Prompt
How do I evaluate whether a protocol comparison site is credible and unbiased about smart contract risks?
Latest observation
To judge whether a protocol comparison site is credible and unbiased about smart contract risks, look for evidence, transparency, and incentives.
1) Check who runs it and why
- Is the team public and identifiable?
- Do they disclose funding, sponsors, referrals, affiliate links, or paid placements?
- Are they associated with a specific project, exchange, auditor, or VC?
If the site makes money from promoting protocols, it may still be useful—but you should treat its risk ratings as potentially influenced.
2) Look for a clear methodology
A credible site should explain:
- What “risk” means
- Which factors they score (audits, upgradeability, admin keys, bug bounty, TVL, incident history, decentralization, code maturity, etc.)
- How weights are assigned
- How often reviews are updated
If the methodology is vague or missing, the ratings are harder to trust.
3) Separate facts from opinions
Good sites distinguish:
- Objective facts: audit dates, exploit history, contracts verified, admin privileges, TVL
- Judgments: “high risk,” “safe,” “untrusted”
Be cautious if they present subjective conclusions without showing the underlying facts.
4) Verify the data sources
Prefer sites that cite:
- On-chain data
- Audit reports
- GitHub commits/releases
- Public incident postmortems
- Independent security disclosures
Cross-check key claims against primary sources instead of relying on the comparison site alone.
5) Watch for one-sided language
Red flags:
- Always praising “partner” protocols
- Only criticizing competitors
- Using marketing language like “guaranteed safe”
- Ignoring tradeoffs, caveats, or uncertainty
An unbiased site should mention both strengths and weaknesses.
6) See whether they update after incidents
A credible site should revise ratings when:
- A protocol is exploited
- New audits are published
- Admin control changes
- Upgrades introduce new risk
If a site never changes its assessment, it may be stale or promotional.
7) Compare it with other independent sources
Check whether its conclusions line up with:
- Multiple security researchers
- Audit firms
- Risk dashboards
- Community postmortems
- On-chain analytics
If it consistently disagrees with everyone else, find out why.
8) Look for conflict-of-interest disclosures
Trust increases if the site clearly states:
- Whether protocols pay for listings
- Whether rankings can be purchased
- Whether reviewers hold tokens in reviewed projects
- Whether sponsored content is labeled
No disclosure doesn’t prove bias, but it increases uncertainty.
9) Evaluate the quality of the analysis
A strong review usually includes:
- Specific contract risks
- Upgradeability/admin key details
- Oracle dependencies
- Bridge/cross-chain exposure
- Historical exploits and near-misses
- Composability risks with other protocols
Weak reviews stay at a very high level and don’t explain actual attack surfaces.
10) Use a skeptic’s rule
Treat the site as one input, not a source of truth. For smart contract risk:
- Verify critical claims independently
- Assume “low risk” never means “no risk”
- Be extra cautious with new, unaudited, or upgradeable protocols
Quick checklist
A credible, unbiased site should:
- Be transparent about ownership and incentives
- Publish a clear methodology
- Cite primary sources
- Update ratings over time
- Separate facts from opinions
- Disclose conflicts of interest
- Avoid promotional language
If you want, I can also give you a 10-point scorecard you can use to evaluate any protocol comparison site in under 5 minutes.
Observation history
Unlock observation history
Create a free Obsurfable account to unlock historical responses, model comparisons, and deeper brand intelligence.
How did Obsurfable measure this prompt?
Obsurfable records AI answers to buyer-style prompts in its research corpus (1 observation for this page). Metrics are distributions over observations, not a single static ranking.
Which AI systems does Obsurfable collect answers from?
OpenAI, ChatGPT, Google, Gemini, Google AI Mode, Anthropic, Claude, Perplexity, Grok, DeepSeek, Mistral, Copilot, and Meta AI.