The Android Paradox and Computing Inequality

Benedict Evans wrote a great article on android fragmentation. He and I have covered this theme a variety of ways but I wanted to add a few more elements for you to think about. I also tweeted over the weekend on this theme and faced fire from the twittersphere.

It is a paradox that one can do so much with Android’s “open” OS yet the majority of its users do less than the “closed” iPhone’s iOS.

— Ben Bajarin (@BenBajarin) May 17, 2014

Furthermore, Benedict makes a point in his post key to understanding the Android paradox:

Again, this is a paradox: Android is the platform best for early adopters and iOS the one best for late adopters who just want something that works, but the market adoption is the other way around.

There is no question Android is a unique beast. I’ve long heard Android compared to Windows. However, I have never been comfortable with this comparison. There are certainly some fundamental similarities but there are also a great deal of fundamental differences. They are similar in that they are both software platforms available for third parties. OEMs can take the software platform and create their own hardware. While those are basic similarities, Android has a fragmentation problem Microsoft never really had.

In the PC era, Microsoft maintained very strict hardware constraints for products running Windows. Intel, and AMD to a degree, also assisted with this a great deal to make sure OEMs had a certain bar that was maintained in terms of computing experience. This came crashing down with Windows Vista. This was an example of Microsoft having too computationally complex a piece of software which made machines with underpowered CPUs, and more importantly weak graphics capabilities, have many issues with Vista. I remember telling the OEMs around the time if everyone just shipped a discreet GPU on all their Vista machines it would be fine. But that would have driven costs up at a time they were all trying to drive costs down. For the most part, we have not had the same fragmentation issues with Windows we have with Android. Curiously, we saw the beginnings of the issue with Netbooks. For the first few years, these devices ran underpowered CPUs which consumers then attempted to use to do things they would do with more powerful computers. This is why early Netbooks didn’t play flash video well, or games, or other CPU intensive tasks. It brings us down an interesting thought trail of where we are today with computing power in lower cost devices. Which leads us to the Android paradox.

Where Windows seemingly was always designed to run computationally capable silicon, Android is and must be designed for the lowest common denominator. Android must run on an extremely low end CPU and an extremely high end CPU. This problem is outlined well in this video and in this post by Game Oven.

The point the folks at Game Oven highlight is the challenge of building a computationally complex piece of software and getting it to run on every Android device in the world. This is a significant problem for the future of computing and it begs a fascinating question. Architecturally speaking, is ARM or x86 better suited to address this particular issue? I do not have this answer yet, but it is a key question.

With my background in semiconductors, I look at this problem and doubt Google can solve it. Google can not maintain control of the hardware in smartphones the same way Microsoft could with Windows PCs. Mostly this has to do with the fact Android OEMs want to make a phone that can be sold for $100 dollars or less. Those devices, by sheer economics, will have to use an inexpensive and low powered CPU. That vendor will also have to make decisions on which sensors or other chipsets to include or not include in order to hit that price. Which means software developers like Game Oven simply can’t run their software on those devices. It emphasizes the point that a software developer Android addressable market is limited by hardware if they are looking to push the envelope of computing.

Going further down this rabbit hole gets even more interesting for the future of computing. With roughly 80% of the Android install base being lower end, underpowered devices, and even more so as we add another billion plus first time computer owners with a smartphone costing less that $150, we have an issue of exposing those first time owners to the full potential of handheld computing. Perhaps that isn’t necessary, since it is their first computer, but it is from the standpoint of personal computing.

However, this may only be a short term problem. When you look at silicon road maps and Moore’s Law, it seems it’s possible that, 4-5 years from now, some incredibly powerful CPUs will be able to run in devices that can be extremely low cost. That being said, as long as Moore’s law exists, software developers will hopefully exploit the new capabilities of latest generation silicon and push software to the limits. Until Moore’s law is passed, conceptually, there will always be an inequality gap with regard to computing.

Intel may have something to offer to solve this problem, as x86 brings many efficiencies to the table for complex operations. Should they get to a very low power yet high performance benchmark over the next few years and gain traction in smartphones, I will be curious to see if the fragmentation issue around what Game Oven highlights could be solved through an x85 Android environment. We will have to wait to see, but I am still skeptical fragmentation can be solved simply by all computing devices running the same application processor architecture.

Remarkably, Apple is one of the few companies who can solve computing’s inequality gap. Should they be able to bring their experience to hardware at lower costs, it could be a huge benefit to the future of computing. Until then, we remain firmly set in an environment where the capabilities of hand held computing are unequal.