Tuesday, September 8, 2015

Article on Energy and ICT

[This post summarizes the arguments presented in the op-ed article by  Nischal Regmi and Shailesh Pandey appeared in the NagarikNews oped of September 8, 2015. Regmi and Pandey are researchers affiliated to Martin Chautari The article is in Nepali and can be found here].

The vision documents related to Nepali ICT/IT such as the proposed ICT policy and e-government master plan are critically short on discourses related to the energy infrastructure to support such a large-scale development. The reality of the proposed ideas can be attacked from various angles. We make two technological arguments on why Nepal is not prepared and might not be able to sustain the demands of a pervasive technology.

System Argument: The energy profile of the large scale ICT projects can change drastically based on what devices and units are excluded from the system boundary. If we are to take out personal computing devices, data centers and amplification and refrigeration units from the energy equation then the energy used by other components of ICT/Internet are very low. By excluding/hiding these components a hopeless plan can look realistic and sustainable. In 2013, the data centers in the USA consumed electricity that could power New York twice over. Similarly, including personal devices, servers and network devices the energy consumption can change by one or two orders of magnitude (i.e. from 10 to 100 times). The argument that efficiency derived from innovative applications of technology to industry does not work in case of Nepal. The industrial sector only consumes about 5.25% of total energy production. And, the increase in efficiency is being caught up by the demand, aka Jevons effect.

Technology Limits: There are fundamental limits to how fast the computation and communication technology can improve. The computers (chips) cannot get powerful and efficient indefinitely at a blinding pace we are accustomed to for the last two decades. Some say that the halt will come in effect as early as 2036. Similarly, the user demands are stretching the improvements (increased capacity) that an optical fiber can give. The industry will only take interest in the development of a new technology as long as it finds an investment in its return. This could effectively move the deadline even closer.

No comments:

Post a Comment