Tuesday, October 19, 2010

Depth-First vs. Breadth-First Discovery of the Future

Steve Job's rant about Android and the Google model (http://techcrunch.com/2010/10/18/steve-jobs-android-audio/) is fabulously thought-provoking - not to be outdone by Andy Rubin's hysterical response (http://twitter.com/#!/Arubin/status/27808662429).

On one hand, you have the Internet-bred nerds that claim that an open model is fundamentally superior & will eventually win. On the other hand, you have Steve who  convincingly articulates that open vs. closed isn't what matters - it's about fragmented vs. integrated. He claims (and recent history does support him) that the integrated approach is more effective at delivering the best innovations to consumers.

Here is a proposed thought exercise. Imagine that progress is a 'search' for the discovery of continually better products for consumers. Think of it like the traversal of a tree - the root is today and branching from today are a multitude of options for how to design the future. Each decision is an edge on this tree. Choose your level of detail, but imagine that each potential product decision is represented in this tree. So, in the Apple-style integrated approach, you have a consolidated decision making process that decisively selects a *single* path - say ten steps down the tree, where they bet that humanity will elect to take its future. If you have a phenomenally visionary leader such as Steve, he is able to drive his part of the ecosystem to select a *single* path and more often than not, this path is a successful one. On the other hand, you have the open / fragmented approach, that Google is taking with Android, where the leader/organization takes only five steps decisively, but then lets the ecosystem try out a bunch of branches.

>> Fundamentally, the integrated approach is a depth-first search for the future, whereas the open approach is a breadth-first search for the future. 

The integrated approach plays two important roles
1) Unadulterated Expression of a Vision: Once in a very rare while, a true visionary comes along. These people are not just creative. They don't just see glimpses of the future. They are able to draw a complete self-consistent view of what the future will look like. I feel that many who try create the future (myself included) will have insights and ideas about how specific items will evolve. We might have a fantastic idea about how the spoon of the future will look - and the spoon we imagine is in a kitchen as we know it today; in a house as we know it today. But it's a damn nice spoon. Then, once in a while, you get someone like Steve who comes along, who is able to truly shed the shackles of the present. He tries to imagine the *entire* house of the future - unencumbered by the house of the present. He is able to take system-level evolutions and apply them in ways no one else would have even considered. This, in my opinion is one of the key roles of the integrated approach. It enables the selection of a specific path down the tree of innovation - unencumbered by constraints of the present.

2) Clear High Activation Energy Hurdles: In some cases, it takes a massive company to clear the activation energy required to open up a whole avenue for the future. For example, it took Apple Computer to get AT&T to finally open up the mobile market. All of us who'd spent years fighting to deal with the shit-show that was WAP and trying to convince carriers to give us space on their "home decks" had a quasi-religious experience the first time we interacted with the iPhone. It took Steve Jobs and the entire weight of Apple Computer to change the laws of physics in a world where any innovation immediately fell into the carrier-black-hole. There is no way that an Android-style open "let's throw a party and invite all of our friends to hack on your network" would have ever worked with the carriers (in addition to them losing their marbles everytime Google was mentioned). It took Steve's "my way or the highway" style, as well as the realization that Apple would easily be the best bet for someone to tightly control their environment (can you imagine Google or MSFT trying to approve apps?).

For the company (Apple) following the integrated approach, there are two key advantages
1) Resource efficiency: Overall, the effort needed to end up at node x down the tree is lower than waiting for an open model to breath-first its way to that same node. It is also arguable that the open model is perhaps likely to never even end up at some ideal/stable node, given the baggage it carries from all of the wrong paths that were explored. When mistakes are made, it is often impossible to fully eradicate their impact on the system (I bet you can still find some code related to Clippy in Windows :). So, the integrated model gives you a pure-bred creature, no scars from past battles, no fat from past excesses. Lean mean future-discovering machine.

2) Head start: Moreover, the integrated approach gets Apple there earlier than everyone else. If Apple's bet is correct, being in the future for a year ahead of the others allows them to benefit from the image of being an innovator (which they are), they have more time to learn and continue running ahead, and finally, they can build a protective mote to block others from getting to that point through market share, filing patents, etc...

But eventually, the open model wins
Now that I've spent all this time discussing how awesome it is to be integrated and drive with a singular vision, I will still contend that the open approach will eventually win. I would argue that it is a question of timing and what phase of an innovation cycle we are talking about. Once in a while, you need massively discontinuous steps for the future to happen (ex. introduction of the iPhone). However, once a new era begins, consumers at large will have a sufficiently diverse set of interests that no singular vision is optimal. Steve Jobs designed the house of the future, but it only comes in white and no fridge or hot water. Now, come into the picture all the roughnecks from Google and their contractors (HTC, Motorola, etc...), who give you tons of the great features they copied from Steve's new house - in any color and also allow you to build the shed however you like. Oh, and you can get a solar heater for free and it comes with a Froyo machine integrated!

< rant >
I do have to mention that all players involved tend to be disingenuous about their definition of openness and how they relate to it. On one hand, you have Google that claims full openness, but conveniently is decisive at key junctures. I do think that Google is the best search engine, but I can't remember seeing another search engine being offered when I was activating my Android (not that I would have chose anything else - just sayin'). On the other hand, you have Steve railing against the open model, whereas the entire empire he's built over the last decade is based on the OpenBSD operating system (crap! it even has open in the name!) and WebKit - two important open source project.
< / rant >

My point here is that the integrated approach does enable the discontinuous innovations. However, these system-level steps are not needed every year. In fact, they are most often needed to overcome the cruft that has been laid by past innovators that are still trying to hold on to their crumbling empire. Once the integrated assault opens the gates, the breadth-first exploration becomes a more efficient mechanism to discover successful paths to the future (for the current cycle). To see this clearly, compare an iPhone 3G with an iPhone 4. Then, compare last year's Android to the current versions (both hardware and software). While Steve has a staggering hit rate in terms of envisioning where the future lies, his vision is a singular one that has no way to keep up with the diversity of human interests and needs. I'm typing this on a MacBook Pro, but the switch from an iPhone to and Android 24 hours ago has so far been the most satisfying encounter I've had with technology this year (the iPad is a close second).


Vishal Sood said...

Great article. I do agree with your thoughts. It resonates very similar to the book I am reading by Clayton 'The innovators solution', where he talks about the integrated vs modular approach. And how the modular approach wins mostly.

But in this case don't you think that by the time modular approach (open model) wins, Apple is moving on to something next i.e. iPad. It has already sold close 5M of them. And who knows what they shall be coming up with next as the true innovator in the industry?

tyg said...

Also seems to me that the approach reflects each company's computing culture. Apple's computing power is based on individual machines (admittedly these days with multicore processors). Google's is based on lots and lots of machines working together.

Brandon Badger said...
This comment has been removed by the author.
Brandon Badger said...

There is no perfect phone. There are only perfect phones.


othman said...


That's certainly a valid point. I wonder if two questions to consider are (1) How realistic it is it to keep up a cadence of truly discontinuous innovations. While the iPad is certainly innovative, it's a different ballgame than the impact the iPhone had on the mobile market. So, wondering whether - once the big leap has happened - whether iPad-scale innovations are sufficient to provide a sustained advantage. (2) How effective is the integrated approach at making products that truly go mainstream - relative to an open/fragmented model. Both have advantages, but Android's competition with the iPhone seems to indicate at least that the jury is still out...