Technology will either deepen the chasm between the best and the worst SMET education or help the very best SMET education reach all learners. It will require a conscious and sustained effort to avoid the former and realize the latter. If, but only if, extending the reach of the best SMET education is made an explicit goal, a national digital library for SMET education can have a significant and salutory impact.

Although the cost of hardware, software, and network access are well-known obstacles to wider use of information technology, there are other important obstacles. NSDL will need to work on all the potential obstacles, including but not limited to:

• Hardware and network limitations. Wherever possible content should run on the least expensive feasible platforms. The NSDL might periodically establish "target platforms." By promising users that target platforms would be able to run most library content, this would establish incentives for upgrading equipment. By promising developers that content running on target platforms would reach a larger audience, this would provide incentives for less demanding content. By setting new target platforms once a year, an NSDL would be able to evolve.

• Human infrastructure. Many of us are fortunate enough to work in environments with good technical support or have dogeared yellow sheets of paper with the names, email addresses, and phone numbers of knowledgable people. Many potential users, however, lack that support. The NSDL might include "help" services of various sorts including FAQs, email help, and toll-free numbers. NSDL might also provide training services for users who are less experienced with information technology.

• Limited time. Potential library users can be discouraged by one or two failures or by heavy demands on their time. NSDL resources should be reliable and user-friendly. Wherever possible resources should be "zero-install" or "turnkey."

• Curriculum materials. Some of the most promising new curriculum materials are built on powerful but expensive sofware packages -- in mathematics, for example, computer algebra systems like Maple, Mathematica, MathCAD and DERIVE. There is a huge chasm between students and faculty at institutions with site licenses and those at other institutions.

• Licenses and memberships.In general, NSDL should be wary of the possibility that site licenses and institutional memberships can exacerbate the gap between students and faculty at wealthier institutions and those at poorer ones. We need to seek and encourage more creative pricing strategies with accessibility and reach as an overriding goal.

• Cumulative effects of individually reasonable requirements. Many of the resources currently available on the Web have specific requirements that are by themselves entirely reasonable. The cumulative effects of many such requirements, however, can become prohibitively expensive or even impossible.