Google I/O 2016 Quick Thoughts
I totally forgot to transcribe my notes from Google I/O this year. All of my previous years are lost in the data migration gone bad from WordPress. Sorry. This will just be a high level set of my thoughts after seeing the keynote this year.
Google Assistant: Well it looks to me like Amazon Alexa or Google Now on steroids. Integration across all devices running Android. Add the new Google Home “speaker” device and you have a more generic version of the Amazon Echo or Tap. (I really like my Tap BTW.) The integration was impressive in the demos. We’ll have to see how it really plays out.
Allo and Duo: Now this part interested me. Google announced two products that may bring some ubiquity to messaging for the home user. Allo (a la Messages on iOS and Mac) is a messaging app tied to a phone number. Hmmm. How does that really play with their already existing XMPP stuff? Mobile only? Not sure I’m sold but it is interesting. I think it does bring the cross answer for IM on Android and iOS mobile… bring those users together. Same with Duo, their 1:1 video app (a la FaceTime on iOS / Mac). Great for that 1:1 conference and it is nice that it will be across platforms. Would like to see both of these tools on Windows devices as well as desktop OSes (Windows, Mac, Linux???).
Android N still doesn’t have a name. I was hoping for Nougat I like saying nougat. :-) Although Nutella would be pretty cool too.
It looks as though there won’t be major UI changes but a big focus on performance and batter life. That is a GREAT thing. Vulkan is the new N high performance game rendering “engine”. I’m guessing similar to Apple’s Metal. Tied with the new Just in Time (JIT) compiler they are claiming much better performance, live profiling to ART(again about performance), and smaller apps. Again, time will tell. Also they are touting that the Ahead of Time (AoT) compiler with the JIT improve runtime performance, memory, and storage requirements.
I’m not 100% clear on the implementation but it sounds good. One example cited was that, “profile-guided compilation lets ART manage the AOT/JIT compilation for each app according to its actual usage, as well as conditions on the device. For example, ART maintains a profile of each app’s hot methods and can precompile and cache those methods for best performance. It leaves other parts of the app uncompiled until they are actually used.”
There were a lot of other developer oriented changes that could be great if they come to fruition as planned. Accelerated Mobile Pages (AMP). Not sure of the real world impact of that. I haven’t researched it deeply yet. Will need to look into it sometime. See the Android N Developers Guide for more info: https://developer.android.com/preview/api-overview.html
On the security side some good improvements were announced like: file based encryption (Please!), hardening of the media framework, seamless updates, and deeper app security in the Google Play Store.
Unicode 9 emoji standard. OK. Good. Not a big feature really, but good to know for you messaging types out there.
VR on N was a big push as well. I’m not totally sold on that yet but it will be interesting to see what they can bring to bear.
Firebase. Now here is something to get super excited about. Realtime DB with built in mobile analytics (FREE AND UNLIMITED). Cloud messenger. Notifications. IAM. All right there. Cross-platform. Nice acquisition by Google / Alphabet. I started to play around with it and I have to say I’m pretty impressed, once I could get it working. Had a few hiccups along the way. Overall I think there is potential there. Will it be enough to compete with Azure and AWS? Once again I say, only time will tell. For the Android only community of developers, most likely a nice and easy to use. I’ll have to dive deeper to see if it has the richness and ease of integration of the offerings from Microsoft and Amazon. https://firebase.google.com/
Android Studio evolution. C++ support. Awesome. That rocks. g.co/androidstudio.
Android Instant Apps. Basically you could run an app right from the app store without fully installing it. Say you hit a link on LinkedIn that wants the app for some feature but you don’t want to fully install it (not sure why). Only the pieces of the app needed for the functionality you are trying to use are downloaded and installed. I guess for some mega apps (Salesforce?) it might be nice to have only the pieces you need. I’m not there yet but we’ll keep an eye on it.
Android Wear 2.0 looks to have some nice new features included watch only apps. That’s awesome. I will have to write something up about AW2 when I get my hands on it.
Multitasking. Oh yeah. My favorite announcement of the day. New multi-window options. Obviously both this implementation and Apple’s aren’t near what Microsoft has with the unified base of Windows 10 (which, is just Windows and can do all the cool things Windows can on many devices). When are we getting OS X on mobile? ;-) I digress. I like the implementation they are headed towards with N. Seems like it will work well on any form factor. I think this will be a major feature for the business community of multi-taskers.
Google Cloud… TPUs. Tensor Processing Units. I’m not expert in this area but it was intriguing enough to read a little more about. Basically Google had a shadow project to figure out how to get hardware that was optimized for machine learning. One explanation from the project team states, “ a custom ASIC [an integrated circuit built for a specific application / purpose] we built specifically for machine learning — and tailored for TensorFlow [Google’s open source machine learning library].” Supposedly this reduces computational precision resulting in the use of fewer transistors per operation. Interesting.
How that fairs in the huge machine learning space remains to be seen. The team stated they have been using this for some time with success. Google has access to a ton of data and they know how to get hidden gems out of our data (yes I said OUR data). Microsoft, IBM, Oracle, name your vendor. All are putting significant investment in machine learning. There is so much data already out there and more (that’s a LOT MORE) is being added every day. Gleaning patterns for making decisions out of that data is critical. This is a very exciting area in data science. I’m looking forward to having some more time to dive deeper into ML and what these vendors are bringing to market now and on their roadmaps.
Oh well… that was a dump of what I had written down from the keynote. Hopefully it is a decent enough summary that someone out there gets some value from it.
Have questions or comments? Let me know. ~ Vince
