@AJoel, & @meganomaly

I'd like to flag a pattern I noticed on the leaderboard for Zindi Team to verify. I'm not accusing anyone — just presenting what's publicly visible and asking for an official check, since the rules clearly prohibit multiple accounts per user.

1. Metric-identical pairs (identical to 9 decimal places):

• SaSalonim / SaSaSalonim — identical composite (0.925919356), AUC (0.948131724), and F1 (0.911111111), submitted the same day, 4 submissions each
• MetaLearner / okech — identical composite (0.930086368), AUC (0.944896772), and F1 (0.920212765), same day
• F1Hugh / AucHigh — identical AUC (0.962779877) with F1 differing by only ~0.0004 (essentially the same probability ranking with 1–2 labels flipped at the 0.5 boundary)

2. Six fresh accounts with the exact same F1 (0.925531914 = 87/94): foneo, Mirandaw, Siram, YouAnR, TKrose, KAteAi — all created within the last ~16 hours, all with only 3–4 submissions, all with this identical F1 value. On ~309 public rows, that F1 corresponds to a unique TP/FP/FN combination, so six independent models landing on it within hours seems extremely unlikely.

3. Fresh low-submission accounts matching established accounts:

• btToram (4 subs) and Guelmbaye (54 subs) — identical F1 0.917333333
• biot (4 subs) and GIrum (99 subs) — identical F1 0.917127071

4. Timing pattern: nearly all of these accounts submitted within two bursts (~15–16 hours ago and ~3–4 hours ago), with scores forming a tight band (0.9269–0.9406).