An excellent article on the future of smartphones puts hard numbers to a trend I’ve been watching for two years. In so doing, it points out one of the fundamental competitive drivers in the smartphone market. More than that, it displays a powerful if sometimes less than obvious advantage of open-source software, and implicitly relates that to a seminal concept in military theory called the OODA (observe/orient/decide/act) loop.
The single most interesting feature of the article is a graph of how the average lifetime of Android handset products has been falling over time; 3 years as recently as 2007, it’s now down to 6-9 months. Unstated in the article is that to sustain that pace, development cycles actually have to be shorter than that. If you look at the pace of announcements from (for example) HTC, it’s clear the development time of the more nimble Android players is actually down to approaching 90 days.
The article also talks about the “Quadroid” standard, how intense competition in handset design has been fueled by a combination of inexpensive phone chipsets from Qualcomm and the Android OS. The effect of this combination is to drastically reduce both engineering expenses and time to market. It is, fundamentally, why development cycles can drop to around 90 days and product lifetimes to 6-9 months.
There was a fighter pilot named John Boyd who became the most important strategic theorist writing in English during the 20th century. He began with E/M (energy-maneuverability theory), which became the basis on which modern fighter aircraft are designed and modern fighter tactics taught. He ended his career as one of the architects of the winning “left-hook” strategy in the 1991 Gulf War. Connecting these was a general theory of military effectiveness (and more generally, organizational effectiveness) centered around what he called the OODA loop.
OODA theory is worth reading about in more depth for anyone interested in…well, any kind of competitive dynamics, actually – military, commercial, individual, anything. The Wikipedia article is a good start. Stripped to its essentials, the theory is that competing entities have to go through repeated iterations of observing conditions, relating observation to a generative theory, deciding what option to pursue, and acting. Victory tends to go to the competitor who can run this cycle the fastest.
OODA theory was originally generalized from the observation that in fighter design, maneuverability (especially a shorter turning radius) beats straight-line speed. When you get inside your opponent’s OODA loop, either physically or conceptually, you can attack him from unexpected angles. You can be where he isn’t. You have the initiative; he is reduced to reacting, often with too little and too late.
Now let’s look at the smartphone market, and consider Android vs. single vendors like Apple, RIM, or (now) NoWin (Microsoft/Nokia). Your single vendor has a product development cycle on the close order of a year – exemplified by the fact that the NoWin alliance announced just a week ago won’t commit to shipping a phone before 2012. Apple iPhone releases have been issuing on a predictable once-a-year schedule.
The Android army, and some of its individual members, now has a development cycle approaching 90 days. It gets three or four cycles around the OODA loop to each one executed by Apple, RIM, or NoWin. This is, in particular, why the Android army has been winning the race to exploit new networking standards such as HSDPA and LTE. More generally, it means the Android army can exploit technology changes and the availability of new components such as SoCs (System-On-Chip ASICs) at a pace single vendors can’t match. Most generally, it means the Android army has a faster reaction to consumer demand.
No OODA theorist would be surprised at the result; Android is clobbering the crap out of its less nimble competition – not just on price but on features as well. If you know OODA theory, you would predict this result as soon as it became clear that Android’s OODA loop was going to be shorter.
Of course the Android army has had other things going for it as well. Lower engineering costs is an obvious one. Greater collective financial mass than any single vendor, implying the ability to spend more on development and marketing. All these tie together with tighter OODA in a way that has open-source software at its base. Open-source software is the key.
Because Android is open source, developing and shipping handsets becomes less expensive. Because Android is open source, handset makers and carriers can join the Android army confident that they’re not setting themselves up for extortion by Google or any other single software source. Because Android is open source, Qualcomm and Nvidia can front the enormous investment to make SoCs specialized for Android in confidence that they’ll have a market of more than one customer.
All these drivers increase competition and shorten time to market; Android’s OODA loop tightens, while Apple and RIM and NoWin stumble around trying to respond and fighting last year’s war against products designed last quarter. And come up short, as in Apple’s anticlimactic iPhone 4V launch last week.
There is no possibility that the single vendors can win such a battle; Android’s OODA will inexorably tighten into a noose around their necks. I said this confidently in late 2008 when Android first shipped, and it wasn’t wishful thinking; the consequences of open source were as obvious then as they are now to anyone who was paying attention.
I’m repeating the point now only partly as a study in the competitive dynamics of the smartphone market. Really, there’s nothing special about smartphones in this respect; the general effect of open source tightening the OODA loop operates elsewhere as well, with predictable consequences. We actually should have learned this a quarter century ago from the way that TCP/IP surpassed and almost completely destroyed proprietary networking protocols.
The story doesn’t end in 2011. Android SoCs aren’t generally deployed yet. When they start shipping later this year in third-generation Android phones, the parts count on a minimal smartphone is going to drop to a single chip, a capacitive display, a speaker/mic, a couple of microswitches, and the PCB to mount them on. Qualcomm is already predicting retail unit costs of $75 or less, and it’s going to be less once the chip development costs are amortized out.
The key to this possibility is that most of what used to be hardware costs in a smartphone have been ephemeralized – converted into information complexity inside the SoC. The combination of ephemeralization and open source is the fundamental of the fundamentals here. (Not entirely by coincidence, every single one of the billions of these Androids will have substantial pieces of my software inside it.)
It’s anybody’s guess what the lower limit of the OODA loop is. Not so long ago, 90-day product development cycles would have been considered an insane dream – but the rise of 3D printing and commodity SoCs might very well cut time to prototype much further. I can now easily imagine a handset designer with good CAD tools and a 3D printer producing a testable phone in a week.
What is certain is that single vendors going it alone will not be able to match this pace. They won’t have the SoCs and they won’t be able to match the number of engineer-hours going into the software. Their development and production costs will be higher. Consequently, they’ll be more risk-averse and unable to make decisions on a fast turnaround even if their technology permitted it; their OODA loop will never be able to tighten as much as Android’s.
All these problems – all of them – are predictable consequences of having a business plan stuck to a huge heavy blob of closed-source software. That is the weight that will ultimately drag down and kill Android’s competitors.