Sunday, June 26, 2016

Responsible Problem Management

Our present approach to producing products tends to limit the amount of information needed by those potentially affected by the products to adequately avoid waste and other adverse impacts. This can be traced to how businesses treat problems.

Some problems are known (identifiable and understood) and can be fixed based on experience and a basic learning curve; while others, or parts of others, are unknown along with how they can be fixed. A problem is typically identified by the differences between observations and expectations, which is the function of testing. Understanding the problem is more challenging because it involves finding the causes of those differences (troubleshooting). Testing and troubleshooting together are considered a task that is often performed by a person or team, just as fixing the problem can be considered a task performed by a person or team.

Both "finding" and "fixing" tasks are often performed simultaneously, resulting in some fraction of the overall problem being fixed over a given time. Fixing the problem stops when its total cost exceeds either the available resources or the cost of living with what's left to fix. This economic approach to fixing problems has the important downside that the finding task is also stopped, which limits knowledge about how much is really left to fix and makes its assessment more of a guess.

Some simple modeling using basic learning curves shows that, as a rule of thumb, the effort required to find and fix a practically unknown problem is three times the effort required to fix an entirely known problem, assuming that each task uses the same number of people and everyone performing those tasks has at least average preparation and proficiency. If guessing is unavoidable, then prudent planning would include this margin. If knowledge about the minimum effort is unavailable, then we can assume (also as a rule of thumb) that it will be ten times the effort that an ideal team under ideal conditions would need to expend.

In my experience, most of the "find" task is typically done during research and development, before large-scale production of a product is even attempted. Most potential problems are identified and fixed during redesign of the product. Also, it is common to complete only up to 80% of the find task prior to production, which my modeling suggests is the result of a plan to use the minimum effort for both finding and fixing (instead of just fixing).

The 20% difference in quality may or may not be found by customers, whose feedback could be incorporated into future "products" along with new features that justify them paying more. This is a way of masking and evading twice the acknowledged economic cost (whose largest component is essentially effort). The physical cost still exists, however; and can become much larger when the product is used in a dependent relationship with other products and conditions that are combined in a system. If a system's performance degradation becomes large enough due to multiple problems that haven't been totally fixed, then an additional cost may be added to fix it, some of which is indirectly due to deficiencies in an individual product and is thus passed along to the customer of the product.

One way that a producer can ensure keeping their own cost down is to define product performance to only include what they choose to totally fix. The success of this strategy depends on knowing what that is before they set expectations for the product with their customers, and having a way to deflect responsibility for the consequences of what they missed. This may be one reason that businesses try to eliminate government oversight in the form of inspections and regulations beyond what they explicitly promise to their customers, because accountability for consequences of what they missed could force the cost back on them.

Application of the precautionary principle would also force the cost on the producers, by making them assess the impact of their products on people and environments where the products might be used, and not deploying products where any of those impacts are negative. A major argument against this is the stifling of innovation which might result in advantages over existing products, an argument that is implicitly built into biological evolution, a process of creating "products" (species) that die out if they don't provide an advantage to procreation.

Since the main test of products is their continuing sale to customers, and sales can't be made if customers die before they buy, it might make better sense to ensure that full investigations of problems and potential problems are conducted, either by businesses or agencies of the public, and the the results released to everyone for evaluation. The investigations and release of results would continue as the conditions, environments, and underlying understandings change beyond those originally considered.

Key to enabling evaluation would be development of truly universal and quality education that includes unbiased access to basic knowledge, understanding, and skills, along with experience in the investigation of products over a wide range of types. If the complexity of products exceeds the ability to adequately investigate and evaluate them, then people would become less likely to accept them and they would ultimately be discontinued.