We are now beginning to understand computational systems as living systems. This living computation framework helps us see that digital computers over the last half century have succeeded despite, and sometimes due to, major gaps in their fundamental designs compared to most living systems. Over the coming decade, the IOU's on those design lapses---security, robust operation, self-administration---will start coming due in earnest wherever computers are networked, from the internet to financial markets to defense, and probably to the living room and kitchen as well.
Budgetary constraints notwithstanding, increasing defense reliance on conventional existing COTS systems will be an increasingly unwise proposition. Although, eventually, COTS software will become more robust and secure, evolutionary theory suggests that process will tend to be disaster-driven---and given the leverage implied by large-scale system deployments, the scope of such disasters may be significant.
Policy goals in this context can aim at mitigation as well as encouraging anticipation rather than merely reaction, both in research and in COTS development and deployment. Focused research funding can help, as can equipment purchasing and insurance contracting decisions, even if reform of software liability law is deemed unwise or unachievable. Technical goals span computation and communication, from robust self-healing hardware, to opportunistic and paranoid networking protocols that are sensitive to social networks, to self-aware diversifying operating systems and software systems that put loyalty to their human---whether couch potato or forward fighter---first and foremost.
In the 1990's we started to understand how to build larger, more responsive, and more evolvable software systems. Increasingly urgent research and policy questions for computation in the 2000's ask us how we should handle the living systems we have built, and what kind of life should we build next?