Showing posts with label negotiation. Show all posts
Showing posts with label negotiation. Show all posts

Friday, 22 February 2013

The Siamese Emperor (Part 2)

Why should the Lords Prasopgarn and Sukhum negotiate?

(See The Siamese Emperor Part 1 to see the framing of the problem.)

---------------------------------

Prasopgarn and Sukhum are in a rather difficult situation.  In the ‘game’ they are being forced to play there is only one other player.  They have no reason to play against the Emperor because the Emperor’s outcomes are equal across the board.  They could decide to play to win, but the most likely outcome of that is that they both die.  They do, however, have a better alternative than to play to win.

The Lords could decide to play in such a way so as to not lose.

Consider the likelihood of winning.  Of the four outcomes, only one is a clear ‘win’, where the rival is liquidated and his entire birthright passes to the victor.  Two outcomes are losses in which the rival either prevails or joins the other in death.  The final outcome is not a win; then again it is not a loss either, but rather a draw.

Now consider the likelihood of not losing.  Only two outcomes are losses.  If the decision to co-operate or defect was made randomly, each of the Lords would have a twenty-five percent chance of winning but a fifty percent chance of losing.  As has been discussed, a non-negotiated, non-random outcome may be even more likely to result in a loss to both parties.

While a win may be preferred, there is a significant factor common to both non-losing outcomes – on-going survival.  This fact is salient because it gives the Lords motivation for negotiating towards peace and a draw.

Neither Lord would be able to negotiate openly to a win, since neither Lord is suicidal; neither of them will accept a negotiated settlement which requires their death.  However, they could both justify negotiating towards not losing.  And both have a very good reason for attempting to convince the other that he will honour a peace treaty.  Their very survival depends on how successfully they can cultivate trust in the other.

As a rational actor, Prasopgarn knows that Sukhum will only honour a peace treaty if he believes that Prasopgarn will do the same.  Therefore, Prasopgarn must convince Sukhum that he will not act treacherously.  He could appeal to reason, but that seems fraught with danger as such an argument must be based on a lie.  Prasopgarn cannot argue that maintaining a peace treaty is in his personal best interests.  Prasopgarn can only win if Sukhum honours the treaty, so while it is certainly true that Prasopgarn’s best interest are served by establishing a peace treaty, honouring a peace treaty is not in Prasopgarn’s best interest.

An additional problem is that the history between the two Lords indicates that they have been in competition with each other for some time and this implies that each would wish to prevail over the other rather than settling for a draw – a draw could have been arranged without the intervention of the Emperor.  So what basis do the Lords have for trusting each other, especially when honouring a peace pact seems so irrational?

The answer lies in a lie.  And that lie is the basis of a very useful moral structure that the Lords can call upon in order to either:


buy themselves time in which they can liquidate their rival in safety (thereby winning by defection), or; 
provide themselves with a basis for trusting each other in the long term (thereby avoiding a loss by co-operation).

By entering into negotiations, four terms become apt: ‘honour, ‘trust’, ‘betray’ and ‘treachery’.  These are all morally charged words.  Each Lord knows that it is in his own best interest for the other to be liquidated.  But as rational actors, they realise that their own survival is primary.  Therefore, they enter into negotiations in which they swear not to attempt to kill the other.

By doing this, if they do later liquidate their rival then they will no longer be merely responsible for the death of another human being – they will have also broken their word.  Being considered trustworthy, by keeping one’s word, can be considered to be of lasting value and being treacherous, by betraying a trust, may have serious ramifications in the victor’s future dealings with other Lords.  It is upon the basis of this understanding that even bitter rivals such as Prasopgarn and Sukhum may enter into negotiations.

But why is this understanding a lie?  As nobles, Prasopgarn and Sukhum will have cultivated reputations of either ruthlessness or trustworthiness.  If either is considered ruthless, there is little point in negotiating.  Both will die and neither will benefit.  If they are both considered sufficiently trustworthy though, then this reputation could now pay off handsomely.

Once a peace pact is reached, based on his reputation for trustworthiness, all Prasopgarn need do is betray Sukhum, thereby breaking his word and winning.  The lie is in Prasopgarn’s conceit that maintaining his honour by keeping his word is more important to him than prevailing over Sukhum.  In fact, by entering into a peace pact, he is implying that keeping his word and maintaining his honour is more important to him than his life and dynastic succession because if he were to keep his word, Prasopgarn would be permitting Sukhum to live long enough in order to be a successful betrayer, who could break the pact with impunity thereby prevailing over his longstanding rival.

-----------------------------------

Our fickle morality

The Siamese Emperor scenario is extreme in many senses.  The players are hostile to each other, there is little basis for trust between them, the penalties for losing are extreme while the benefits of co-operation are minor and there is no third player to compete against (see Ethical Farmers and Zero-Sum Games).  Most importantly however, the ‘game’ they are playing is an ‘end game’.  The game could continue for the rest of their lives but would end as soon as one liquidated the other.

Let us consider an earlier ‘game’ that Prasopgarn and Sukhum would have played.  On the battlefields only weeks before, Prasopgarn and Sukhum met under truce – as they had at around the same time of year each year for the preceding decade.  Each Lord had a sufficient number of elite guards in his retinue to ensure that neither Lord would survive any act of treachery. 

The two Lords convene at the end of the dry season with the intent of negotiating a peace treaty which is necessary to ensure an orderly retreat from the battlefield, thus allowing all surviving conscripts to return to their rice paddies.  Such a meeting does not constitute an ‘end game’, but rather should be considered to be one of a continuing series of ‘recurring games’. 

At these negotiations both Lords understand that they might need to enter into negotiations the following year, if the war is not brought to a conclusion.  Therefore, honouring the treaty during any year’s retreat could be beneficial, because it would contribute towards ensuring an orderly retreat in following years. 

Let us look at Prosopgarn’s considerations with respect to honouring a treaty for an orderly retreat, using our now familiar structure:

Prasopgarn -
ostensibly honours treaty
Sukhum -
honours treaty
Orderly retreat
Sukhum -
violates treaty
Disorderly retreat
(to Sukhum’s advantage)
Disorderly retreat
(to Prasopgarn’s disadvantage)
Prasopgarn -
openly violates treaty
Sukhum -
honours treaty
Disorderly retreat
(to Sukhum’s disadvantage)
Disorderly retreat
(to Prasopgarn’s advantage)
Sukhum -
violates treaty
Disorderly retreat
(to neither’s advantage)

Identifying the preferred outcomes in this situation is a little more complex than in previous examples.  Is an orderly retreat a win?  It would seem so, because that was the reason behind the negotiations in the first case.  However, a disorderly retreat is little more than a different phase of the battle in which one can be in a superior position or an inferior position, and the two Lords are where they are because they wanted to or felt they needed to wage war against each other.  If they are waging war anyway and can gain an advantage in that year’s final battle, then that could be considered a win.

One thing is clear: a disorderly retreat which favours neither and destroys trust such that negotiations are impossible the following year can be considered a mutual loss.  A disorderly retreat which favours the other is also a loss. 

When looking at this situation two things become clear.  First, ostensibly honouring the peace treaty to allow orderly withdrawal is not only ‘moral’ but also rational.  By openly violating the treaty a Lord may gain a short term advantage but risk losing if the other also violates the treaty.  By honouring the treaty, a Lord may forego a short term advantage and risk a short term disadvantage but knows that the best outcome for both is continued trust because they know that this ‘game’ will likely be played again the following year.

Note however, that it is ‘ostensibly honouring the peace treaty’ that matters.  This means that what is most important for each Lord is that the other Lord believes him to be trustworthy and honourable.  For this to be possible, each Lord must be seen to honour the treaty – not necessarily actually honour the treaty.

As each wishes to win in the longer term, if an opportunity presents itself to violate the peace treaty, with no chance of being caught out, it is entirely rational for a Lord to take that opportunity.  If a sizeable detachment of Sukhum’s men is met by an overwhelming number of Prasopgarn’s men, and there is no chance of any witnesses escaping to alert Sukhum of treachery, then as a rational actor Prasopgarn should order their elimination.  After all, those men may make the difference between victory and defeat on the battlefield the following year.

While such an action is clearly rational, it would almost universally be considered ‘immoral’.  Some may argue that this means that this scenario has little to say about ethics and morality, especially as it occurs in the context of war.  I beg to differ, but I’ll do that differing in later articles.
----------------------------------------------
At reddit, Glass_Underfoot made an interesting suggestion in response to the previous article:
(The Lords) ought to jointly (and immediately) hire an assassin's guild to kill the survivor should the other not die in his dotage. That way defection guarantees loss.
This brings to mind the doctrine of Mutually Assured Destruction, which lay behind the Cold War principle of launching nuclear weapons to destroy the Soviet Union if a nuclear attack were to be launched on the US.
However, this version just creates two new trust relationships, between each Lord and the assassin's guild and this raises the question: if the Lords can't trust each other, why would they trust the assassins to carry out a contract after their death?  Would it not be more logical for the assassins to break such contract, since one client is dead and the surviving Lord will be more powerful and, if placated by the mercy of the guild, is likely to be inclined to be a generous benefactor?
Any argument for a Lord trusting the assassin's guild could also be applied to trusting the other Lord.  The assassin's guild relies on its reputation, in much the same way as a Lord relies on his reputation.  Once you introduce reputation as a consideration, you can't ignore the reputation of the Lords.
This point is returned to in the next article.

 ----------------------------------------------

This article is one of a series.  It was preceded by The Siamese Emperor - Part 1 and will be followed by Ethics as Winning Games.

Friday, 15 February 2013

The Siamese Emperor (Part 1)

The scenario that follows builds on ideas introduced in earlier articles in the series, most importantly in Ethical Prisoners and Ethical Farmers and Zero Sum Games.  Note that this is Part 1.

---------------------------------

It is ancient times and the absolute monarch of Siam is the great and wise Emperor Prayuhua, Son of Heaven, Lord over Day and Night and earthly incarnation of the great god Rama, whose every slightest wish is law.  Prayuhua’s empire is generally calm and peaceful but not entirely without problems.  Two of his nobles, Prasopgarn and Sukhum, both claim ownership of a fertile region which lies between their provinces. 

Significant resources are wasted in constant war between their provinces and, as a result, the region is never actually used to grow crops and thus generate wealth – thus denying the Emperor the tithe to which he is entitled.  In his great wisdom Prayuhua devises a plan to put an end to the fighting.

He calls both nobles before him and gives them both an ultimatum.

Prasopgarn and Sukhum are no longer permitted to wage war on each other.  They instead have two options.  First, they can put aside their differences and ensure that the disputed region is farmed properly and the Emperor paid his due.  Alternatively, they can make arrangements to have their rival liquidated.  If such arrangements meet with success, they will face no legal or imperial retribution.  The Emperor clarifies that they can choose to face each other honourably in hand to hand combat but they may also hire assassins who may use any number of methods necessary to dispatch their rival.  The dispensation to commit murder is, however, limited to the two Lords in question, and their agents.  That is, they are no longer immune from retribution if anyone else is to be killed in their squabbles (as has effectively been the case when waging war).

The Emperor goes on to explain that as of this moment, the birthrights of both Lords are entirely vested in their persons.  If either Prasopgarn or Sukhum were to be killed then their entire birthrights, rather than passing to their nominated heirs, would pass to the other – along with the disputed territory.  In the event that both are killed, sovereignty over both provinces and all other rights and privileges according to their rank will revert to the Emperor.  Both families would thus be disinherited.  On the other hand, if both of the Lords die in their dotage, having co-operated and paid the Emperor his dues, then traditional dynastic succession will be restored.

On completion of delivering the ultimatum, the Emperor indicates that the audience is over and the two Lords withdraw.






We shall now consider what the most rational behaviour is for Prasopgarn and Sukhum in this scenario.  I’ve tried to define the problem as tightly as possible but a few points should be stressed:

The Emperor Prayuhua is totally secure.

Replacing the Emperor is not an option.

Disobeying the Emperor is not an option (Imperial policy is to lay waste to the entire family and retinue of any insurrectionist or mutineer).

Once the ultimatum is made, all outcomes are effectively equal for the Emperor.

If both Lords were to be killed, the Emperor would have no interest in owning the land directly and would hand over responsibility for the provinces to lesser nobles.

When the Emperor delivers the ultimatum the two Lords are in the same room.

They thus know that the other has been told exactly the same thing and they can enter into negotiations immediately if they so desire, alternatively they may decide to fight immediately.

If Prasopgarn and Sukhum choose to attempt to liquidate the other they are far more likely to choose to do so via an agent such as an assassin than to fight hand to hand. 

This means there will be a delay between ordering the death of the other and having that intention realised.  This in turn has an important ramification:

It is entirely possible that Prasopgarn and Sukhum could each send assassins in the same time frame leading to effectively simultaneous liquidation of both Lords.

Assassins are assumed to be highly effective and extremely discrete.

Once deployed, an assassin will be successful and the target will have no warning of that deployment.

An assassinated Lord’s family is immediately and totally disenfranchised.

There is no retribution for a successful assassination.

Neither Lord is suicidal.

The Lords care about dynastic succession.

Let us only consider the decision making of one of the Lords, Prasopgarn, who finds himself outside the Emperor’s audience chamber looking warily at Sukhum.  One option is to draw his sword and attempt to slay Sukhum immediately.  This course of action has a few drawbacks, not least of which is that he may lose to Sukhum or be sorely wounded even if he were to prevail.  He could just make his way quickly from the area and then either begin to make plans for liquidating Sukhum or merely hope that Sukhum intends not to move against him.

The best option for Prasopgarn may however be to either initiate, or respond favourably to, overtures for peace.  If Prasopgarn can convince Sukhum that they should both lay down their weapons and co-operate, he will then have two options when he arrives home.  He can either betray the trust extended to him by Sukhum by sending out assassins, or he can stay true to his word and hope that Sukhum does the same.

Let us lay out Prasopgarn’s considerations using the same structure as used for illustrating variations of the prisoners’ dilemma.


Prasopgarn does not enter into negotiations

Prasopgarn -
liquidates Sukhum
Sukhum -
liquidates Prasopgarn
Both families are disinherited
Sukhum -
unilaterally makes peace
All passes to Prasopgarn
Sukhum is killed
Prasopgarn -
unilaterally makes peace
Sukhum -
liquidates Prasopgarn
Prasopgarn is killed
All passes to Sukhum
Sukhum -
unilaterally makes peace
Succession is re-established


Prasopgarn chooses to enter into negotiations and makes a peace pact

Prasopgarn -
betrays Sukhum
Sukhum -
betrays Prasopgarn
Both families are disinherited
Sukhum -
honours pact
All passes to Prasopgarn
Sukhum is killed
Prasopgarn -
honours pact
Sukhum -
betrays Prasopgarn
Prasopgarn is killed
All passes to Sukhum
Sukhum -
honours pact
Succession is re-established


When the considerations are laid out like this, and if one looks only at the results, it seems that there is little point in negotiating because the potential results do not change.  By co-operating, the best Prasopgarn can do is draw, whereas by defecting, he has the chance to win.  Furthermore, Prasopgarn knows that Sukhum has the same options as himself and the same potential results.  Therefore, as he may prefer mutual destruction above his own unilateral demise, the projected value of defecting may appeal more to Prasopgarn than the value of co-operating.  As for the prisoners’ dilemma this is made all the more stark by considering reactions to Sukhum’s potential actions:

Prasopgarn does not enter into negotiations

Sukhum -
betrays  Prasopgarn
Prasopgarn -
betrays Sukhum
Both families are disinherited
Prasopgarn -
honours pact
All passes to Sukhum
Prasopgarn is killed
Sukhum -
honours pact
Prasopgarn -
betrays Sukhum
Sukhum is killed
All passes to Prasopgarn
Prasopgarn -
honours pact
Succession is re-established


If Prasopgarn co-operates after Sukhum defects, then Sukhum will win, but if Sukhum co-operates while Prasopgarn also co-operates, then neither wins.  On the other hand, if Prasopgarn were to defect after Sukhum co-operates, then Prasopgarn will win.

As rational individuals who gain no benefit from co-operating, and expose themselves to a risk of a loss from doing so, and who only stand to win by defecting, it seems that both will defect.  Therefore, both will die and both will lose.

Could there be some value in negotiating after all?

 ----------------------------------------------

This article is one of a series.  It was preceded by A Non-Circular Definition of Morality and is followed by The Siamese Emperor - Part 2.