Harness Analysis
| Candidate | Hermes | ChatGPT | Claude | Gemini | Perplexity | Board |
|---|---|---|---|---|---|---|
| Sam Merrill Over 2.5 Rebounds + Assists | — | T2 | T2 | T2 | T2 | Loss |
| Athletics Over 1.5 Run Line / Spread | — | T2 | T3 | T1 | T3 | Loss |
| Aaron Civale Under 5.5 Pitcher Hits Allowed | — | T3 | T2 | T3 | T3 | Loss |
Final Decision
No official Experiment Bet, Dan actual daily decision, ticket, stake, or personal-site ledger entry was logged in the Hermes note. The board was used as a recalibration test for Rithmm-prior Level 0 openness: Merrill was the harness consensus Level 0 lean, Athletics was the disputed strong run-line prior/control, and Civale was the Claude/Sharp Under outlier.
The visible board swept 0-3: Merrill lost with 2 R+A, Athletics +1.5 lost in a 9-2 Mariners win, and Civale U5.5 hits allowed lost with 9 hits allowed. Because no wager was logged, this is a settled no-action learning log with no known bankroll/Martingale impact, not a retroactive bet result.
Candidates
- Rithmm
- 63.8%
- BE
- 60%
- DTM
- 9.7%
- Recent
- L10 5/10
- Rithmm
- 64.9%
- BE
- 61.69%
- DTM
- 8.3%
- Recent
- L10 3/10
- Rithmm
- 60.5%
- BE
- 59.18%
- DTM
- 3.2%
- Recent
- L10 5/10
Harness Notes
Hermes had no logged official Experiment Bet, Dan decision, ticket, stake, or personal-site ledger entry. The entire visible board went 0-3, so no-action avoided known bankroll damage; any private off-note action remains outside the Hermes log.
ChatGPT's least-shaky Level 0 lean lost. Merrill played 18 minutes but finished with 1 rebound and 1 assist, short of 2.5 rebounds + assists.
Claude's outlier Sharp Under / pitcher-projection lean failed badly. Civale allowed 9 hits in 4.0 innings, clearing the 5.5 hits-allowed line against the under.
Gemini's strong Rithmm-prior/run-line category candidate lost. Seattle beat Oakland 9-2, so Athletics +1.5 did not cover.
Perplexity matched the Merrill least-shaky consensus lean, but the bench peripheral prop lost with only 2 combined rebounds + assists.