Lecture 7: Fallacies & Risks
Four Points About
Arguments
1. Building your case on mere possibility
-In general, a weak strategy
-Anything’s possible
-Reasonable possibility is generally much more interesting
-Remember that when arguing about which engineering outcome to
choose, all options will typically carry some possibility of a catastrophic
outcome
Fallacies
A fallacy is a common error in reasoning which people (even well educated,
careful people) often fail to notice in their own arguments or which
devious people might use in their arguments in the hope that we won't notice
them.
Four Points About
Arguments
2. Mistaking Possibility for Probability
-E.g., there is a chance that containment will fail, therefore
containment is likely to cost much more than $20 million
-Here, you need an argument to establish probability
-Beware of committing this fallacy by going from possibility on
top of page 1 to probability on the bottom of page 2
Four Points About
Arguments
3. The Straw Man Fallacy
-Misrepresenting the position
of the opposing side in such a way that the opposing position appears obviously
false or ridiculous.
-Creating a 'straw man'
just so you can knock it down.
-E.g., MacLellan hasn’t considered that containment might
fail
Four Points About
Arguments
4. The Concorde Fallacy (a.k.a. the Sunk Cost Fallacy)
-What’s wrong with the following argument?
-One reasoning for choosing incineration over containment is that so much money
has already been spent on pursuing the incineration option. If we go with containment, that money will all have been
wasted.
Four Points About
Arguments
The Concorde Fallacy
-Thinking that the fact that a certain amount of money has been
spent pursuing option A is by itself a reason for continuing to pursue option
A rather than some other option.
-Sunk costs are sunk whether we stick with option A or not.
-Similar arguments were used during the building of the Concorde
Risk
- Risk is inevitable, people won’t always do their jobs well
- Two Phenomena:
-Normalization of Deviance
-A phenomenon to be resisted
-Normal Accidents
-An unavoidable aspect of engineering?
Normalizing Deviance
-Accepting anomalies or increasing boundaries of acceptable risk
-The Challenger case (153-4)
-a long string of acceptance of differences from predicted behavior
-“the joint was officially certified as an acceptable risk, even
though the joint’s behavior deviated from design predictions.” (153)
Normal Accidents?
It has been suggested that two features of high-risk systems make
accidents both likely and difficult to predict and avoid.
Tight Coupling
-Processes affect one another in a short time
-Little time to correct and contain a failure
Complex Interactions
-Parts of the system can interact in unanticipated ways (150-1)
Are Accidents a ‘Cost of
Doing Business’?
According to sociologist Charles Perrow “accidents
in complex tightly coupled systems are inevitable.” (151)
Should we simply accept that engineering (and medicine and …) will
sometimes lead to catastrophies?
Perceptions of Risk
As MacLellan pointed
out, there is a difference between how experts and the public perceive risk
Expert: Risk = hazard+pathway+receptor
Public: Risk = hazard
Likewise, the public and experts often differ when
it comes to estimating acceptable degrees of risk
Acceptable Risk
Degree of risk: “the product of the likelihood and the magnitude
of the harm.” (155)
-A high likelihood of a minor harm may be a greater
risk than a low likelihood of substantial harm
-Rooted in a utilitarian, cost-benefit approach
Acceptable risk: “An acceptable risk is one of where, given the
options available, the risk of harm is at least equaled by the probability
of producing benefit.” (156)
Limitations on Acceptable Risk
As a utilitarian notion, this idea suffers the usual
problems with that view
-Estimating likelihood
-Estimating cost of outcomes
Most would argue it is also in need of deontological
limitations
-Suppose project has risk of catastrophic harm
to some small group of individuals
-Respect for person may rule out the project even
if cost/benefit analysis is OK
Estimating Risk
-In general, people are not good estimators of probability
-In a study asking subjects to estimate risk of harm from smoking,
driving, skiing, etc.
-Expert estimates were ~10 times too low
-Public estimate were ~100 times too low (157-8)
Tendency to focus on dramatic cases
-Driving with cell phones vs. fiddling with radio, coffee
Estimating Risk
Risks are seen as greater if caused by humans, than if they have
a natural origin. (158)
The Washington sniper?
Risks are discounted over time – an immediate risk is seen as being
30 times greater than a delayed one (158)
Voluntary Risks
“voluntarily assumed risks are seen
as inherently less risky, not simply more acceptable.” (158)
“laypeople are generally willing to
take risks that are 1000 times … as uncertain as voluntary risks.” (158)
The Public Conception
of Risk?
Some argue that the public are not actually bad at estimating risks,
they simply do not distinguish between risks and acceptable risks (159)
In other words, the public has a normative conception of
risk while experts have a descriptive conception
A Normative Conception of Risk
If we accept this idea, then what is involved in a risk/acceptable
risk?
-Consent
-Fairness
-Control
A deontological notion of risk?
Reconsider the Tar Ponds dispute in light of this
The Principle of Acceptable Risk
“People should be protected from the harmful effects of technology,
especially when the harms are not consented to or when they are unjustly distributed,
except that this protection must sometimes be balanced against (1) the need
to preserve great and irreplaceable benefits and (2) the limitations on our
ability to obtain informed consent.” (167)
Moral Risk?
To what extent, if any, is a person’s moral status (i.e., how good
or bad he is) subject to risk
One view: how good or bad you are is not
vulnerable to luck (i.e., risk) at all
How things turn out is irrelevant morally
Is this true?
Legal Luck
The preceding view is not true where the law is concerned. Some actions carry a legal risk
i.e., how much wrong you have done legally is connected to how
things turn out
Attempted murder vs. murder
Dangerous driving vs. vehicular homicide
Moral Luck
Some argue that just as we are subject to luck legally, we are
also subject to moral luck
-How good or bad we are is dependent on luck
-Contrast utilitarianism vs. deontology on this point
-How troubling an idea is this?