Systematic Solutions that Crack Under Pressure
- What happens to a bridge in an earth quake? It shatters.
- What happens when a massive surge hits a dam? It cracks.
- What happens to your small web site under a massive DDOS attack? It folds.
- What happens when your smartphone gets hit by an EM pulse? It fuses into a brick.
It's usually not the fault of the designer. Abnormal circumstances can rapidly cause a functioning system design to fail.
At first we construct solutions that are simple and secure enough to thrive in ordinary conditions (benign weather, light traffic, etc). But over time security demands that systems should survive an endlessly growing number of hypothetical, extraordinary conditions while gracefully degrading in performance. Those types of designs and constructions are heavily specialized, and far from cheap.
Engineers of both hardware and software are faced with a pretty ridiculous task. Designing systems that are near invincible to a diverse set of unknown threats. It's a fools errand. By definition, non-adaptive systems will only be able to function under limited specific toxic environmental elements until they fail.
Cost Driven Solutions
The solution is to drive the system design in a direction that is cost driven. If the cost per unit in a massive system is made cheap enough, even high failure rates due to expensive to endure problems are manageable. Each unit in a network should be simple and durable to local conditions. The ubiquitous nature of network power should be leveraged. Instead of a single hyper expensive system the solution is to rely on millions of cheap "dumb" units.
There are exceptions and other design limitations but consider the following choices as examples.
- Scalable remote computing resources are incredibly cheap and getting cheaper. Why optimize a software package for a 5-10% improvement over a few year lifetime, when you can scale the hardware much cheaper. Although dependant on software lifespan, most application specific software can be phased out and replaced instead of expensively maintained with unnacceptable unit costs.
- Disposable/Recyclable Smartphones that cost pennies per unit to create. Tethered to solve mobile Internet even in remote areas. Identity and other important data are stored in cloud datacenters. Local resources are limited but high functionality is available at each node
- Massively parallel Pebble Bed Reactors or Micro Reactors for remote power. Imagine miniscule versions of reactors distributed everywhere.
The trend towards system designs with many units, simple, independent, & cheap (MUSIC) is just one cost driven solution to solve the creeping and unknown failure problem.
In the future we may leverage massive network solutions to build grandiose structures. Imagine a huge bridge that forms from many miniscule independent bots. Just a spin on the draw bridge :).