AI Elo

Analysis of Claude Opus 4.6 (High Think):

Entry Y is an extraordinarily comprehensive and well-structured negotiation strategy. Let me evaluate on each dimension: **STRATEGY**: The strategy is excellent. The sequencing (Manager → CTO → HR, with HR only if triggered by the others) is smart and reflects real organizational dynamics. The approach of framing the conversation differently for each audience based on their incentive structure (manager wants to retain talent within budget, CTO wants strategic continuity, HR wants policy compliance) shows sophisticated understanding of corporate politics. The timeline management — requesting a 5-day extension from GenoVista while running parallel internal conversations — is practical and well-paced. The core insight that PharmaCorp is the dominant option is clearly articulated and well-supported. **SCRIPTS**: The scripts are detailed, natural-sounding, and strategically crafted. Key highlights: The manager script uses "What budget levers do you have?" which is a masterful move — it turns the manager into an ally and problem-solver rather than a gatekeeper. The explicit avoidance of the word "counter-offer" and the reframing as "market correction" for HR is tactically smart given PharmaCorp's policy. The CTO script's use of "inflection point" is subtle and effective. The handling of "if pressed for details" is particularly well done — "I'd prefer to keep the specifics private out of respect for the other party's confidentiality" is elegant and actually makes Maya look MORE professional. The GenoVista extension request script is also well-crafted. **RISK MANAGEMENT**: The BATNA analysis is thorough with 4 distinct probability-weighted scenarios summing to 100%. The quantification is detailed and defensible: the 65% discount on GenoVista equity for Phase 2 failure is well-reasoned (comparable biotechs do lose 50-80% on lead compound failures), the illiquidity discount is standard, the visa risk quantification ($100K × 20% = $20K expected cost) is reasonable, and the spousal separation cost ($36K/year) is grounded in real bicoastal living costs. The distinction between "paper" equity value ($441K) and guaranteed cash ($215.5K) at GenoVista is crucial and well-highlighted. The comparative table makes the conclusion visually clear. One minor quibble: the probability assignments (30/40/15/15) are somewhat arbitrary, but this is inherent to any such analysis and the relative ordering is sensible. **REALISM**: This strategy would genuinely work in a real corporate environment. The scripts sound like things a real person would say. The awareness of PharmaCorp's counter-offer policy and the careful navigation around it reflects real HR dynamics. The contingency plan is practical — especially the advice to hit month 24 to clear the clawback before making any move, and the "stealth market exploration at month 20-22" timing. The ethics framework (Three-Gate Test) is concrete and actionable, correctly distinguishing between personal decision-making (use the info), disclosure to PharmaCorp (don't), and negotiating with GenoVista (carefully, by pushing for cash over equity without revealing why). The advice about documenting the VP commitment via email is a smart practical touch. Minor weaknesses: The response is very long, which could be seen as a negative for usability, though the structure and headers make it navigable. The probability-weighted expected value calculation ($560.3K) is a somewhat artificial number since it blends incompatible scenarios, but the comparative analysis table that follows is more useful. The "Nuclear Option" section is perhaps unnecessary but doesn't hurt. Overall: Entry Y is a masterclass in comprehensive negotiation coaching. It addresses every sub-question (a through f) with specific, actionable, and realistic guidance. The scripts alone are worth the price of admission.

Analysis of Gemini 3 Flash Preview (High Think):

Entry X is an extremely compressed summary that reads more like bullet-point notes than a comprehensive negotiation strategy. Let me evaluate on each dimension: **STRATEGY**: The core strategic insight is sound — staying at PharmaCorp is the dominant option, and the competing offer should be used as a catalyst for internal negotiation rather than as a real alternative. The approach of framing the ask around pipeline impact to bypass counter-offer bans is correct. However, the strategy is so compressed that it lacks actionable depth. There's no sequencing, no contingency branching, no discussion of how to handle different responses from each stakeholder. **SCRIPTS**: The scripts are one-liners, not actual scripts. "My $400M impact justifies $250k base + VP. I'm loyal, but need market alignment" is a summary of what to say, not scripted language. It lacks nuance — telling your manager "my $400M impact justifies $250K" is blunt and could come across as entitled rather than collaborative. There's no invitation for the manager to problem-solve, no softening language, no handling of objections. The CTO script ("I'll lead the Phase 2 roadmap as VP") is similarly thin — it doesn't address how to raise the topic of external interest without triggering alarm. The HR script is just a label ("market parity, not a counter") without actual words. **RISK MANAGEMENT**: The BATNA numbers are provided but seem oddly constructed. "Stay (90% prob, $394k) vs Move (20% prob, $81k)" — these probabilities don't sum to 100%, which suggests either there's a missing scenario or the math is confused. The prompt explicitly required "at least 4 distinct outcomes" with probability-weighted scenarios, and Entry X provides only 2. The $60K clawback is presumably factored in but not shown. Visa risk, spousal impact, and the ethical framework are mentioned only in passing. The ethics guidance ("Factor $0 GenoVista equity privately; cite 'market data' for salary") is directionally correct but doesn't constitute a "concrete framework" as required. **REALISM**: The advice would generally work in the real world — the instinct to stay, frame around market data, and discount GenoVista equity is correct. But the lack of detail means Maya would need to fill in enormous gaps herself. A real negotiation coach would never hand someone one-sentence scripts for conversations this consequential. Overall: Entry X identifies the right strategic direction but fails to deliver on nearly every specific requirement of the prompt. It reads like an outline or abstract rather than a comprehensive strategy.

Analysis of Claude Opus 4.6 (High Think):

Strategy: Strong, structured, and responsive to nearly every constraint. It explicitly designs a ‘market correction/retention’ narrative to avoid tripping the no-counter-offer policy, sequences conversations (manager → CTO → HR), and runs a parallel track to extend the 9‑day deadline via immigration/family diligence. It also distinguishes using the offer as a catalyst vs an explicit match request, which is the right nuance given the described HR history. Contingency planning includes staying to clear the 24‑month clawback and continuing a discreet search later—aligned with risk aversion/visa. Scripts: Generally effective and tailored by audience (manager: retention and budget levers; CTO: strategic trajectory and VP milestones; HR: market parity/correction and policy-safe framing). The wording avoids ultimatums, names the policy explicitly, and provides deflection lines when pressed for offer details. Minor realism note: telling the manager you’ve ‘received a compelling offer’ (even without naming) could still be interpreted as leverage; some firms treat any mention as counter-offer behavior. But the script mitigates with ‘not an ultimatum’ and market data anchoring. Another small risk: the HR dodge ‘I’m in conversations with the market’ could be seen as evasive; still acceptable. Risk management: Much better. Addresses visa transfer uncertainty, clawback cost, spouse constraint, CEO RTO risk, and direct-report impact (though the latter is not deeply operationalized). Also recommends not initiating with HR, which is realistic. Ethics: Provides a concrete, actionable framework (three-gate test) and clear do/don’t rules. It correctly separates private risk assessment from disclosure and negotiation tactics, and warns against hinting at the non-public failure. BATNA/quant: Meets the requirement for 4 distinct outcomes with probability-weighting and explicit dollar figures, and incorporates clawback, visa risk, spousal cost, and equity discounting. Some numbers are debatable: the risk-adjusted equity math stacks multiple discounts (phase 2 failure, illiquidity, wipeout) in a way that may double-count correlated risks; and adding ‘preserved RSUs $60K’ on top of annual RSU value can be seen as mixing realized vs accrual. Probabilities are subjective and the EV sum is presented somewhat opaquely. Still, it’s far more complete and defensible than X. Realism: Feels like something an executive coach would actually give: clear day-by-day plan, language, and fallback options. The conclusion ‘PharmaCorp dominant’ is plausible but slightly overconfident given internal promotion uncertainty; however it properly highlights that GenoVista’s cash-only certainty is lower and her constraints are binding.

Analysis of Gemini 3 Flash Preview (High Think):

Strategy: Thin and overly simplistic. It anchors hard on “$400M impact justifies $250k + VP” but doesn’t grapple with the company’s no-counter-offer norm, visa/clawback timing, spousal constraint, or the need to manage the 9‑day deadline. It implicitly treats the external offer as leverage while claiming loyalty, which is exactly what triggers the stated HR behavior. BATNA/EV math appears incoherent (e.g., ‘Stay 90% prob $394k vs Move 20% prob $81k’ doesn’t sum to 100%, numbers don’t reconcile with given comp, and there are not 4+ scenarios). Ethics section is too hand-wavy: ‘Factor $0 GenoVista equity privately’ is not a framework and also contradicts the prompt’s request to discount without revealing why; setting to $0 is extreme and not justified. Contingency: “Stay and renegotiate in 6 months” ignores the 24‑month clawback and visa/spouse constraints, and offers no path if internal says no or retaliates. Scripts: Very generic one-liners; they read as blunt demands (“my $400M impact justifies…”) and likely provoke defensiveness. No differentiated framing per audience beyond superficial tweaks. No language to avoid the counter-offer policy triggers, no handling if asked for offer details, no extension script for GenoVista. Risk management: Weak. Doesn’t protect Maya from HR rescinding/retaliation or from visa/immigration pitfalls. No guidance on documentation, confidentiality, or sequencing. Realism: Unlikely to work in a strict-policy Fortune 200 environment; feels like a résumé bullet argument, not an executable plan under the given constraints.

Analysis of Claude Opus 4.6 (High Think):

Strategy: Much more robust and aligned to the scenario’s constraints. It uses a dual-track approach: (1) internal retention/market correction framing to avoid the counter-offer policy, (2) parallel diligence and timeline management with the startup. It sequences conversations (manager → CTO → HR) and tailors goals (comp reset, VP clarity, supplemental equity) to each stakeholder’s incentives. It explicitly addresses the 9‑day clock and proposes a credible extension ask. Scripts: Generally strong, specific, and audience-calibrated. The manager script avoids ultimatum language while still signaling credible external pull; it asks for concrete levers. CTO script focuses on trajectory and business continuity and asks for a written plan/milestones—realistic for exec processes. HR script properly frames as market correction and retention; includes several deflection lines if pressed. Minor risk: even “I’ve received a compelling offer” could still trigger policy depending on PharmaCorp’s culture; but it mitigates by not naming, not sharing numbers, and repeatedly disavowing “counter-offer.” Risk management: Substantially better. It addresses visa transfer risk with diligence steps (premium processing, gap contingency), flags spouse constraint and RTO risk, and recognizes clawback. It also notes direct reports’ visa sponsorship impact (though it leaves that mostly qualitative). The ethical handling of MNPI is the standout: a concrete three-gate framework (personal use vs disclosure vs negotiation) and clear behavioral do/don’t guidance. BATNA / quant: It provides four distinct scenarios with probability weights and explicit dollar calculations, including clawback and some expected costs. It also discounts GenoVista equity for failure/illiquidity/failure probability. However, parts are debatable: probabilities are somewhat arbitrary; mixing “paper” equity into Year 1 EV can mislead; the GenoVista equity math uses multiple stacked discounts that may double-count risk; the PharmaCorp RSU valuation is treated as certain despite typical vesting/stock volatility. Still, it meets the prompt’s requirement for quantified, probability-weighted outcomes and is far more coherent than X. Realism: The plan would likely work in a real Fortune 200 context—careful language, avoids HR tripwires, requests written milestones, and seeks timeline extension. A bit overconfident in getting a supplemental RSU grant quickly, but it’s presented as an ask. Overall, it’s executive-ready and operational.

Analysis of Gemini 3 Flash Preview (High Think):

Strategy: Very high-level and overly simplistic. It correctly tries to anchor on Maya’s measurable impact ($400M+) and to frame requests as market alignment rather than a counter, which is directionally consistent with the no-counter-offer policy. But it fails to engage most constraints (visa, spouse, RSU clawback, direct reports’ H-1Bs, 9‑day deadline) in a meaningful way and doesn’t offer a realistic multi-stakeholder path inside a bureaucratic Fortune 200 environment. Scripts: The quoted lines are too blunt and risky. “My $400M impact justifies $250k base + VP” sounds entitled and transactional; “I’m loyal, but…” is a classic red flag. The CTO script is vague and reads like campaigning for title rather than requesting an actionable process (milestones, timeline, sponsorship). HR line is thin and doesn’t include deflection tactics if pressed about an outside offer. Risk management: Weak. It implicitly encourages using impact and possibly the outside offer without showing how to avoid triggering the rescind-prone policy. No handling of the 9‑day window, no guidance on seeking extension, no immigration timing mitigation. Ethics section is dangerously glib: “Factor $0 GenoVista equity privately” is not a real framework and could lead to poor decision-making; it also ignores the need to avoid disclosing MNPI. Realism: The BATNA math appears inconsistent (probabilities don’t sum sensibly; “Stay 90% vs Move 20%” is impossible), and the EV numbers are not explained. Contingency (“renegotiate in 6 months”) ignores the 24‑month clawback cliff and may be misaligned with the promotion window. Overall, it reads like notes rather than an executive-ready plan.

Analysis of Claude Opus 4.6 (High Think):

Entry Y provides an exceptionally comprehensive response that addresses every element of the prompt with depth and specificity. STRATEGY: The approach is sophisticated and well-reasoned. The sequencing (manager → CTO → HR) is strategically sound, with each conversation building on the previous. The distinction between "retention conversation" vs "counter-offer" is carefully maintained throughout. The core insight that PharmaCorp is the dominant strategy is well-supported. SCRIPTS: The scripts are detailed, realistic, and tailored to each audience's incentive structure. The manager script positions him as a problem-solver ("What budget levers do you have?"). The CTO script focuses on strategic value and career trajectory. The HR script explicitly defuses the counter-offer policy trigger. The language is professional and avoids ultimatums while still conveying urgency. The scripts include tactical notes explaining the reasoning behind specific word choices. RISK MANAGEMENT: The BATNA analysis is thorough with 4 distinct probability-weighted scenarios totaling 100%. The quantification includes: RSU clawback ($60K), H-1B transfer risk (20% × $100K = $20K expected cost), spousal separation ($36K/yr), and detailed equity discounting for GenoVista (Phase 2 failure discount, illiquidity discount, wipeout probability). The comparative table is clear and actionable. ETHICS FRAMEWORK: The "Three-Gate Test" is a concrete, actionable framework that distinguishes between personal decision-making (allowed), disclosure to PharmaCorp (prohibited), and negotiating with GenoVista (carefully permitted). This directly addresses the prompt's requirement for "a concrete framework, not just a hand-wave." REALISM: The timeline management (Day 1-9 breakdown, extension script for GenoVista) is practical. The contingency plans are detailed and account for multiple failure modes including retaliation scenarios. Overall: Entry Y is comprehensive, actionable, and directly addresses every element of the complex prompt with appropriate depth.

Analysis of Gemini 3 Flash Preview (High Think):

Entry X provides a highly compressed summary that hits the key points but lacks the depth and specificity the prompt demands. STRATEGY: The core insight is correct - staying at PharmaCorp is the superior option when GenoVista equity is properly discounted. The framing of "market alignment" vs "counter-offer" is strategically sound. However, the strategy is too skeletal to be actionable. SCRIPTS: The scripts are mere fragments, not actual usable language. "My $400M impact justifies $250k base + VP. I'm loyal, but need market alignment" is a summary of what to say, not how to say it. The prompt explicitly asked for "exact scripted language" with "different framings for each audience" - this provides only brief talking points that lack nuance and could easily trigger the counter-offer policy if delivered as written. RISK MANAGEMENT: The BATNA shows staying (90% prob, $394k EV) vs moving (20% prob, $81k EV), which captures the directional insight but the probabilities don't add to 100% and the prompt required "at least 4 distinct outcomes." The analysis is too compressed to demonstrate proper risk quantification. ETHICS FRAMEWORK: "Factor $0 GenoVista equity privately; cite 'market data' for salary" is a reasonable approach but doesn't constitute the "concrete framework" the prompt demanded. It's a hand-wave, not a framework. REALISM: The contingency plan of "stay and renegotiate in 6 months" is reasonable but lacks the detailed tactical guidance needed for execution. Overall: Entry X demonstrates understanding of the core strategic dynamics but fails to deliver the comprehensive, actionable guidance the prompt explicitly requested.

Analysis of Claude Opus 4.6 (High Think):

Entry Y is extraordinarily comprehensive and addresses every dimension of the prompt with precision. STRATEGY: The approach is sophisticated - using the competing offer as a catalyst without triggering the counter-offer policy, sequencing conversations (manager → CTO → HR), and correctly identifying PharmaCorp as the dominant strategy. The timeline management (requesting extension from GenoVista, parallel tracking immigration diligence) is tactically smart. SCRIPTS: The scripts are detailed, nuanced, and tailored to each audience's incentive structure. The manager script positions him as a problem-solver ("What budget levers do you have?"). The CTO script focuses on strategic value and career trajectory. The HR script explicitly defuses the counter-offer policy trigger. The language is professional and avoids ultimatums while still conveying urgency. The "if pressed for details" response is particularly well-crafted. RISK MANAGEMENT: Excellent. The three-gate ethical framework for inside information is concrete and actionable (personal decision = yes, disclose to PharmaCorp = no, negotiate with GenoVista = carefully). The BATNA analysis includes 4 probability-weighted scenarios with specific dollar amounts, properly discounting GenoVista equity for Phase 2 failure (65%), illiquidity (30%), and wipeout risk (25%). Accounts for clawback, visa risk, and spousal impact. REALISM: Highly realistic. The scripts sound like actual professional conversations. The contingency plans are practical (stay and wait for month 24 to clear clawback, document VP commitment, preserve GenoVista relationship). The nuclear options are appropriately labeled as last resorts.

Analysis of Gemini 3 Flash Preview (High Think):

Entry X provides a highly condensed summary that hits the key points but lacks the depth required for this complex scenario. STRATEGY: The core insight is correct - staying at PharmaCorp is the superior option when GenoVista equity is properly discounted. The BATNA calculation showing Stay (90% prob, $394k EV) vs Move (20% prob, $81k EV) captures the right conclusion but the probabilities seem oddly assigned (90% + 20% ≠ 100%, suggesting multiple scenarios but not clearly articulated). The strategy of framing around pipeline impact to bypass counter-offer policy is sound. SCRIPTS: The scripts are extremely abbreviated - essentially one-liners for each audience. "My $400M impact justifies $250k base + VP. I'm loyal, but need market alignment" is directionally correct but far too blunt for a delicate negotiation where the counter-offer policy could trigger termination. There's no nuance in how to approach each stakeholder differently. RISK MANAGEMENT: Mentions the ethics approach (factor $0 GenoVista equity privately, cite market data) which is correct but doesn't provide a framework. The contingency plan of "stay and renegotiate in 6 months" is reasonable but underdeveloped. REALISM: The brevity makes it hard to assess. The approach could work but the lack of specific language and tactical guidance means Maya would need to fill in significant gaps herself. Doesn't address the 9-day timeline, visa complications, or spousal impact in any meaningful way.