palmsource and Microsoft. I know that future Palm models will be coming out with both operating sytems in order to test market reaction.

I asked if he knew whether any of the Treos slated for later this year were expected to run Palm's new Linux-based system. He said they would run Windows and Palm OS Garnet and that the "new operating system will be introduced in 2007 (don’t know if fiscal or calendar)."

So there you have it, folks. It seems that we now have not one but two companies working on a successor to the Palm operating system: PalmSource, the company that Palm spun off in 2003 to continue the development of the Palm OS but which was recently acquired by ACCESS; and Palm Inc. itself, which own's the Palm brand and is therefore the only party that can actually name their platform "Palm OS." It's thought that ACCESS, heavily staked by DoCoMo, will be looking largely to Japan and China for its customers. Meanwhile Palm, dominant in the US smartphone market, has been conspicuously quiet about the announcement of the ACCESS Linux Platform in February. Now we know why.

The more I think about it, the more I wonder if Palm Inc. couldn't surprise us with their own "Palm Linux" on one of the Treo's they're supposed to release later this year. Last week I looked at the evidence that Palm really is putting money and manpower against a new Linux platform. On the plane to Chicago tonight I whipped up a box diagram that shows how I think Palm could be using Linux to solve the sticky business problem I talked about:

This is a much more modest system than what ACCESS and PalmSource have announced, but it's got some interesting things going for it. The first thing to notice is that in addition to running Palm OS applications in an emulated 68k environment (same as Palm OS Garnet) this runs a handful of built-in native Linux apps. These native applications are carefully chosen based on the benefit they can create for the user in terms of multi-tasking: They either deal with tasks like a phone call or incoming text message that are likely to cause an interruption to a task in progress, or they deal with long-running tasks that would benefit strongly from being able to run in the background so the user can attend to something else--email, browser, and audio player come to mind.

Palm OS Garnet can only run a single application at a time because Palm applications are launched as subroutines of a single-tasking UI Application Shell. That won't change here since these apps will run in an emulator that precisely mimics Garnet. But in this architecture, anything inside the yellow boxes runs in it's own process and has protected memory. If Palm develops a UI framework that enables easy, intuitive switching between a running Palm OS app and one or more native apps the impression to the user will be that the Palm OS is now a multi-tasking OS--at least for operations where that multi-tasking is of the most value.

That's nice for users, but it's absolutely critical for GSM operators that would like to sell Treos that use their shiny new broadband networks. The UMTS standard they must meet requires that the phone be able to simultaneously place data and voice calls without dropping or pausing either call. Even though Garnet has a premptive multitasking kernel it's UIAS won't allow that to happen since it would require running two event loops and two stacks within dynamic memory--something Garnet just can't do. So this platform fixes the problem Palm Inc has of bringing the popular Palm OS into compliance with all the major 3G standards.

Before ACCESS acquired PalmSource, Palm had made statements that they were enthusiastic about the Linux-based successor to the Palm OS. But since the announcement that ALP will be that successor there hasn't been a peep from Palm. Not even a supportive quote that PalmSource could put in their press release. Spooky.

That could mean what it seems to mean on the face of it: Palm isn't interested in the direction PalmSource is going with the platform since ACCESS redirected it. Or as Mike Mace has pointed out there could be other good reasons for their silence. But even if they are privately interested in ALP, Palm has a problem when it comes to supporting their large and avid Palm OS user base. The outline of this predicament is as follows:

• ALP will run Palm OS apps, but the earliest time Palm could ship a device that runs ALP would probably be two years from now
• Palm OS Garnet will in all likelihood not support the UMTS standard used on GSM 3G networks, which differs from the standard employed on CDMA data networks like EV-DO in that it requires the capability of concurrent voice and data transmission
• Operators are losing interest in new handsets that aren't 3G

Barring what I gather would be a minor engineering miracle, don't expect Palm to try to push any more Garnet Treos after this year. And don't expect them to just sit out the global GSM market for a year or two until something like ALP is ready. That market is just too big and important and the competition is moving much too quickly. So what are their alternatives?

They could do what a lot of folks suspect they plan to do: ditch Palm OS completely on the Treo line after the currently planned releases and go 100% with Windows Mobile. That would certainly be a streamlined solution from an engineering standpoint. But if there was any way to avoid it, why burn the bridge to the loyal Palm OS user base, especially when so many of their customers keep moving up to the high-margin Treo 600 series from the lower-margin Palm OS handhelds? I think this move would confirm in just about everyone's mind that the Palm OS is going away and accelerate the collapse of Palm OS PDAs, which still constitute a substantial part of Palm's revenue. It would also put Palm in a much harder position for avoiding commoditization. The customizations that Palm can do to improve Windows Mobile are customizations that other vendors will be able to match much more easily than the value that Palm could add to their already popular Palm OS Treos. I think we all expect Palm to become a major vendor of Windows Mobile smartphones in the coming years, but I'm skeptical of the idea that they'll be abandoning their dual platform strategy any time soon.

Some (myself included) have speculated that Palm may have started working on their own Linux platform a couple of years back, once they determined that Palm OS Cobalt was not going to fly for them. The source that first reported that Palm was working on a Windows Mobile Treo reported at about the same time (November '04) that Linux-based devices were in the works. (Consider: PalmSource may actually have been prodded into it's original Linux announcement in December '04 by news that Palm was moving in this direction.) Also, throughout most of last year and as of this writing Palm has continuously posted job openings for Linux handset engineers. Consider this job description that's on their site right now:

At the risk of adding to the endless "Palm OS is dead" drumbeat, it really should be emphasized that PalmSource is not working on a new version of Palm OS. Computing Unplugged received a correction from PalmSource when they stated merely that the MAX application framework would "inherit much of the traditional look and feel of the Palm OS." From Maureen McConnell at PalmSource:

It's not particularly our intention that MAX inherit much of the traditional look and feel of the Palm OS. While this paradigm works fine on PDA-like devices with touchscreens, it's not as effective on more 'phone like' devices -- MAX is intended to address both effectively.

Some folks are still having a hard time wrapping their heads around what ACCESS and PalmSource actually announced last month. PC Pro, for example, read the above statement to mean that the new platform will be "split," with a special new interface just for phones and a something different (perhaps more Palm-like) on PDAs.

I don't see any evidence at all that that is the case. Everything seems to point to the idea that MAX is designed to support both touchscreen and non-touchscreen devices so that there doesn't have to be a different look and feel between the two. We heard enough about the Rome project at last year's DevCon to know that its incarnation in MAX is going to be a fundamental rethink of usability in mobile devices, adapted to the new multi-tasking capabilities of the platform. So while ALP will include an emulator to run legacy Palm OS applications I think it's best to take PalmSource at their word that applications written against the preferred MAX API are not likely to look or work very much like Palm OS apps have in the past.

I think from ACCESS's standpoint, the inclusion of the GHost emulator ("G" for Palm OS Garnet) is intended as a bridge for the large user and developer base to start moving them to a new platform, rather than a statement of how great the Palm OS is or how important it is to their future plans.

That's a good thing--potentially, at least. Palm OS is 10 years old this month and was designed for dramatically simpler devices than what it's running on today. The original Pilot 1000 had only 128k of RAM, a 16MHz processor and no connectivity other than a cradle sync to files on a desktop PC. There was no file system (virtual or otherwise), no TCP/IP stack for networking, no wireless connectivity (not even infrared beaming) and for all intents and purposes no concept of it being more than a personal information manager. ALP is being developed for devices that can handle multiple simultaneous networking operations--telephony, cellular data, Bluetooth, WiFi, etc.--plus much of the rest of what people expect to do with a modern PC these days. And it has to somehow make this interaction simple enough that people who don't want a PC in their pocket will have a good user experience doing the things they do want to do with a mobile device.

The original "Zen of Palm" philosophy was that users would interact with the device the way they do their wristwatch: frequently, but only for brief moments. The thinking that was going into MAX as early as a year ago was that the system needs to be built from the ground up to enable users to deal smoothly with interruptions of sustained tasks when other tasks (like an incoming call or message) impose themselves. That's a very different kind of thinking, and if it didn't result in some significant changes from the old Palm OS I'd be pretty disappointed.