Self-doubts & self-dilemmas

[Philosophe]

It seems to me that we can doubt our own intentions & hence our willingness to carry out our own or others' positions. Hence we can have dilemmas with respect to ourselves.

Thus, we can have one-person dramas.

Here's a simple example.



I cannot create this example using Confrontation ManagerTM, since this does not allow self-doubts. Nor does it allow a stated intentions column in an options board with only one party. So I've had to create it using Windows Paint. But I would argue that it makes perfect sense.

-- The drunk takes the position that he ought to stop drinking (column D).
-- However, he does not intend to (column t).
-- He has a persuasion dilemma with himself (the arrow pointing to t) because he prefers to continue drinking. To get rid of this dilemma, he must either give in to himself by changing his position or find additional reasons, not present in the current options board, that will persuade him to change his intentions.
-- Also, he does not trust himself to stop drinking, even if he intended to (the question-mark).
-- He therefore has a trust dilemma with himself. To get rid of it, he must either give in to himself by changing his position or create a new options board by finding additional reasons why he should trust himself.

All this makes sense to me. Note that though the drunk's persuasion dilemma & trust dilemma are similar, they are not the same. The trust dilemma is about his perception of how he would behave if he decided to stop drinking. The persuasion dilemma is about the fact that though he agrees he should stop drinking, he hasn't decided to & prefers not to.

If he preferred to stop drinking while not doing so (ie, if the arrow's direction were reversed), he would have a rejection dilemma. There would then be a question-mark in his intention. To get rid of it, he would either have to prefer drinking (giving himself a persuasion dilemma instead) or decide to stop-- leaving himself with his trust dilemma.

This example concerns individual psychology. But exactly the same options board could model dilemmas of organizational management. Replace the drunk with Russia's President Putin. Replace his option "stop drinking" with the option "stop murdering political opponents in foreign countries". According to some analysts, Putin may face the same dilemmas as the drunk. Instead of problems controlling himself, he has problems controlling security service personnel that report to him & are supposed to take his orders.

Of course, these problems could be modelled by making the security personnel a separate party. In the same way, we could model the drunk's conscience & his desires as separate parties. But suppose we are focussing on Putin's interaction with foreign countries. Then it is simpler & better, I think, to model the problem in terms of Putin's self-doubt & self-dilemmas. Otherwise, the same model would be operating on different levels & would bring together parties (Russian agents & foreign countries) that are not in communication & know little about each other.

What do others think of this suggestion?

[drama theorist]

Philosophe's idea of self-doubts & self-dilemmas has a major drawback.

Philosophe not only wants to allow self-dilemmas, he also wants to abandon the rule that I can't have a persuasion, rejection or threat dilemma with someone whose position is compatible with mine. He abandons this rule by talking about a drunk's persuasion & rejection dilemmas with himself.

The difficulty is that if there's disagreement (meaning stated intentions don't suit every position) I normally prefer my position to the threatened future. In fact, Confrontation ManagerTM software assumes I do. Of course, I might be a "bully" or "flirt", pretending to want a resolution on my terms but actually preferring conflict. But once this is seen by others (ie, communicated) then the threatened future is seen as my position. So the software assumes I prefer my position.

The end result: when there's disagreement, Philosophe's idea would give every party a rejection & persuasion dilemma with itself. It can't get rid of this dilemma unless & until all parties' intentions agree with each other.

That seems wrong. I can see how the mere presence of disagreement might make me uncomfortable. But should it give me a dilemma I can't do anything about? That seems to undermine the concept of dilemmas being spurs to action.

[The management]

We've moved this topic to a new forum called "New ideas in drama theory".

It's interesting, but concerns ideas that aren't yet established & aren't supported by Confrontation ManagerTM software. So it seems better to put it in a separate forum, so as not to confuse current users.

Some other topics that have been discussed may be moved for the same reason.

[Nigel Howard]

This is how it seems to me. Philosophe's idea is a good one. Drama theorist's criticism destroys it.

I have a radical solution. Redefine persuasion & rejection dilemmas in terms of doubts rather than preferences.

The result, I think, is a simpler, more insightful & more useful analysis of dilemmas. Also, Philosophe's idea of self-doubts & self-dilemmas is rescued from Drama theorist's criticism.

At present, A is said to have a persuasion dilemma with B if B prefers the threatened future to A's position. My suggestion: define A's persuasion dilemma with B as the set of intentions that are controlled by B, not doubted by A & flout A's position.

This means I have a persuasion dilemma with you if I believe you can & undoubtedly will block my position.

Makes sense, I think. It pinpoints the location of my dilemma: what I have to persuade you of.

So much for my persuasion dilemma. But what if I doubt your flouting of my position?

In that case, I suggest, you have a rejection/threat dilemma. (Note that these two dilemmas become one, which kind of makes sense because, under the previous definition, they often coincided anyhow.)

What is your threat/rejection dilemma?

My doubt means I think you prefer not to flout my position. Therefore I see your flouting of it as a "threat", if a threat is defined as something you'd do only to punish me. And your dilemma is to convince me you mean your threat: you'd really do it.

Again, your dilemma is pinpointed. It consists of intentions of yours that flout my position & are doubted by me.

We'd still have arrows to show these dilemmas. An arrow pointing to the threatened future would show a persuasion dilemma. One pointing away would show a threat/rejection dilemma. The difference is that arrows now follow automatically, given doubts.

The user would simply input parties, options, positions & intentions -- plus doubts. Stronger results would then follow from fewer assumptions.

Preferences would still be part of of the picture, inasmuch as it's clear that all doubts come from beliefs about preferences. I doubt your (actual or potential) intention because I think you'd prefer not to carry it out. But doubts would be used instead of preferences because, being communicated, they're observable. They're the observable manifestation of preferences.

I'd suggest some other changes to go along with these. First, we no longer need the positioning dilemma. This arose because, having defined dilemmas in terms of preferences between columns, questions arose about preferences between all pairs of columns: what if I preferred the threatened future to my own position? What if I preferred your position to my own? Under these new, simpler definitions such questions don't arise.

Also, what we've called the cooperation dilemma should be classified as a response to a trust dilemma, rather than as a dilemma on its own.

Each dilemma A has with B should be seen as possibly generating a response by B. B's response is generated by A not "giving in" to the dilemma (by weakening its own position or intentions), but "fighting" it (by trying to change own & others' beliefs & preferences).

Since the threat & rejection dilemmas would be made one, we'd be left with three dilemmas: persuasion, threat & trust. The response to A's persuasion dilemma would be B fighting or giving in to the potential threat dilemma caused by A's fight. The response to A's threat dilemma would be B fighting or giving in to the potential persuasion dilemma caused by A's fight. And the response to A's fighting a trust dilemma would be B fighting or giving in to the potential trust dilemma caused by A's fight (inasmuch as A would fight its trust dilemma by making B think, "Maybe A's mistrust of me makes it doubtful if I can trust A as I'd like to.")

Clears up a lot of problems. In particular, makes dilemma-elimination necessary & sufficient for all parties to agree & trust each other.

[Politico]

I like Nigel's idea. It solves another problem I've had -- how to model the fact that a party shares another party's position but disagrees with the threats it's making to achieve that position.

I didn't realize at first that it solved this problem. Instead, I thought I had a criticism that destroyed Nigel's idea.

Let me explain.

My criticism was this. Suppose I make a threat or promise to pressure another party to accept my demands. For example, suppose I, as the US, pressure Iran to give up nuclear development by threatening to bomb nuclear sites & promising to lift sanctions.

My position, it seems, should be that if my demands are met I won't carry out my threat & will carry out my promise. But given disagreement over my demands, my stated intention must be to carry out my threat & not my promise.

If my threats & promises are credible, this gives me a persuasion dilemma with myself.

Here's the US-Iran model. (BTW, since I'm using Windows Paint to adjust the model, I also followed a suggestion of Manuel's & put the intentions column to the right of the board, instead of the middle. I find it less confusing.)



In regard to bombing Iran, the US is making the "non-commitment" threat: "We're not taking this card off the table." This way of making a threat makes it quite credible (you've made no commitment, so can't break your word) but rather weak: the other side will discount the threat, seeing that you're not prepared to risk making it credible.

Iran nevertheless has a persuasion dilemma with the US: it wants them to lift sanctions & commit not to bomb. The US has one with Iran: it wants them to cease enrichment. But in Nigel's new system, the US also has a persuasion dilemma with itself. What can this mean?

It could mean the US has internal critics that don't like the US threats. This is like the self-dilemma of Philosophe's drunk.

Fine in that case, but what if it hasn't? This seemed to present a problem, till I realized that if the US "gave in" to its persuasion dilemma (by putting dashes in its position), it would just be a way of asserting its threats & promises. It would mean "We don't take a position on whether we carry out these threats & promises. You choose."

Just the way parties do make threats & promises. "Whether we do this or not is up to you." So it seems to me that for a party to put dashes in its position against its threats & promises should always be taken to mean it has no internal conflicts over this threat/promise. Here's the model in the US-Iran case.



Likewise, another party that agrees with your position but not your threat or promise can show this by taking a position on your threat/promise. In the US-Iran case, the Security Council agrees with the promise (to lift sanctions) but not the threat (to bomb). Arab countries disagree with both (they think the threat shouldn't be made & the promise shouldn't be conditional). Here's the model.



What dilemmas do we have here? The Security Council (S) & Arab countries (A) agree with the US (U) that Iran (I) should cease enrichment.

U, S & A have trust dilemmas with I (shown by the question-marks) & persuasion dilemmas as well (they believe Iran won't cease enrichment).

I has persuasion dilemmas with U (believing that U won't lift sanctions or make a commitment not to bomb).

In addition, S & A have persuasion dilemmas with U. Both think it shouldn't threaten to bomb.

A also thinks it shouldn't promise to lift sanctions -- it should do so unconditionally.

I conclude that Nigel's idea allows us to model internal & external disagreements within a coalition as to which sanctions should be applied. That's an advance, in addition to making dilemmas fewer, simpler & clearer.

[Manuel]

Politico,

I am not sure I follow your reasoning from the options boards. If you dash the US's options of bombing enrichment facilities and sanctioning Iran non-militarily (in the US position), I understand that's because the US is not making any explicit statements about those two options, correct? If the US communicates that they wish Iran to stop enriching uranium or else they will have to resort to sanctions, surely you have to mark that option as 'rejected' in US's position, not as dashed?

As I see it, not saying anything about military actions (dashing that option) may mean that there is internal disagreement about that option. So the US may not want to make the threat explicit yet, until internal consensus is reached. It may also mean that they don't want to make it explicit to avoid losing the power of the threat by spelling it out. In these two cases you avoid communication of what you intend to do regarding a particular option, and therefore you can dash it. But if the option is communicated, then it has to be even accepted or rejected.

Do you agree? Is this the way you are using the dash in your options board?

[Politico]

Manuel, I think we need to be clear about the meaning of a dash in a position column.

I agree that if Party A says nothing about option x, that will often mean we should leave a dash against x in A's position. But not if it's obvious what A's position is going to be. In general, a party "saying nothing" has to be interpreted in light of the fact that we don't usually waste time stating the obvious. Because of this, we must always answer the question -- what does this party communicate by saying nothing?

For example, if I point a gun at you & say "give me your wallet", only one option is openly referred to: your option to give me your wallet. My position is openly stated as "that option should be adopted".

Nothing else is said. But a lot of important things are communicated without being said -- because they're obvious. They're communicated because the "default" (obvious) thing is assumed about them unless/until something different is said.
These unsaid, communicated things are:

-- I have an option to shoot you.
-- Your position is that I shouldn't shoot you.
-- Your position -- unless & until you say or act differently -- is that you shouldn't give me your wallet.
-- That's also your stated intention -- unless/until you say or act differently.
-- My stated intention (credible or not) is to shoot you.

My position on my "shoot" option (as distinct from my stated intention) is also, surely, being communicated in this unsaid way. I'm communicating that my adoption of this option is contingent on your adoption of the "give wallet" option!

But that surely means I'm leaving my option open (ie, making it a dash in my position). I'm leaving it to be filled in by your position -- once that has been made compatible with mine.

What I'm communicating is: "I know/assume your position is that I shouldn't shoot you. That's fine, I'll let you decide that -- if you let me have my position."

The way to model this is to leave the "shoot" option as a dash in my position, because the very function of a dash in a position column is to allow positions to be compatible. By leaving a dash in an option, a party is saying, "To get an agreement, I'm willing for this to be decided by others' positions."

So my position "adopt wallet-giving; leave open shooting" says: "I'll accept your position on shooting if you accept mine on wallet-giving."

Here's the model. (You'll note that I've modelled the fact that I really don't want to shoot you.)



I therefore claim that the general rule (though it may not have been adhered to in the past) is that an option used as a threat or a promise should be "left open" -- dashed-- in the position of the threatener or promiser. This expresses its contingency. I think it will be interesting to go through previous models (we have a rich treasury on this site) & see how the rule works with them. It seems to me that it must improve them.

The exception to the rule, discussed in my post, is when a party has internal conflict over whether an option should be used as a threat or promise.

[Philosophe]

To me, Politico's approach seems theoretically sounder, but I think Manuel is right; it'll clash with the way users want to do dt.

How does the problem arise? Because in making a promise or threat, I'm making my option contingent on what others do. The question is, how to model this in a way that comes naturally to users without giving me a spurious persuasion or threat dilemma with myself.

Politico's solution (to "leave open" the option in my position column) appeals to me for the reasons Politico gives. Another argument for it is that a position is defined as preferred to the threatened future if some future belonging to the position is preferred. So I appeal to your preferences (in theory) just as much by leaving my "promise" option open as I do by taking the position that I'll fulfil the promise. Also, I clearly leave it up to you whether I do so or not -- rather than seeming to demand, request, or propose it (alternatives which could, as Politico suggests, be used to model the position of one who thinks the promised action shouldn't be conditional).

It's typical of a good theoretical answer that there are many good reasons for it.

Nevertheless, Manuel is undeniably right in pointing to the importance of what people explicitly say. Real people often stress eagerness to fulfil their promises -- provided the other side will meet the necessary conditions. Such stress on my promise (or preference for not carrying out my threat) seems inadequately modelled by "leaving it open".

Sometimes not-leaving-it-open may lead the other side to see the contingent action as unconditional, but usually it's clear it doesn't mean this. Parties that want it to be unconditional (for example, Europeans that wanted unconditional abandonment of the US threat to invade Iraq, or nowadays of the threat to bomb Iran) are the ones that demand clarity, asking for statements that "under no circumstances" will the threat be carried out.

Why do we "accentuate the positive" in this way? Because it generates positive rather than negative emotion, & hence is often the best way to get agreement.

Politico may be right that the message "this is conditional" is best modelled by putting dashes in my position, but users will be puzzled to be told they mustn't model positions as they're clearly & emphatically stated.

I therefore suggest a compromise. It messes up the simplicity of the theoretical model a bit. Politico will have to give up his suggested way to model disagreements over whether threats ought to be made or promises ought to be conditional. But I think it'll suit users better.

Let's leave theoretical purity to theorists.

My proposal. Use Nigel's definitions of persuasion & threat dilemmas (see his post above) when they are with another party.

-- I've a persuasion dilemma with you over options where I don't doubt you'll flout my position.
-- You have a threat dilemma with me over options where you say you'll flout my position, but I doubt you will.

When it comes to persuasion & threat dilemmas with myself, add a complicating condition. Say that to have a persuasion or threat dilemma with myself, I must flout my own position without at the same time flouting the position of another party that flouts my position. In other words, I must be flouting my own position without the excuse that I'm pressuring someone else to [i]stop flouting it[/i] (so that their compliance would mean I'd stop flouting it too).

As said, this makes the definition slightly more complicated, but should make the program more intuitive for the user. It won't throw up dilemmas that are puzzling to interpret, as I fear the simple formulation would.

Take Politico's example, where he holds up Manuel for his wallet:



Both Politico & Manuel have threat dilemmas with each other: Politico doesn't want to shoot & Manuel will happily give up his wallet -- he just hasn't agreed to yet. (Under the existing Confrontation ManagerTM system, they'd also have rejection dilemmas. But under the new system suggested by Nigel, these are the same as their threat dilemmas: the arrows show both or either.)

Fine. The problem is to avoid Politico having a threat dilemma with himself. Politico's own suggestion: Politico puts a dash in the top left cell. But suppose Politico is swearing he doesn't want to shoot, he just wants the wallet? Users will want to make this clearly expressed position clear.

My suggestion: the program should recognize that Politico's "threat against himself" is also a threat against Manuel, who's flouting Politico. So it shouldn't give Politico a self-threat dilemma -- & so I haven't drawn an arrow to show one -- until we move to this options board.



Here, Manuel has convincingly said he'll hand over his wallet, making this both his position & intention. Politico trusts him (no question-marks in his position), so what's his problem? Confused, he's still pointing his gun at Manuel, threatening him. He might be threatening him not to pursue him or jump him -- but these options are not on the board. No, he has a self-threat dilemma, as well as a threat dilemma with Manuel. He gets over both by ceasing to threaten to shoot, instead running off with the wallet. There is agreement & a resolution of the drama.

My suggestion means that Politico would have to use other ways to model disagreement with making a threat or conditional promise. In his model, he'd have to give the Security Council & Arab countries specific options whether to " support US bombing threat" & "support sanctions". This is less theoretically neat, but I suggest, more realistic in terms of the way users will want to use the system.

[Manuel]

I like the idea of simplifying the number of dilemmas, because it can make DT more accessible to others, although it doesn't look like the change is going to be without pain. The new approach to options board analysis means that the focus is now on doubts about intentions, and I find it difficult to grasp the implications using Confrontation Manager - which focuses on preferences. So spare me if I make mistakes.

My understanding is that, under the new approach, conflict dynamics is based on beliefs and doubts about the capabilities and intentions of others to block my position. That's clean and, as Nigel points out, it bases modelling on what is actually communicated - doubts.

My first concern, though, is whether eliminating dilemmas under the new approach subtracts validity to options board analysis. Let's look at the case shown above of Iran enriching uranium. The usual way of modelling that conflict (including the issue of the disagreement over the means of coercion) would be this:



This shows the same persuasion dilemma for Iran wrt the rest, and the trust and persuasion dilemmas of the various parties wrt Iran. In my options board I also see that U, S and A have rejection dilemmas wrt each other. These dilemmas stem from the fact that they all prefer each other’s positions (in which Iran won’t enrich uranium, even if they have doubts about them complying) to the threatened future (in which Iran will certainly continue the enrichment programme while being subject to sanctions). So even if the way of addressing the rejection dilemmas points out at the disagreement over the means of coercion, the focus is on the preference of position, not on the specific beliefs and doubts that cause the disagreement. In that sense, Politico’s model captures the situation better, by focusing on the beliefs that the parties have on others flouting their positions regarding coercive measures.

Politico’s model seems to work better than the traditional approach to some extent. However, the main issue is that of dashing threats and promises that are communicated as adopted or rejected. It would be good if we found a simple but intuitive way of modelling this. I am interested in understanding how to model the Iran-U/S/A confrontation following Philosophe’s comments. This is how I’d do it:



I am not sure of the results of this options board, because Confrontation Manager forces me to think in terms of preferred positions and the board gets quite complicated with four characters. Philosophe, is this how you suggest to model this confrontation?

[Politico]

Manuel, I think there's a serious confusion here. I never suggested that the contingency nature of threats & promises should be represented by dashes in the intentions column. I suggested dashes in the position column of the threatener/promiser.

I suggested this as a way for a threatener/promiser to get rid of a persuasion dilemma it would have with itself if it its position was that its threat shouldn't be carried out. It would have this dilemma with itself because its stated intention (to carry out the threat) would be flouting its own position.

This problem wouldn't be got round by dashes in the intentions column! A dash here means "I won't commit myself". But I still have a persuasion dilemma with a party that won't commit itself not to carry out a threat against me!

A dash in the positions column, on the other hand, means "this is up to you." Position-column dashes mean "I leave this open for others to choose". So the threatener wouldn't be stating a position as to whether the threat should be carried out. It would be leaving this to be decided by the party being threatened.

So my way of modelling the enrichment problem was the same as yours in regard to the intentions column. I suggested putting dashes in the US position against the options to bomb & sanction Iraq, & a dash against the "sanction" option in the Security Council's position. This got rid of the need to define options like "support sanctions" & "support bombing threat".

However, I must say I'm persuaded by Philosophe's argument that this model would seem unnatural to some users. So I accept Philosophe's suggestion as to how to get round the problem -- which is to redefine the persuasion dilemma so that I don't have a persuasion dilemma against myself if my intention that flouts my own position also flouts the position of another party that flouts my position somewhere else.

Under Philosophe's suggestion, I think your first model of the enrichment problem would be the right one.

[Manuel]

Thanks for the clarification about the dashes. The bit that is still not clear is that of the self-dilemma. It seems to me that, by default, every time you make a threat you give yourself a self-dilemma, because you don’t really want to implement that threat, it’s just a last resort if your demands are ignored. So if you introduce the concept of self-dilemmas, you always have to dash threats, unless you reject them. Is this correct? If so, this may be a rule that would help model confrontations, in spite of the fact that sometimes parties may be making statements about a threat that require it to be adopted or rejected in their position.

I have tried to model a case I heard this morning in the radio, using your approach, to see if I understand it. This is the case of a country ruled by a dictator that refuses to give way to the demands of Western Democracies with the excuse that after the elections, extremists parties may end up ruling the country and threatening those very same Western Democracies – the case given was Egypt, but it could equally apply to Pakistan and other countries. The Western Democracies face the dilemma of impose sanctions to the anti-democratic regime, or else support the dictator to keep extremists under control. This is the board:



According to the approach described in this topic, both parties experience persuasion dilemmas, as they are flouting each others positions. The Western Democracies also experience a threat dilemma, because the dictator does not believe that they will cease to support him, for fear of extremists. Both parties also face trust dilemmas.

In your view, is this analysis correct? Would you model this in a different way?

[drama theorist]

Manuel, your analysis is in line with Politico's suggested approach. I think it's the right analysis.

Before discussing your application, I want to say I support Politico's clean, simple, theoretically correct solution to the "problem" of spurious self-dilemmas.

I think (as you yourself suggest) that this so-called "problem" is easily solved by presenting dt analysis in the right way. We need to recognize that "adoption, rejection & leaving open" mean different things depending on whether we're talking about a party's position, its intention or the user's assumption about a consequence. The user's edit of a cell in the options board needs to be directed accordingly.

When editing a position cell, the user should be asked to choose one item from a check list looking like this:

Under A 's position, this option is:
-- an absolute requirement
-- excluded altogether
-- negotiable or conditional.

The rationale for this way of presenting it is simple. If I indicate that something is not my final, non-negotiable, unconditional position, I implicitly invite others to decide how it's filled in. My current position is therefore what I currently present as my minimum requirements or "needs".

This should solve the problem of "spurious" self-persuasion & self-threat dilemmas. If A is making a promise or threat, it's using an option conditionally. Hence, the user is explicitly directed to put a dash against that option in A's position. In your example, this is the reason why the dictator puts a dash against "help in war against terror" and the West puts a dash against "impose sanctions".

It's interesting, though, that the first of these dashes might be wrong in the case of some dictatorships. Saudi Arabia, for example, credibly states that helping the West against terror is unconditional, being in its own interest. At the same time, it tries threatening not to do so (eg over the BAe contract investigation). The result: its threat is incredible, & it has a self-threat dilemma. The dilemma: either don't use this threat or try to make it credible, starting with making it credible to yourself.

I'd say it's a virtue of the new approach that it asks such questions & enables us to model the answers. (Note that when your threat is incredible to me, it's because I don't think it's credible to you. I think you don't think you can or want to carry it out.)

Another virtue is that it makes your persuasion dilemmas explicit in terms of options -- they're credible intentions: by the dictator, not to hold free elections; by the West, not to support dictatorship.

Also, rejection dilemmas (which are now the same as threat dilemmas) can now exist alongside persuasion dilemmas -- ie, I can have a rejection/threat dilemma with you regarding one option, while you have a persuasion dilemma with me regarding another. This makes sense to me.

Let me also say this. I think it's important, when doing the 3-dilemma analysis, to analyze B's response to A's reaction to its dilemma with B, as well as A's reaction. The rationale is this:

-- A's dilemma is whether to "give in" or "react" to its dilemma.
-- If A reacts, B faces a putative dilemma. It's
-- -- a putative threat dilemma if A is reacting to a persuasion dilemma,
-- -- a putative persuasion dilemma if A is reacting to a threat dilemma,
-- -- a putative trust dilemma if A is reacting to a trust dilemma.
-- B's response, therefore, consists of "giving in" or "reacting" to this putative dilemma caused by A's reaction.

(The response to a trust dilemma is what we used to call a cooperation dilemma, but I'd call it a response rather than an independent dilemma because it only arises if A does not "give in" to its trust dilemma.)

Let's apply this to your model. How might W react to its threat dilemma? It could try to make sanctions credible by showing that public anger or its own interests demand them. This gives D a putative persuasion dilemma, to which it responds either by giving in (putting a dash against "impose sanctions", thereby saying "we don't care") or by reacting -- eg, arguing that W would be better off if it did not impose sanctions (& also supported dictatorship), saying, "we're a bulwark against terror", etc.

The general point is that emotions, rationalizations & redefinitions of the frame go both ways when A has a dilemma with B.