video: Movie shows extreme sub-second behavior in the networks underlying the current financial system. Such extreme behaviors can offer fresh insight into the function -- and importantly the malfunction -- of a brain. The entire episode lasts less than 500 milliseconds, which is below the human response time and faster than the typical timescale of human consciousness. This material relates to a paper that appeared in the Feb. 24, 2017, issue of Science, published by AAAS. The paper, by N.F. Johnson at University of Miami in Coral Gables, FL, and colleagues was titled, 'To slow or not? Challenges in subsecond networks' view more
Credit: Neil Johnson
In a world increasingly reliant on high-speed networks, introducing microsecond delays into such systems can have profound effects. In this Policy Forum, Neil F. Johnson discusses the need to better understand how these micro-delays may drive more extreme behaviors, particularly in the context of the financial market, but also in other fields. In fall 2016, the U.S's fastest and largest financial network was subjected to its first ever intentional delay. The seemingly minute delay of 350 microseconds was significant enough to give traders the false impression of an increase in market activity, which can trigger trading algorithms and create a feedback effect on the price dynamics. Johnson says that policy-makers justify this delay because it levels out highly asymmetric advantages available to faster participants. The impact of these delays needs to be better understood, however, before their broader use, he says. For example, data-capturing a 500 microsecond delay and bunched delays resulted in substantially more extreme trading behavior. The need for a better understanding of the impact of delays is increasingly urgent, as similar techniques are poised to be extended beyond the financial market, for example to navigational networks in driverless cars or drones. The core scientific challenge, Johnson says, is to be able to predict the types of extreme behaviors that an intentional time delay will generate. Classifying a system's sensitivity to such delays could help policy-makers tailor policies to cope with different market scenarios, Johnson concludes.
###
Journal
Science