(this is page 38 of 121)





Latest

Despite people complaining that Apple didn't introduce cold fusion this time around (talk about winner's curse), I found my brief attendance at WWDC this year well worth it, and came away with a better understanding of how what Apple is doing with the iPhone serves as a harbinger of a trend that will likely redefine all of the consumer electronics industry.

I started the day by attending the "iPhone OS networking" session which laid out all of the incremental improvements that Apple has made to the TCP/IP routing algorithms in 3.0 so as to allow a seamless experience between the various radios in the device. Now the iPhone is not the first device I've had with both a cell modem and a Wi-fi radio, but it certainly was the first that attempted (and mostly succeeded) in doing the right thing switching between one and the other during truly mobile scenarios.

And it looks like this is about to get even better. Publicly announced was the whole "captive network" thing around hotspots at public places (think AT&T wifi at Starbucks) that require authentication which is likely to be a huge boon to folks that use these networks and have thus been trapped in authentication hell while walking by a Starbucks. But there are many other problems that follow from chatty network protocols that are suddenly hoisted from one IP address (the WAN) to another (the Wifi LAN). It's clear that the Apple team has thought through this (read the stupid NDAed docs to see how), especially along the theme of the entire conference: "refinement."

While there were many great sessions and meetings in between, my day was bookended with another networky session: an introduction to developing with the Apple Push Notification Service (APNS). Thanks to the stupid NDA again, I'm going to just make two high level observations about APNS: first, it smells of a highly efficient implementation that was likely the result of Apple still smarting from last year's MobileMe fiasco, which is to say, binary protocols are cool again.

Second, the implication of the implementation is that each of the 40MM iPhone devices out there is going to maintain a persistent TCP connection to an Apple service at all times. Seen in this light, the investment in the network stack refinements from the morning makes even more sense.

But more significantly, think about what this means for the consumer electronics industry as a whole. As it sometimes does, Apple is showing us what I think will become an increasingly common design pattern: devices that are "phoning home" via the Internet to their makers's servers in order to receive/send new information that is relevant to their overall use. This is no different than what we saw with last week's Air France's speed sensors reporting realtime telemetry back to Airbus— but instead of talking about a $35MM plane, we are going to start seeing this in all sorts of low cost consumer electronics devices.

Could this substantiate the rumors about Apple building a billion dollar data center in North Carolina? Maintaining persistent connections to that many devices alone certainly wouldn't require that level of investment— after all, large IM networks have done this for years— but it might if APNS is just the beginning and in time Apple hopes to use the connection for much richer forms of data exchange.

Two last points on this: first, every other device manufacturer is now going to have to get their heads wrapped around this type of deployment model— something which may be quite painful to the folks that have spent their entire careers "shipping boxes" and building business models that stop when the customer pays you for that box.

And second, it is interesting to see that APNS is being offered by Apple and not by the wireless carriers in each of the countries the iPhone is sold. It has got to be obvious to AT&T that APNS is the beginning of the end for SMS charges for iPhone customers, and more importantly, the continuation of something much much better (if you are an enduser): Apple wresting control of the non commodity parts of the pipe from carriers who are more addicted to their own business models than pleasing users.

Let the era of cloud consumer electronics begin!

Comments Permalink

It may be a sign of the times, but the most striking feature of the 2009 Apple Worldwide Developers Conference is that it would be hard to assemble this many people that are that familiar with all of the different layers of the computing stack under one roof were it any other company running this conference.

And it shows. From the low level tools like LLVM and their own compiler front end and the Open CL stuff to make a general purpose computing platform out of the gates on the GPU to the high level improvements to Finder and the native apps, these guys seem to have it all covered.

Snow Leopard and iPhone OS apparently share 85% of the same codebase which is a bit nutty given that the former runs basically super computers and the latter is made to fit in pockets. Grand Central provides new ways to enable multicore threads-based programming while Bonjour proxies allow peripherals on your home network to remain asleep for the 99% of the time they are not in use. Lithium polymer is the new black when it comes to integrated laptop battery technologies (that are supposedly greener) while the ARM 7 instruction set is the new jacks-to-enter for mobile instruction sets. The XCode debugger now can step backwards while at the same time allowing source level variable inspection. And to boot we're also getting 1,000 new improvements to the speed, stability, and look & feel of the applications we've come to depend on thanks to this new build toolset.

I'm droning on to make the point that the sheer breath of stack traversal from silicon to tooling to enduser applications and even web services that power new forms of micropayments is all being covered here. By one computer company— albeit one that stands on the shoulders of open source and the power of the Internet.

And while I'm on that topic, the new iPhone 3GS looks fantastic— a worthwhile upgrade for anyone who has come to depend on it as a key piece of their personal technology. AT&T is going to make no bones of the fact that this is an honest-to-goodness $700 mobile computer they are arming you with, which is why this is the new unlocked price of the high end unit— or $500 for the folks who rolled their hardware last year for the iPhone 3G and who don't want to wait until December to upgrade.

But you know what? This is actually not a bad thing. Subsidized business models are evil and it's time we consumers put a stop to them. They misalign the interests of the end users, the sales channel, and the folks who slave all day to make the products. I've seen it first hand in the printer industry, and I imagine that it's not much better when it comes to nickel-and-diming wireless carriers with their roach motel plans and inability to execute on behalf of the user.

So before folks start complaining, let's stop to consider that these are real honest-to-goodness computers being put in our pockets by what is perhaps the world's last standing Real Computer Company.

Comments Permalink

Google's giveaway of the "G2" (officially called the Google Ion), preloaded with Android 1.5, was a generous and thoughtful way to engage developers in building apps for the Android platform. And in any world that didn't have the juggernaut that is the iPhone and the AppStore, it may have worked.

Sadly though the overall Android experience still lacks polish and leaves something to be desired, especially in the realm of small devices that come out of a pocket and need to be usable for short bursts. The software is coming along nicely, and the soft keyboard is a welcome addition, but either the OS's broadcasting touch events, the capacitive screen driver, or the hardware itself is weak because most times you get going with the new keyboard and hit the space bar it fails to register, as well as random slowdowns in event processing on any user action (I blame the JVM's garbage collector and not as people have written, the multiple apps in the background which is perhaps the best thing about Android as opposed to the iPhone). Furthermore, the "notifications" system where you have to hit a terrifically small area on the top toolbar to see what's happened as well as the "app tab" metaphor both seem like they were not designed at all—or at very least, not polished enough.

But alas, Android is the open option so I'll keep rooting for it and using every new release. In the meanwhile I hope that the Android-on-post-PC-devices push yields some interesting new alternatives in a world less defined by the iPhone (even Palm's Pre is getting compared primarily to the iPhone on its launch weekend).

Comments Permalink

Making Sausage at Google I/O

Spanish chorizo is among my favorite foods in the world. And this is despite the fact that during my last trip to Spain I visited someone who showed me how it is made. And trust me, if you are a fan of chorizo, this is one sausage you just don't want to see getting made.

I felt similarly this week while attending talks at at Google's developer conference, I/O, and specifically those around the company's hosted web platform, AppEngine, and the ones around its browser-based UI toolkit GWT. In neither set of talks did Google engineers seek to sugarcoat the fact that working on a performant web application is messy and ugly, especially when you've got scaling ambitions similar to Google's.

Most of the talks I saw on either topic were worth watching (as they come out on video), and especially all of the ones titled "advanced X." During those I felt that Google was doing its best to give webapp developers a good best practices cheat sheet that will save you lots and lots of time.

What is good for the web is good for Google, and nowhere was this more obvious than at I/O this week. Unlike other developer conferences where vendors line up to tell you how everything is as simple as click-click-poof!, the Google folks seem to take great pleasure in marinading in the web's complex, organic, and most importantly, open nature. All of the technologies showcased seem to take the long view which implies a sort of patience that probably only Google could muster. Betting on HTML 5? Letting Android cook slowly while everyone declares it "dead?" These are all good things for all of us.

Oh and did I mention that they gave us all Ion phones (G2 Android)? 5,000 phones hoping that a few result in killer apps. Now that is investing in the future.

Next up, my first impressions of Wave (having digested it for a few days).

Comments Permalink

For a few months now, one of my favorite weekend activities has been to spend some time at Apple's AppStore, looking around for cool new apps to put in my pocket. Lately though I suspect I am not alone in feeling "AppStore fatigue" by which I mean that it is beginning to sense that most new apps are not that interesting, or in the best of cases, little more than a native interface for something that should have probably been left as an iPhone-optimized website (casual games at the AppStore are the one exception).

I've heard the argument that this sameness is being driven by the limitations of Apple's sandbox, and that with the coming of OS 3.0, we're going to see a whole bunch of new innovation. Outside of the rumored video recording capabilities, beefier hardware, and Bonjour over Bluetooth though, I'm not sure that this is right.

In fact, I'd be willing to bet that until true multi-tasking comes to the iPhone platform, as well as multiple hardware form factors, it's going to fail to reach the true potential of the kind of pocket computer that could kill PC use for most regular folks.

Why? Because it takes too long to get to most tasks when what you've got is a pull-based model for information that is relevant to a particular context. Whenever a use case is just a bit more complex than getting a restaurant review nearby, checking your email, or catching up on the latest headlines, the iPhone's obligatory 15 second time-to-info is fast becoming a deal breaker (how quickly we become spoiled!). And forget about actually authoring content, not so much because of the time it takes to get started, but because any type of rich asset upload without background processing is just an exercise in watching the grass grow.

Having upgraded my G1 to "Cupcake" (Android 1.5), I've recently had a chance to rediscover the importance of multi-tasking. Cupcake should really be called Android 1.0 (because it is what they should have shipped) or 2.0 (because it is so much better than what they did ship). Overall the OS is much snappier, and apps seem to crash much less when they are backgrounded. The battery still gets crushed but fetching data in the background makes a huge difference and automatic YouTube integration is one of those technology-as-magic things worthy of seeing.

Interestingly enough, the Android team had its own host of issues to solve with the sub-60 second use case that had more to do with the G1 hardware than with software. Because the G1 has a Sidekick-like slide-out keyboard, most entry-based tasks required two hands and an awful context switch (having the screen rotate and hoping that app elements stay more or less in place). With Cupcake's new soft keyboard (which is just as good as the iPhone's, many tasks can now be performed one-handed and much more quickly.

Finally, I hate to admit this, but my recent experience with Android and the G1 is making me sort of excited for the Palm Pre which— at least of paper— seems to address both the software (a truly modern muli-tasking OS built for a mobile phone) and the hardware (a physical keyboard that is aligned with the most common phone orientation, plus a willingness to vary the form factor model to model) challenges that handicap the current crop of pocket computers. On paper though, it is easy for anything to be better than reality, so we'll have to wait until June 6 to see!

Comments Permalink