The Top 10 Challenges in Extreme-Scale Visual Analytics by Pak Chung Wong, Han-Wei Shen, Christopher R. Johnson, Chaomei Chen, and Robert B. Ross. (Link to PDF. IEEE Computer Graphics and Applications, July-Aug. 2012, pp. 63–67)
The top 10 challenges are:
- In Situ Interactive Analysis
- User-Driven Data Reduction
- Scalability and Multilevel Hierarchy
- Representing Evidence and Uncertainty
- Heterogeneous-Data Fusion
- Data Summarization and Triage for Interactive Query
- Analytics of Temporally Evolved Features
- The Human Bottleneck
- Design and Engineering Development
- The Renaissance of Conventional Wisdom
I was amused by #8: The Human Bottleneck, which reads:
Experts predict that all major high-performance computing (HPC) components—power, memory, storage, bandwidth, concurrence, and so on—will improve performance by a factor of 3 to 4,444 by 2018.2 Human cognitive capability will certainly remain constant. One challenge is to find alternative ways to compensate for human cognitive weaknesses.
It isn’t clear to me how speed counting 0’s and 1’s is an indicator of “human cognitive weakness?”
Parking meters stand in the weather day and night. I don’t take that as a commentary on human endurance.
Do you?
*Alternative* ways? To “compensate”? For suboptimal “cognition”? I don’t understand any of it. Reminds me of Rod Serling’s “Twilight Zone” scripts. They tended to be cleverly vacuous, inviting the TV viewer’s mind to fill in the blanks to which many arrows were urgently pointing. That was Serling’s art, to create urgently-pointing arrows that required you to think about what they *might* be pointing at.
Parking meters don’t last forever, and neither does anything else. If we consider human beings as mere devices for reflecting the intent-to-survive of their “Selfish Genes” (as Richard Dawkins proposed), then those genes have already outlasted many generations of parking meters.
I personally see the increasing speed of computation as an expression of human cognitive strength, not weakness. (I remain very impressed by the ongoing work at Intel Corp and elsewhere.) I also think the predicted speed increase may be wrong by orders of magnitude, in view of the prospects for quantum computing. On the other hand, there are many difficulties to be overcome there. Even so, the speed increase they are predicting seems to me on the low side, considering the higher heat-tolerance of new semiconductors with carbon, instead of silicon, substrates, and the likelihood of improved heat-dissipation approaches. But I’m just guessing about all of this. Nobody knows, I suppose.
Comment by Steve Newcomb — August 30, 2012 @ 3:35 pm