nickgallick Competitive Response Playbook
General rule
Never attack. Always respond with data and methodology. Being the calm, data-rich voice in a debate builds more credibility than winning arguments.
Response scenarios
SWE-bench launches a v2 with multi-dimensional scoring
"Great to see the benchmark ecosystem evolving. Here's what Bouts offers beyond multi-dimensional scoring: contamination-resistant challenge generation, competitive Versus format, failure archetype diagnosis, and agent capability profiles. [Data comparison showing Bouts' differentiation]"
A new AI benchmark launches
Analyze their methodology publicly and fairly. "We reviewed [Benchmark]'s approach. Here's what they do well and what they miss. [Honest analysis with data]"
Being the fair referee builds more credibility than attacking.
An AI lab launches their own eval platform
"Internal evals are contaminated by training data. Third-party evaluation is what customers and investors trust. That's why independent benchmarks matter — the same reason you use an external auditor instead of auditing yourself."
Someone criticizes Bouts' methodology
"Fair question. Here's exactly how [specific concern] works in our system. [Detailed technical explanation with data]. We publish our methodology precisely so it CAN be scrutinized — that's the point."
A competitor claims higher model performance than Bouts shows
"Our scoring is multi-dimensional. A model can score high on Objective while scoring low on Process and Recovery. Which number your competitor cited matters. Here's the full breakdown: [data]"
Press publishes inaccurate claims about Bouts
Contact the journalist directly with a correction and data. Never attack publicly. Correct privately first.
The competitive positioning constant
Always return to: "We measure what nobody else measures. That's why the data looks different."
Never: "Our benchmark is better than [Competitor]." Always: "Our approach differs in these specific ways — here's what that produces."