#TFOP (The Future Of Prediction) --- SynTalk


Episode Artwork
1.0x
0% played 00:00 00:00
Mar 17 2017 74 mins  
How unpredictable are we? Do our desires and constraints make us more detectable? Is data collection (somehow) always human? What’s not predictable? Can terrorist behaviour be predicted with consistency? Must data be balanced? Are cheats (terrorists, hackers, spiders, bots) more easily detected only if they are in the minority? Must predictions necessarily have explanatory power and granularity? Would tsunamis become completely predictable in the long run, as solar eclipses now are? Are predictions, like for gravitational waves, sometimes very difficult to confirm? Must we sometimes capture 'all' the data (say via sky surveys) even before deciding on all the analyses? Would natural history be more predictable if there were no astronomical events or human beings? Do very rare events (like 'zero day attacks’) often go undetected? Can we even predict if (and where) life might be elsewhere in the universe? Why is it so difficult? Will we know if/when aliens invade the earth? Could we be gone in the blink of an eye? Is it comforting to know (…precisely) how the night sky would look on 14th September 2400 AD? SynTalk thinks about these & more questions using concepts from statistics (Prof. G. Jogesh Babu, The Pennsylvania State University, Pennsylvania), natural history (Dr. John Mathew, IISER, Pune), & computer science (Prof. V. S. Subrahmanian, University of Maryland, Maryland). Listen in....