Here are the notes for my talk at ToorCon2009 on 24 October 2009. At the moment, this file is also at
http://rohan.sdsu.edu/faculty/vinge/toorcon2009/index.htm

I don't know how far computational ubiquity can be extended, but here is a fairly conservative trajectory, starting in the 1980s:

• Embedded systems, becoming:
• Networked embedded systems, becoming:
• Fine-grained distribution
  • Smart Dust and MEMS
    http://robotics.eecs.berkeley.edu/~pister/SmartDust/
  • Localizers -- location knowledge from distributed computation on node-to-node latencies
  • Ad hoc networks
    See Tim Shepard thesis: Decentralized Channel Management in Scalable Multihop Spread-Spectrum Packet Radio Networks
• Most physical objects know where they are, what they are, and can (in principle) network with any other object. Those objects with effectors can modify their surroundings.
  • Warneke, Last, and Pister, IEEE Computer, January 2001, p44-51: "We will program the walls and the furniture, and some day even the insects and the dust."
  • Ultimately, the greatest hardware problems may be similar to problems faced in biology:
    • Power for high duty-cycle nodes
    • Cleaning up node guano (dead communication infrastructure nodes)

• Wearables: (almost) the ultimate interface
  • The closest system outside ourselves
  • The minimum responsibility of a wearable system:
    • Act as communication endpoint
    • Provide trustability (we hope, since serious failures here render the rest of security a charade).
• Importance of wearable I/O devices
  • Augmented Reality
  • Multiple consensual virtual enironments are possible, each oriented to the information needs of its constituency. If we also have open standards, then bottom-up social networks and even bottom-up advertising become possible.
    This is essentially the situation in my novel Rainbows End

• Reality becomes its own database
• Most physical world outcomes are the product of software running on top of the uniform (network/localizer/sensor/effector) substrate.
• Most inventions and new capabilities become the common property of all players -- and very quickly.
• Most physical failures (and physical attacks) are -- at least at the beginning -- software and computer and communication failures.
  • "Our flight wouldn't take off; the software balked." This could be a bug in the software -- or it could be a rather mellow alternative to the airplane really going down in flames.
  • Identity theft instead of muggings.
  • Software stabilized systems replace intrinsically stable systems.
  • http://xkcd.com/c149.html and beyond.
  • % kill -9 -u user ...
    

Civilization faces orders of magnitude more threat vectors than in the past. The people in the this room may be among the few who have an idea of how great the variety may be.

Coping with this new environment will require some serious attitude adjustments:

• Do not attempt to legislate reality! Avoid the seductive attractions of SHE
• Do not install national emergency network shutdown (or takeover) mechanisms. Such mechanisms simply provide another attack vector for your enemies.
• When it comes to national infrastructure, there is no backdoor that is your friend.
• Work publicly with all, to find and fix software security issues, and provide safe, secure, and open infrastructure.

• Work with all, to find and fix software security issues, and provide safe, secure, and open infrastructure.
• Support heterogeneous solutions at all layers.
• Bless those hobbiests who keep alive the old technologies that don't need semi-conductors. (Worst case, we'll really need those folks!)

It seems likely that (absent physical disaster) the trend toward computational ubiquity is unstoppable, and is leading to complexity comparable to the complexity of the biological world. Biological mechanisms had a very long time to develop robustness. The trend toward ubiquity may have to make similar improvements in just ten or fifteen years -- if it is to avoid ending itself (and the civilization that supports it).

Even with our best efforts, it seems that safety will eventually depend upon oversight that proceeds faster than committee (or even programmer) review -- and this fact may seem to violate the spirit of the advice given above.

In the long run, the tools we create will have to be very reliable and very smart:

The Technological Singularity

The Technological Singularity is the notion that in the fairly near future, we will create or become creatures of superhuman intelligence. There are various ways this might happen:

• The AI Scenario: We create superhuman Artificial Intelligence in computers.
• The IA Scenario: We enhance human intelligence through human-to-computer interfaces -- that is, we achieve Intelligence Amplification (IA).
• The Biomedical Scenario: We directly increase our intelligence by improving the neurological operation of our brains.
• The Internet Scenario: Humanity, its networks, computers, and databases become sufficiently effective to be considered a superhuman being.
• The Digital Gaia Scenario: The network of embedded microprocessors becomes sufficiently effective to be considered a superhuman being.

[So my talk here at ToorCon has mainly been about the run-up to the last two scenarios on this list.]