Questioning what you think you know

Last Friday, Eugene Spafford became the first of four speakers in a new Distinguished Colloquia series offered by the School of Informatics. Spafford’s talk—“The Value in Questioning What You Think You Know”—was a reflective look at current hardware and software practices, many of which are based on the computing landscape in the 1950s.

Patterns for early solutions become constraints for later ones.

With the intention of exploring non-traditional thinking about computing, Spafford began by sharing a half dozen exercises meant to illustrate several themes of his current work:

• Beware conditioning
• Be alert to ambiguity
• Don’t assume context
• Be sure you understand boundaries
• The usual is not the rule
• What computes is not the same as reality

These take-home statements set up Spafford’s main message: examine the current context of computing and come up with new best practices that make sense.

Much of the time spent updating computers deals with security problems owing to an outdated set of rules for how software and hardware are best built. Spafford notes when the first computers started using transistors, the challenges of computing were quite different than they are today. Transistors cost 1/500,000th of a dollar today compared to about $60 (current value) back in 1958. 2005 marked the first year that we produced more transistors in the world than grains of sand rice. Disk storage maxed out at 200 bits per cubic inch in 1958, costing about a dime per byte. Today, we can store 425 GB in the same physical space at a cost of about a dime per 100 MB.

“Why are we still using some of the same basic paradigms today that were applicable in 1958?” asked Spafford.

Minimization is tantamount to good security, he added, so the elimination of unnecessary protocols—accepted in the name of efficiency but born out of constraints that are no longer relevant—can potentially reduce opportunities to compromise computer systems. The distinguished lecturer identified several available opportunities to re-think how we design computers: use of memory, paging, programming, stacks, shared libraries and all-in-one operating systems.

Even if Spafford’s thought experiment succeeds in inspiring the creation of a new modern computer, that re-imagination might come with a price. Innovation is a difficult proposition. The time when radical changes are accepted on merit tend to come in early adoption, a phase of the technology life cycle long since passed in computing. Business and personal users alike have invested heavily in a now familiar way of using the technology, and thus are less likely to accept any improvement that requires uncomfortable transitions to a new paradigm.

Likewise, the sustainable design of computers must be considered. The world may not be able to survive a massive replacement of existing toxic hardware, even if the new toxic hardware is superior from an engineering point of view. In the Q&A after the talk, Spafford failed to address the environmental issue directly, suggesting that “we replace our computers now” as an off-the-cuff justification for why it would be OK (or no worse) to do so in the future. The outside-the-box creativity encouraged in this talk needs to be extended to our practices of what we do with the things we make obsolete.

Criticism aside, the ideas Spafford put forth are very relevant to design in general. The moment we create, adopt and use our designs, we establish a set of constraints befitting only the current context, in the best case. Time shifts all situations, helping things to become more powerful and needs to fade in priority. In their place, new needs emerge from the changed conditions. The objects, though, remain fixed. At what point do we fight the unwillingness of a user base to accept change through re-examination of their needs? Can we anticipate common changes and embed that flexibility into the initial design?

Spafford, a professor of computer science at Purdue University and executive director of CERIAS, is known for work in information security and privacy, software engineering, and computing policy. According to the colloquium bio, “Some people consider him a polymathic futurist, and others simply think he’s an iconoclastic crank.”