Frankfurt Thanksgiving ‘07
Hello There!
I’m a media darling!
Last week Friday Anne Baker, the VP of Marketing here at my new gig (ActionEngine in Bellevue, WA) asked me if I had downloaded and looked at the new Google Android software development kit. (Android is a new mobile phone operating system and is the presumptive basis for the oft-rumored gPhone.) Being the geek that I am, of course I had. “Great!” she then said, “Can you do an interview with the Seattle Times in about 30 minutes?”
To be sure, skimming the documentation in an SDK and being able to speak authoritatively on the subject (with a member of the press no less) are two wholly different matters. I tore back into the docs, while simultaneously watching a YouTube stream from the Google evangelist corp and rapidly jotting down notes of my impressions and interpretations of what I was digesting and then subsequently regurgitating. Anne and I then sat down with the reporter for about an hour’s worth of chat.
I knew the story would run in the Seattle Times business / technology section — but what I didn’t know was that the story would then be picked up by two blog outfits: one covering exclusively mobile content, the other an Android fanboy site. All three of them reference me by name, and quote me directly as saying:
Chris Lihosit, a senior creative technical architect at Action Engine, said one of the first things he did last Monday was download the developer kit, which encompasses all the tools needed to build an application. Although some of the features look great, he said, Action Engine is not interested.
“To be clear, I’m excited about firing it up and banging away to my heart’s content whenever I have a weekend to burn,” Lihosit said, “but I don’t know if it will be part of my job description.”
While the quote is indeed accurate, I *am* excited to play around with Android and I also suspect that it *won’t* be part of my job anytime soon — the article misrepresented the points I was trying to make. (This generally seems to be par for the course when giving interviews, which makes talking point memos seem all the more reasonable in hindsight.) The points I really laid out were:
Android Pros
- Android is built on the shoulders of some great open source technologies, like an OpenGL api, a great text rendering engine, the webKit backend of Safari and SQLite, a full relational database. All pretty cool, well thought out stuff.
- Android allows for multiple java process to run on the JVM, in either a synchronous or an asynch mode. This is big stuff to mobile developers; right now only 1 app can run at a time with mobile Java so things like soft switching and data sharing are right out the window for the majority of handsets in the global market.
- Android opens up a fairly universal access policy to a phone’s underlying capabilities — this is a massive improvement over somewhat spotty JSR implementations in the mobile industry today.
- The entire handset experience is apparently meant to be swappable. Don’t like the dialer? Get a new one. Want a new address book? Pop a better one in place. Looking for eyecandy on the idle screen? Here ya go. Only a few of these can be accomplished today mass market (Windows Mobile being the notable exception) and they all require certification and cash from gatekeeper organizations like Qualcomm.
Android Cons
- The Android release doesn’t do anything to solve the glaring problem in mobile: the ~6,000 device types used in the world, each with its own set of characteristics. Entire companies have been built upon device databases which contain profiled phone capabilities. With the replacement cycle of handsets being what it is the likelihood of this problem going away any time soon is nil.
- Implicit in the release are some baseline assumptions for development, such as a screen size that will provide enough real estate to provide a meaningful user experience. Volatile memory constraints, persistent file storage and guaranteed network access are all bottlenecked at the ODM fab. The goodness of Android can be spoiled by an accountant looking to shave $2 from the component price of a handset.
- Android has an XML based layout language, which looks to be similar to XAML in the Vista release. This approach has been tried before with mobile and has consistently failed. Exactly how an experience will gracefully degrade from a designed-for device of 320×240 to a run-on device of 172×240 isn’t particularly specified. And, *if* there isn’t a guarantee of great UX change, then Android is exactly the same as other mobile development languages. Which is to say, shitty.
- All the development for Android occurs in Java (horray!). However, it runs on a proprietary to Google (i.e., not open sourced) virtual machine called “Dalvik” (boo!). Dalvik, it turns out doesn’t actually run J2ME / MIDP2 … it runs J2SE or the desktop version of Java. (Note: Sun has been rumbling that J2SE and J2ME are converging, and therefore they might cut J2ME.) As a result of this code change, none of the existing mobile applications (I mean ZERO, NADA, REIN.) will work on an Android device — zero forward compatibility. Runtime file extensions are even different, instead of “.jar” they’re “.dex” (hurray for patent attorney billable hours!). Android starts with a clean slate … every ported app for Android is a baseline refactor and rewrite. That sucks primetime.
My real take on the Android release is that this is Google’s play to bring the power of their datacenters onto the mobile phone. The Android APIs integrate well with the online suite of Google Apps (Picasa, gMail and gMaps are all mentioned by name as application possibilities in the evangelist videos) so it’s clear that Google’s headed to the third screen. In a roundabout way, I see this also as Google’s entree to the embedded marketspace; there isn’t anything stopping Android from running on a set top box either. This incidentally is the space where Microsoft has long drooled as a future market for growth. Microsoft has met with very limited success in this market so far.
I think that Android will have great success. Well, let me qualify that for a second. I think Android will have great success in APAC and India, both places that are easily 2-3 years ahead of the US in terms of mobile adoption and utility. Domestically, the wireless carriers are all walled gardens the likes of which evoke the openness of CompuServe and QuantumLink in the 80’s. (This is to say, nearly zero.) I would love to see a Google evangelist convince one of the 8 brand management decision makers at AT&T, Sprint/Nextel, Verizon, or Alltel (yes, there are literally just a handful of these people in the US) that allowing any third party to remove their billion dollar investment in branding is a good idea. Off the bat, carriers would love their ability to form slotting and revenue share partnerships with service providers if anyone can build deep seated applications. Just try to steal that candy from the 800-pound gorilla baby.
As part of its launch, Google announced a $10 million bounty program (snarky comment: perhaps they should value the bounties in Euros soon). Motivating anyone with the carrot of shiny lucre is neato, but who’s really going to be going after that money? A first round prize of $25,000 isn’t nearly enough to bankroll an engineer for a significant length of time. (Back of envelope opportunity cost calculation: Average J2ME dev salary: ~$90k/year, loaded with benefits $135k or $67.50/hour. Doing double duty as both a designer, a developer and a tester, that leaves around 370 work hours or a little over 2 months to develop, test, release and win a prize before the breakeven line is passed. In the terms of a software development business, that’s a bit hard to swallow.) A final round prize of $275,000 would most likely come from an established mobile application house porting an application to Android making the money less of a prize and more of a de facto investment.
Adobe did this sort of thing back in 2002/2003 when it first released FlashLite (back then the new mobile killer platform). FlashLite still doesn’t have a meaningful presence in the US (well, it is on one phone from Verizon). Until there is some there “there” I see this as really being an academic exercise; Google is rewarding the geek elite to do what we would normally do anyway — hack around with cool new tech.