Fractals and design

So my first releasable app is already a month over my initial estimation. The Android learning curve is taking twists and turns, I am graphically challenged so it takes me longer to design a good icon, blah, blah, blah. Everything mentioned is true but what is interesting is that when it comes to design the challenges that need to be solved behave like fractals.

Or in other words in software design and in lesser extent electronics most problem domains are fractals. Design generally deals with fractals, for instance, to build a single engine one must build hundreds of components, that are built off hundreds of cogs and pieces, that have to have dozens of unique features. And once the engine is built it needs to go into a car/boat/plane composed of thousands of parts that they themselves represent millions of smaller problems each visible only once a designer focuses into the problem area.

Because of the fractal nature of the problem domains and the lack of serious standardization in design principles. software design is notorious for it's cost overruns, widening scope and date slips. In software a huge ten day design effort problem can be hiding behind every aspect of the project. A feature might be impossible to test without setting a ridiculous number of data by hand, a function takes too long to complete and concurrency at that stage requires redesign and so many more that volumes of books are not enough to list them all. Due to the number of unknown factors involved one could say that the model of solving a software engineering problem is chaotic and as such very difficult to correctly predict and estimate.

As software engineers we try to use our tools to accurately model and predict our future but when we actually look at our methods we generally look at the sky, lick our fingers and say hmm... it'll be OK. Only to find ourselves in a thunderstorm of epic proportions a week later. Unfortunately because business and money are also involved very often someone warns us of the coming storm but in an effort to keep our bosses, clients and investors happy we tape his/her mouth SHUT and throw them in the basement. Cause if you hope it will work out, it works out right? Ask any of sailors down at the bottom of the sea.

New methods are being developed all the time but rest assured that every tool in the world can be manipulated to predict good times when contracts full of money are at stake. I mean ask any construction company, they'll tell you: "Don't look at the sandy land we are developing on, look at the nice drawings instead and please give us money." Trust us, we won't go over budget before you are way too committed to pull out. And the world keeps on turning.

Till next time,
Stratos out.

Vertically integrated design in a 2.0 world

This is mostly inspired by the troubles faced by RIM and other companies.

Vertical integration in design used to be a good idea. The same company is in control of the architecture, the hardware, the software, the brand, the marketing, everything. Electronics produced this way rely on standardized interfaces to communicate with the outside world, everything is well contained and you as a company you control your products and subsequently your brand tightly.

It was such a good idea in the 70s and 80s but from the 90s onward the world started to accelerate. Design by comity and over-protectiveness slowed down innovation and created bottlenecks. To fight these, companies discovered outsourcing to speed up the mundane parts of design which at least allowed for features to have a quicker time to market. But the outsourcing solution is not enough in the 21st century world, time to market of 1-2 years is unacceptable in most industries if not all of them.

Mega projects are feeling the strain of trying to catch up with newer, leaner, more distributed designs and these growing pains seem to me to be the driving force behind agile methodology adoption by large corporations. If you have read some of the agile books out there the introductions sound like pitches aimed at alleviating the major problems faced by industrial mega projects. Elimination of bottlenecks via feature teams, knowledge transfer via team interactions, better control via transparency, etc, etc.

But agile has had mixed results. Change itself causes some initial slowdown in production or the fear of a slowdown. Management and employees don't easily embrace changes that undermine their job security which is a natural byproduct of some agile methods. Not to mention that often the wrong methodology is applied to the wrong project due to a silver bullet mentality.

And then came crowd-sourcing and the shit really hit the fan! The most affected industry of all is the mobile phone industry. The smartphone wars that have ensued are a testament of the changes going on in the world of electronics and software. Six companies are in the ring, one so far behind it is irrelevant, yes that's you HP, sorry. One already accepted defeat and hopes to band with competitors to survive, nice call Nokia. Microsoft, after having their crappy windows used to wipe the floor with, comes back with WP7 but is it too late? RIM is faced with mounting pressure to do something, market Blackberries to toddlers maybe? And the two kings of the ring right now, Apple and Google. Well Apple was the king to be precise before Google and their Android army started kicking their butt. But Apple, RIM, M$ and co have banded together in a last ditch effort to stop the green menace.

In the battle of Android vs everybody else it seems that Android cannot survive the constant barrage of patent trolling, leveraging the existing power in markets outside the mobile arena (WP7 and Xbox, iOS and AirPlay, MacBooks, iPad, iLife) and challenges posed by a newer and not as mature platform. In all this mess my money are still with Android. Why is that? It is not my natural tendency to side with David against Goliath, I mean Google is close to taking over the whole universe it isn't a David. It isn't even their motto "Don't be evil" that makes me like them so much.

Google will win because vertically integrated design in the 2.0 world is doomed to sink like a monolith. iPhone5s best features are already integrated into Android (smells like counter lawsuits by Google?) or are about to be topped by C2DM functionality. Simply put the practices of industrial secrecy, design by comity, content approval processes are no longer relevant in today's fast moving world. Time to market of over a month is too much at this age. Your only hope is to make your source open, attract as many individual designers as possible, give your platform to anyone who asks for it and let them change it as they please, in one word: Android.

Till next time,
Stratos out.