First Part: Disposable Organizatons

Topic 1: Success and Exploration

What are the effects of success and failure on exploratory behavior?

Success generates slack resources which are often used in increased exploratory behavior. Success-driven exploration is serendipitous, thriving on novelty. It's less efficient and has a lower probability of return than exploitation.

Failure reduces slack and shifts focus toward exploitation, refining existing abilities rather than exploring for other alternatives. In certain technological situations failure reduces the propensity to search, which reduces success and competitiveness and further increases failure (the failure trap).

To what extent do successful organizations experiment more than, or in ways different from, organizations that are failing?

Successful organizations conduct more exploratory search when they are far above their targets and have alot of excess slack. Exploration during success is often unintentional, driven by the organization's underestimation of risk based on over attributing their success to their own abilities. Their search is more variable in their search behavior, with less control and more involved with "free association, madness, loose discipline, and relaxed control". The returns tend to be in the long term and only after they are exploited. Eventually these exploratory inefficiences reduce performance and cause an elimination of slack, or aspirations increase which also reduces slack.

Also, successful organziations may also continue to exploit existing technologies (what made them successful) and continue to gain success, but at the risk of failing to explore enough and being vulnerable to competitive threat.

Those who are failing and are below their targets also tend to engage in exploratory behavior, though I suspect it is not as undisciplined and search based on slack. A failing organization is probably more poised to quickly exploit a newfound technology. Exploratory search after failure is a higher risk proposition than exploration after success. Often companies fall into a failure trap where they keep searching and failing to exploit technologies long enough to realize their benefits. Their impatience condemns them to miss good ideas.

Topic 2: Adaptive Aspirations

What are the consequences for exploratory behavior in organizations in which aspirations adapt to experience?

I think it depends on whether the aspirations are adapted so that the organization is always slightly below or above aspirations.

Theoretically, if an organization continually changes it aspirations to be slightly higher than it currently is, it will always be in a situation where it is willing to take some risk on exploratory behavior. Slack will not be created that would lead to inefficiencies, but the organization will always continue to engage in some exploratory research. The risk of a "failure trap" is strong if the organization becomes impatient with short term rewards (which favor exploitation of existing ideas or underexplotation of explored ideas).

If the organization revises its aspiration upwards after every success they will minimize excess slack and prevent it from creating too much inefficiency. Managers may also not have enough time to attribute success to ability and "rest on their laurels". However, adapting aspirations too rapidly may queill truly frivolous and "mad" search which can possibly lead to great gains.

For example, frivolous search may involve exploring totally new product lines, while failure-induced search may involve innovations to existing product lines. However, if organizations adapt aspirations to always be slightly above them (e.g. "do as good as last year") they will avoid exploration and focus on exploitation.

Does it matter if the adaptation is self-referential (that is, aspirations are a function of one's own past performance) or other referential (that is, aspirations are a function of the past performances of others)?

I think it depends on which group was most successful in the past, self or others.

If you have been more successful, you may tend to be risk-averse (to maintain success), which would minimize exploration. If others have been more successful, you might be more risk-seeking to aspire to others performance levels.

If the aspirations are self-referential, there is consistency in ones actions. However, there are situations where one can progress versus one's own aspirations and yet fall behind the competition (e.g., the computer industry). But if one is currently the market leader, using one's own performance to set aspirations would make more sense (especially if one sets the aspiration high enough to stimulate exploratory behavior.)

IF the aspirations are other-referential, one may either be extremely below aspirations (leading to risk averse behavior), below target (leadingto exploration), or slightly above target (leading to more exploitation) or far above target (leading to slack and more exploration). The organization may become more adaptive but less reilable.

Topic 3: Future Organizations

How will organizations of the future differ from organizations as we know them today?

Future organizations will be:

* less hierarchical in coordination (due to info technology)

* less emphasis on learning by doing and more emphasis on accessing existing information

* organizatons will rely on other organizations for information and learning

* less affected by political boundaries and nation-states

* rely more upon "knowledge inventories" to survive

* increased focus on research and education

Why?

The world is changing more rapidly, and organziations will continue to exist within volitile environments. More changes now happen within the lifetime of individuals and organizations. Global linkages span political boundaries. The information revolution changes the capabilities for coordination, control, and learning. Competition is become more based on knowledge. Science is becoming private property.

Democratic politics is become less certain, with difficulties balancing demands for more services and less taxes.

As a result, organizations have less time to react. The existing changes have weakened the buffers that had protected organizations from volatility. There are fewer protections against competitors.

How will organizations of the future adapt?

Organizations will become disposable. Individual organizations will be temporary constructions within a larger societal framework (maybe under the guise of a parent company). Organizational variety will be maintained at the population level. Adaptation will be through variation, selection, and retention of more rigid, exploitation-oriented organizations.

These disposable organizations will be short-term focused with only modest adaptive durability. A highly efficient organization will be maintained until it is disfunctional and then it will be discarded for another form. There will be an adhoc construction of project groups or collaborations linked by constantly changing hierarchical networks.

What will be the problems of adaptation?

First, these disposable structure goes against existing societal norms of stability of workforces. Disadvanteged will seek redress.

Second, adaptation requires a balance of exploitation and exploration, but both factors tend to undermine the other. The disposable organization may resist eventual dissolution by engaging in exploration themselves, which would undermine their efficiency and make the result less than optimal.

Also, there are normative pressures to adopt the predominant structure. It will be hard to nuture mutant organizations until they can prove their viability.

What is the role of imagination in adaptation?

Imagining possible futures is one way to get a source of new organizational experiments and find different ways of thinking and acting. The maintainence of illusion through imagination also can help shield new ideas from the normative pressures and hostile environment. New forms need to be buffered from the drive to learn only what is technically efficient (currently). By restricting short-term feedback one can shelter the new idea until it can rise above the noise.

The new era of disposable organizations will require illusions to conserve belief in the face of failure. A business plan or a vision are such forms of fantasy.

Yet there is alot to be viewed as unjust to sustain individual madness at the individual organizational level to maintain variety at the population level. Most new ideas are bad or disasterous.

Second Part: Learning to Be Risk Averse

How are risk preferences portrayed in the literature on choice?

In that literature, why are individuals risk averse?

Standard behavior theories show that people are risk averse for gains and risk seeking for losses. In economics it is caused by a decreasing marginal utility for gains and a decreasing disutility for losses, where utility functions are seen as a fundamental primative of behavior.

Rational choice theory states that there are different beliefs about consequences and subjective values associated with those consequences. Theorists in this paradigm state that people try to get above their aspiration level. When alternatives are all above the target they will minimize the chance they will go below their aspiration level. When the alternative are all below the target they will try to maximize their chance of getting above the aspiration level.

What is the alternative suggested in the "Learning to Be Risk Averse" paper?

This paper suggests that risk averse behavior for gains and risk seeking for losses is a result of simple learning behavior applied to the outcome probabilities. Favorable responses lead to increased propensity for an alternative, unfavorable responses lead to a decreased propensity. This learned response behavior will systematically lead to the observed gain/loss risk preference.

What is the theorem?

Each alternative, if chosen, has a probability of occuring. A learner will probabilistically decide based on learning experience with each alternative which alternative to pursue. The chosen alternative will occur based on its inherant probabilty . The learner will use the outcome to update his/her preference for choosing the alternatives.

The theorem states that in a favorable outcome the learner will increase the propensity to choose the same alternative the next time.

How is it demonstrated:

March uses three learning models to simulate learning by experience:

Fractional adjustment model -- fractionally adjust the propensities for A vs B based on the last outcome (with the speed of adaptation = a). This models an extremely short memory.

Average Return Model -- Change propensity based on the % of return gotten by that alternative (If the averag return of alternative A 60% of the total sum of average returns, then A will be chosen 60% of the time). It assumes an infinitely long memory

Weighted Return Model -- This model combines the last result with the memory of average returns. It weights the most recent result more than the memory (with increasing B).

Results?

The results were that all three models produce risk aversion for gains and risk seeking for losses. In fact, he found that the "premium" needed for people find neutrality between the safe and risk alternative ranged between 4 - 19x. Even when the difference between the safe and risk alternatives is reduced to 0, there is never a situation where the more risky alternative is preferred.

Also, faster learning leads to reaching risk averse/risk-seeking behavior more quickly.

What are the implications?

Learning modifies the sampling rate of the alternatives which in turn changes the outcome. If learning is quick, one may avoid a perceived risk alternative before its true riskiness is really evaluated. Thus, small sample learning of risky alternatives may be misguided. Alternatives that are usually fairly poor but occaisionally good are interpreted as worse than they really are. Likewise, alternatives that are usually good but occaisionally poor are intrepreted as better than they really are.

Overestimation of the attractiveness of a risky alternative is self-correcting, but underestimation of the risk of an alternative is not self-correcting.

The results show that a simple learning model can explain how people can become risk averse over gains. It isn't necessarily a trait or a ambiguous utility function. Also, individual learnings are shared across people via rules and procedures for non-repetitive situations as individuals do across repetitive ones.

To encourage risk seeking for gains we have to slow down learning. It can result from indoctrination outside of experience, limited learning, or noise in learning experience.

It also suggests that fast, precise learning may not be that adaptive, by increasing risk aversion and reducing exploratory behavior.