MODELS AND COST ESTIMATES FROM OTHER STUDIES
In addition to this report, we are aware of three studies that estimate the national costs of connecting all public K-12 schools to the NII. We thought it might be helpful for the reader if we briefly summarized the approaches taken by each study and the resulting estimates. The natural tendency would be to directly compare estimates among the studies; however, since each study models different infrastructures, this comparison is difficult. Accordingly, it seems more useful to review the major similarities and differences in approaches and conclusions among each of the studies. We should also note that each study has informed our thinking, and we have appreciated the opportunity to exchange ideas with the authors of the first two studies (the last one is yet to be published). The three studies are:
MIT/Department of Education
The MIT/Department of Education studies informed our approach early on. The 1995 update (referred to simply as MIT from here on) discusses five models of connectivity which include increasing levels of functionality and expense across all elements of infrastructure.
Several factors account for the differences between the estimate from the MIT study and this study. First, the costs for each model in the MIT work are presented as ranges, while we have estimated a weighted average cost by making assumptions about the distribution of individual costs across schools. For example, within each model the MIT work assumes a single type of connection to the school for all schools, while our approach differentiates between rural and non-rural schools. Second, while the models describe similar levels of infrastructure, they are not identical. Third, the MIT models assume that the current costs for deployment and operation/maintenance remain constant over time, while we have adjusted for declining prices in certain items. Fourth, we have included some initial costs that the MIT researchers have excluded by design-for example, certain software (specifically, packaged applications), furniture stations, printers, and security devices. Finally, we have made different ongoing cost assumptions. Relative to this study, the MIT report assumes less training and support, hardware replacement cycles that are (implicitly) over twice as long, and no packaged software or upgrades.
TIAP
The TIAP study is also similar in approach in that it estimates the costs for three deployment models from the ground up. The TIAP models, for which annual costs are estimated based on five- and twenty-year deployment cycles, are as follows:
In addition to assuming broadband in all models, the TIAP study is different from this report in other respects. First, it does not reduce deployment costs by the currently installed base of computers within the schools. Second, it does not include telecommunications usage charges to the schools; instead, it includes the costs to the Local Exchange Carriers (LECs) of providing broadband service. The TIAP study makes this distinction in order to separate the issues of cost and price for several reasons. First, there is no known tariff for broadband access to schools or any suitable analogous tariffed service. Second, it was conjectured that the costs to provide broadband access to schools might be recovered in ways other than the usual tariffing process.
Milken
The Milken study takes an entirely different approach. Researchers at the Institute surveyed the state education superintendents as to what it would cost to complete their K-12 technology plans. Based on the 40 states that responded to the survey, the Institute projected a cost of $31 billion to "fully implement [each state's] vision for technology." While details of the underlying state technology plans were not available at the time of writing, it appears that the state plans are, on average, less ambitious than the Classroom model outlined in this report. Further, the Milken study seems to have focused on the costs to deploy the infrastructure, not to operate and maintain it.