Choice Group - 18 Mar 2020

The good news

About Killing People

Zooming to you live
18 Mar 2020 at 4:30pm GMT

Susanne Burri and Bryan W. Roberts
London School of Economics & Political Science

Cautionary news

About Killing People

and other irreversible acts

Zooming to you live
18 Mar 2020 at 4:30pm GMT

Susanne Burri and Bryan W. Roberts
London School of Economics & Political Science

Plan:

Choosing death
Irreversibility
Cautionary news principle
Expected utility

Choosing death

"Several countries have demonstrated that this virus can be suppressed and controlled. The challenge for many countries who are now dealing with large clusters or community transmission is not whether they can do the same - it's whether they will."

Dr Tedros Adhanom Ghebreyesus
Wed, 11 Mar 2020

Thu, 12 Mar 2020

Predicted deaths after six days: 60

(Average growth factor: 1.42)

UK Corona Response

"you want to protect, those at risk groups so that they basically don’t catch the disease, and by the time they come out of their cocooning, herd immunity’s been achieved in the rest of the population"

David Halpbern
Thu, 12 Mar 2020

UK Corona Response

"If you suppress something very, very hard, when you release those measures it bounces back and it bounces back at the wrong time"

Patrick Vallance
Fri, 13 Mar 2020

Tue, 17 Mar 2020 (Yesterday)

Predicted deaths after six days: 673

(Average growth factor: 1.45)

18 Mar 2020 (Today)

"Herd immunity" has been dropped

Mass gatherings are still allowed

Bars and restaurants are still full

Schools are still open

Testing is only for the hospitalised

Consequences of unsupressed coronavirus

UK Population: 66 million
80% Infections: 53 million
1% Death Rate: 530,000

Choice set

Option 1 (Reversible)	Option 2 (Irreversible)
Now: Benefit	Now: Loss
Later: Uncertain	Later: Uncertain

This happens often

Inaction (Irreversible) vs. Climate action (Reversible)

Surgery (Irreversible) vs. Waiting (Reversible)

Death penalty (Irreversible) vs. Life sentence (Reversible)

Letting go (Irreversible) vs. Pushing (Reversible)

Rights-Based Approach

Death is gravely wrong in most cases.
Exceptions life forfeiting or waving one's right-to-life can be investigated.

c.f. Thomson (1990) or McMahan (2002)

Consequentialist Approach

Rarely studied
...because it's easy?

Consequentialism with Certainty

An act is justified iff
a morally better outcome is produced

Consequentialism with Uncertainty

Utilitarian: just maximise expected utility
Maximin: just maximise the minimum cost

Example: Two equally good expected outcomes

Rights-Based: Choose more life
Utilitarian: Choose either
Burri and Roberts: Weigh irreversibility

c.f. Burri (forthcoming)

Irreversibility

A modern idea

Source: Google Ngram Viewer

Irreversible physics

Decay in the weak sector

Irreversible chemistry

Irreversible biology

Irreversible flow of time

proteron, husteron

Irreversible choice :

A decision that you can't take back

('Take-back': opportunity to consider the same choice set at a later time)

Choice set

Option 1 (Reversible)	Option 2 (Irreversible)
Now: Life	Now: Death
Later: Uncertain	Later: Uncertain

How to weigh irreversibility?

Cautionary news principle

Bernanke : 'bad news' principle

(Princeton; Brookings Institute;
Former chairman of the US Federal Reserve)

Bernanke : 'Bad news' principle

Uncertain future information

Confirmatory: Reason to act irreversibly
Cautionary news: Reason to wait

Cautionary news principle :

When comparing an irreversible decision, e.g. death, to a reversible one, e.g. life:

possible future cautionary news is relevant;
possible future confirmatory news is not.

(SB & BWR, manuscript)

Choice set

Option 1 (Reversible)	Option 2 (Irreversible)
Now: Life	Now: Death
Later: Uncertain	Later: Uncertain

Life option

Future confirmatory news: You can still change your mind.
Future cautionary news: You made the right choice.

Death option

Future confirmatory news: You made the right choice.
Future cautionary news: The wrong choice was irreversibly made.

Possible cautionary news

Matters more

Expected utility

Bernanke-inspired

Quantitative Analysis

Initial value: \(V_0\). Cost of irreversible act: \(C\)
Cautionary news: \(n>0\). Confirmatory news: \(m>0\)

\[ V_{t_1}:=\left\{ \begin{array}{cl} (1+m)V_0 & \textrm{ with probability } p,\\ \\ (1-n)V_0 & \textrm{ with probability } 1-p.\end{array}\right. \]

Future Discount-Factor:

\[ \frac{1}{1+d}\;\;\;(d>0) \]

Decision Model Assumptions:

The reversible act is a choice at time \(t\) only if the irreversible act was not chosen at an early time \(t'< t\).
The choice of highest expected value should be selected (utilitarianism).

Theorem ("cautionary news principle")

The choice to act irreversibly should be selected iff

\[V_0 > \frac{d+(1-p)}{1+d+(1-p)(1-n)}C\]

(Observe: confirmatory news \(m\) does not appear)

Theorem Gloss:

The quantity of future cautionary news (i.e. news supporting not acting) is relevant for the decision to act irreversibly, but not the quantity of of news supporting acting.

Theorem ("cautionary news principle")

The choice to act irreversibly should be selected iff

\[V_0 > \frac{d+(1-p)}{1+d+(1-p)(1-n)}C\]

Low probability supporting news \((1-p)\approx0\) yields an ordinary cost-benefit analysis: act irreversibly iff \(V_0 > C - \text{(discount factor)}\)

Theorem ("cautionary news principle")

The choice to act irreversibly should be selected iff

\[V_0 > \frac{d+(1-p)}{1+d+(1-p)(1-n)}C\]

Low cost supports acting irreversibly: \(C\approx0\) implies \(V_0>0\), which implies that we should always act.

Hesitation:

A real number cannot possibly valuate a human life.

...But our analysis is just a start.
Other decision frameworks can be explored/compared
The 'spirit' of the argument requires no formal analysis
We can do better than existing arguments.

Finis

Burri and Roberts (2020)
"Cautionary news about irreversible acts"