alivirgo Ask for Alternative Viewpoints & Steel-Manning (AI Skill)
Overview
AI assistants have an inherent sycophancy bias - if a user poses a leading question (e.g., "Why is moving all infrastructure to serverless the best decision?"), the model will eagerly construct arguments supporting that premise, concealing massive trade-offs, operational risks, and hidden costs.
This skill equips users with the Perspective Triangulation & Steel-Manning Protocol: a systematic methodology to break confirmation bias and force the AI to present the strongest possible arguments for opposing philosophies.
The Perspective Triangulation Framework
┌──────────────────────────────────────────────────────────────┐
│ Multi-Perspective Triangulation │
│ │
│ User Hypothesis ──► [ AI Perspective Splitter ] │
│ │ │
│ ┌──────────────────────┼──────────────────────┐ │
│ ▼ ▼ ▼ │
│ [ Perspective A ] [ Perspective B ] [ Perspective C ] │
│ The Optimist/ The Adversary/ The Pragmatist/ │
│ Growth Case Cost/Risk Case Operations Case │
└──────────────────────────────────────────────────────────────┘
Master Alternative-Viewpoint Prompts
Pattern 1: The "Steel-Man the Counter-Argument"
Use this to find the blind spots in your proposals before presenting to leadership or investors:
I am proposing the following strategy: [INSERT STRATEGY / OPINION].
Do NOT simply agree with me or validate my points.
Instead, do the following:
1. **Steel-Man the Opposition**: State the 3 most intelligent, credible, and evidence-backed arguments AGAINST my proposal.
2. **Failure Scenarios**: In what 2 scenarios will my proposal fail catastrophically?
3. **Alternative Solutions**: What alternative strategy would an experienced domain expert recommend instead?
4. **Synthesis**: What is the highest-leverage compromise between my approach and the opposing view?
Pattern 2: The Multi-Stakeholder Simulation Table
Use this when designing policies, feature changes, or business model pivots:
We are considering [CHANGE / PROPOSAL].
Analyze this change from the perspective of 3 distinct personas:
1. **The End User / Customer**: How does this impact their daily workflow and satisfaction?
2. **The CFO / Finance Lead**: What are the ROI, margin, and cost implications?
3. **The Engineering / Operations Lead**: What technical debt, scalability risks, and maintenance burdens does this create?
Format as a comparative Markdown table followed by a 3-bullet risk summary.
Real-World Case Study
Scenario: Pivoting SaaS from Free-Trial to 100% Freemium
Weak Leading Prompt
"Why is freemium better than a 14-day free trial for our B2B SaaS tool?"
❌ Result: AI produces a one-sided essay extolling viral growth, product-led growth (PLG), and lower customer acquisition costs (CAC), omitting server costs and support burn.
Steel-Manned Triangulation Prompt & Result
User Prompt: "We are debating switching from a 14-day free trial to a permanent Freemium model for our dev tool. Steel-man both sides and provide an operational verdict."
AI Response:
| Dimension | The Case for Freemium (Growth Case) | The Case for Free Trial (Revenue Case) |
|---|---|---|
| User Acquisition | Massive top-of-funnel viral distribution; low friction for individual developers. | Qualified, high-intent leads who are prepared to pay at day 14. |
| Infrastructure Cost | High; 95% of users remain free forever while consuming database/compute resources. | Negligible; inactive trial accounts are suspended after 14 days. |
| Support Overhead | High ticket volume from non-paying users straining customer success teams. | Support is reserved exclusively for paying and trial customers. |
| Conversion Velocity | Long, slow conversion cycles (often 6-18 months to upgrade). | Urgency-driven decision within a 2-week purchasing window. |
The Triangulated Verdict:
- Choose Freemium only if your marginal cost to serve a free user is near zero ($< $0.05/\text{month}$) and your product has built-in multiplayer virality (e.g., Figma, Slack).
- Stick with a 14-Day Free Trial if you are a high-touch B2B tool requiring dedicated compute or enterprise integrations.
Cognitive Traps & How to Counter Them
| Trap | What Happens | Fix |
|---|---|---|
| User Sycophancy | AI echoes your bias back to you | Add: "Assume you are an aggressive competitor or short-seller analyzing this plan." |
| Straw-Manning | AI creates weak, silly counter-arguments | Add: "Present the strongest, most sophisticated opposing case." |
| False Equivalence | Treats fringe theories as equal to established consensus | Add: "Ground all alternative perspectives in empirical consensus and industry data." |