Redundancy & Iteration
This entry will focus on strategies for developing “the process” in terms of both athletic training and product development, and any endeavor where deep learning is involved. Marked points of “shifting gears” to leverage sudden growth are determined and projected to create direction. The slow, grinding, sometimes grueling energy which goes into repeated efforts can be better focused if designed to do so.
Some definitions
A definition of redundancy points us to “the state of being redundant.”
Redundant - adj : characterized by verbosity or unnecessary repetition in expressing ideas
Iteration - n : a development strategy that involves a cyclical process of refining or tweaking the latest version of a product, process, or idea to make a subsequent version
“...we call something reliable when we can count on it to work as designed.”
Evolution happens when those two concepts are integrated with a process. A study of this process provides us with a couple unique tools to improve the producer’s position during the predictable ups and downs of an idea’s life cycle.
Through incremental iterations, a design can be tested with various evolutions. We walk away with data based on the successes or failures of each iteration. Even in an environment without experienced designers, these iterations present an opportunity for improvement pending other qualities
“We can reliably evolve good designs without being reliably good designers: we need only be good at testing, good at tinkering, and good at being patient.”
Though design skills and foresight are excellent, excellent products are the result of testing, tinkering, and patience even as they are the result of insightful and educated designers. These attributes, plus focus and determination, will force improvements in any design.
Redundancy as a Tool
In two ways we can apply these principles : component redundancy and design diversity. Another definition of redundancy specific to production affords itself to the conversation:
Redundant - adj : serving as a duplicate for preventing failure of an entire system (such as a spacecraft) upon failure of a single component.
Designing redundant components means including a degree of overlap of responsibility. Even better, truly redundant components stand independent of each other while being capable of shared responsibility. In the event of the failure of component A, other compenents B, C, D, E, and F take over and compensate during component A’s repair.
Component redundancy
In Engines of Creation, Drexler reaches a conclusion that “redundancy can bring an exponential explosion of safety,” in terms of lifecycle. Independent, redundant components tap into the power of exponentials as they overlap, building an significant buffer. Adding strength or capacity to one specific component design returns only marginal lifetime expectancy; expanding the component set returns gains in lifetime exponentially.
Consider a suspension bridge engineered to stand with 5 cables but built with 6, each having a one-in-365 daily chance of breakage. Should one break, a crew can be dispatched for repairs while the bridge is still able to work. In other words, for the bridge to be destroyed, two cables must break in one day. This bridge is capable of lasting 10 years.
Consider now the same suspension bridge designed with component redundancy and built with 10 cables. This bridge is capable of lasting 10 million years thanks to the power of exponentials, since 6 cables must break simultaneously for failure. With 15 cables, the bridge can be expected to last more than ten times the age of the earth.
Design diversity
In the prior example, we merely expanded the component set even though each component was roughly the same. But what happens when the components are a) independent to each other and b) each representative of a design iteration of the necessary component? Another example is necessary.
Consider a computer which utilizes 4 separate CPU’s (in two pairs). Each of the 4 is capable of performing the same tasks, but are slightly different in design and operation. Each pair performs the work of a normal computer’s single CPU, and does so with more reliability through constant “accuracy checking” against throughput of the other. With continual measurement for internal consistency, failure of any one component can be replaced by the opposite pair and repaired.
With this method, design diversity, a system designed with redundant components can “correct not just failures in a piece of hardware, but for errors in its [own] design.” However, this diversity requires imagination and vision as well as a unique knowledge of the process under scrutiny.
Note that systems wielding redundant design are historically physically heavier, bulkier, and more expensive than their peers. This is true to a degree, but the margin has
Redundancy in Practice
The benefits of redundancy can be realized by mapping the lifecycle of an evolving technology, determining current stage, and applying the tools to shift gears. We can find the sweet spot to focus R&D efforts by charting this lifecycle.
The Product Hype Life Cycle
The Product Hype Life Cycle has 5 distinct areas. The overall trend is positive growth, though large destabilization is present. We can define the “hype” as the current slope of the curve.
Innovation Trigger. Upon initial documentation and implementation, “hype” begins its rise. Expectations are high because possibilities are imagined but not yet disproved. Here, ideas are rapidly tested and systems of productivity are developed. As we realize how much work is actually necessary to make such capabilities a reality, we reach the…
Peak of Inflated Expectations. More studies are completed which grounds founders and investors. The next period can be traumatic if unexpected, as perceived value plummets into the…
Trough of Disillusionment. In this trough, the community regains a grip of what is possible and begin to implement plans toward these new goals. Should the principles of the technology be sound and if the idea is surrounded by the right people now is the time to invest, before movement is begun up the…
Slope of Enlightenment. It is in this period that the most true value is gained. Refueled by investors while in the trough, this period represents time to put in hard work building their entities around the idea. As productivity reaches equilibrium with true value, we wind up in a strong and steady…
Plateau of Productivity. Here, any surviving organizations built around this technology are developed and expanding. Rather than generate and test ideas (which happened in the first and second phases), the organization has developed complex, profitable systems of operation.
This pattern represents the expected cycle when a new technology or capability is “triggered”. Redundancy and design iteration should be practiced in different forms at different stages.
The trigger and climb to initial peak represent a time when product redundancy should be practiced in an effort to cast a wide net of possibilities and protection. During the fall into the trough, the organization retreats back into the most promising iterations, fortifying them by establishing systems of best practice. Once on the slope up towards plateau, design diversity should be practiced, repeating the process of redundancy on involved systems.
As iterations of system design are tested then implemented or eliminated, the company is able to, metaphorically, shift gears. The wheels are turning with tried and true products; now production shifts and cost-of-production is reduced, drastically in cases.
Conclusion
Repetition is teaching, but redundancy means safety. Armed with an understanding of the evolutionary process of improvement, projects can be strategized depending on their stage in the Product Hype Life Cycle. Current projects can be re-tooled to incorporate component redundancy and design diversity. New projects can be planned strategically to implement those tools at milestones.
Thanks to Will Fistek for editing and providing feedback to this entry.
