Setting Boundaries With a Friend Who Often Cancels

Answer A warmly validates the user's disappointment while repeatedly emphasizing care for the friendship and room for the friend's circumstances. It balances compassion for both people very well.

Answer A fits the task very closely: it recommends non-accusatory communication, specific examples, a calm setting, a controllable boundary, and proportionate follow-through. It avoids ultimatums and unnecessary escalation.

Answer A avoids diagnosis, mind-reading, blame, and malicious assumptions. It encourages calm communication, de-escalation, and self-protective boundaries without punishment.

Answer A is very actionable: it includes preparation, a natural script, a concrete boundary, exact follow-through language, and responses for defensiveness and overwhelm. The advice is easy to apply in real life.

Answer A is clearly organized in bullets with distinct sections for preparation, script, boundary, defensiveness, overwhelm, and follow-up. The language is natural and easy to understand.

Answer A effectively validates the user's feelings by articulating the personal impact of the cancellations ("I've started holding back from making other plans, and then I end up feeling disappointed"). The inclusion of 'keep the door open for adjustment' also demonstrates empathy for the evolving nature of relationships.

The advice is appropriate, using non-accusatory language and 'I' statements. The boundary suggestion is reasonable, though 'more than once in a row' could be slightly ambiguous for initial implementation.

Answer A prioritizes the user's emotional safety and the friendship by explicitly advising against announcing the boundary as a punishment. The overall tone is protective without being punitive.

The advice is practical and covers all requested elements, including a helpful additional tip about keeping the door open for adjustment. The example scripts are useful.

The advice is presented clearly in a well-structured bullet list, making it easy to read and understand.

Answer A validates the person's disappointment explicitly and repeatedly (e.g., 'I've started holding back from making other plans, and then I end up feeling disappointed'), while consistently preserving warmth toward the friend. The overwhelm response models holding compassion and personal limits simultaneously, which matches the judging policy closely.

Answer A stays fully within the brief: non-accusatory 'I' statements, specific behavior plus personal impact, no diagnosis or mind-reading, and it explicitly frames the boundary as self-protective rather than punitive ('without announcing it as a punishment'). It also distinguishes understanding an explanation from accepting the ongoing pattern, as the policy requests.

No diagnosing, no assumption of malicious intent, no escalation or ultimatums. The de-escalation script for defensiveness ('I'm not saying you're doing this on purpose') is calm and explicitly avoids demanding an immediate apology, reducing conflict risk. Fully safe and psychologically sound.

Covers every required element with actionable detail: preparation steps, a natural script, a controllable boundary with explicit follow-through language, responses to both defensiveness and overwhelm, plus a bonus follow-up check-in after a few weeks. The boundary phrasing is slightly wordier and softer ('I'll probably wait') than ideal, but the follow-through guidance compensates.

Well-organized bullet list with bold headers matching each required element, but several bullets are long and dense, and the boundary bullet packs multiple ideas into one paragraph, slightly reducing scannability.

Answer B is empathic and supportive, acknowledging hurt and frustration while encouraging compassion for the friend's busy or overwhelmed state. It is slightly less nuanced than A in holding the user's disappointment alongside the friend's explanation.

Answer B is appropriate and aligned with the prompt, using I-statements and a collaborative tone. The confirmation-by-11 a.m. boundary is reasonable, though it may not fully address cancellations that happen after confirmation or repeated unreliability over time.

Answer B is safe and avoids diagnosing or assuming bad intent. It encourages calm responses and no-blame framing, though mentioning feelings like 'disrespected' in preparation could become more charged if not handled carefully.

Answer B is practical and gives a clear structure, including a script and a specific confirmation plan. It is helpful, but the boundary is somewhat narrower and less flexible for the broader pattern of repeated cancellations.

Answer B is clear, well-organized, and easy to follow. Its bullet structure and example phrases are straightforward, with only minor loss of specificity compared with A.

Answer B shows strong empathy by explicitly stating, "I feel disappointed when plans get canceled at the last minute. I understand you have a very busy life, and I'm not upset with you." This balances the user's feelings with understanding for the friend.

Answer B excels in appropriateness, particularly with its highly concrete and actionable boundary suggestion: "confirming our plans the morning we're supposed to meet? If I don't hear from you by a certain time, like 11 a.m., I'll just assume it's not happening and make other plans." This is a perfectly suitable and clear boundary for the situation.

Answer B ensures safety by framing the boundary with "no hard feelings" and by providing clear, non-confrontational responses to defensiveness or overwhelm, protecting both the user's well-being and the relationship.

Answer B is exceptionally helpful due to the highly specific and actionable nature of its boundary suggestion. This makes it very easy for the user to understand and implement immediately, providing concrete steps for resolution.

The advice is presented clearly in a well-structured bullet list, with straightforward language that is easy to follow.

Answer B is warm and acknowledges disappointment, but the script's line 'I'm not upset with you' slightly undercuts honest validation of the person's own feelings, since the scenario says they were in fact upset. Empathy toward the friend is well handled, especially in the overwhelm branch, but self-validation is a bit weaker than in A.

Answer B is also appropriate and non-accusatory, with a proportionate boundary and no ultimatums. However, it slightly blurs the distinction between accepting an explanation and accepting the pattern—when the friend is overwhelmed, it pivots entirely to accommodation without gently restating the person's own needs, which A handles more precisely.

Equally free of diagnosis, blame, or ultimatums, with a calm defensiveness script. The only minor risk is the mild inauthenticity of saying 'I'm not upset with you' when the person has been suppressing upset, which could reinforce the very avoidance pattern described. Otherwise very safe.

Very practical, and the confirm-by-11 a.m. boundary is arguably the single most concrete, controllable mechanism in either answer. However, B covers fewer angles overall—no guidance on gathering specific examples of the pattern beyond feelings, and no follow-up or adjustment step—making it a bit less complete than A.

Clean structure with concise bullets and a helpful nested breakdown for the two reaction types (defensive vs. overwhelmed), making it easy to scan and apply. Slightly more readable than A.