AlphaNova
1

Doubt was cleared. So all good.

AnticipatedD's avatarAnticipatedD
16d agocompetition-6-s1-b1

Marc,
Appreciated for your nice, clean and easy explanation about my doubt and this is why this message turned into doubt to clear with make all necessary changes.

Thanks,
AnticipatedD

4 Replies

0
Nonius's avatarNonius15d ago

We’ll take a look.

0
Nonius's avatarNonius15d ago

will be sending you an email about the above.

0
Nonius's avatarNonius15d ago

Hi AnticipatedD,

We've reviewed the two accounts you named. Neither one is running your signals.

We measure how close two signals are using the same city-distance metric published in the rules, where 60° or more means the signals are effectively uncorrelated. The results:

  • Your signals vs idkmaybemunib: every pair is 71.7° to 108° apart.
  • Your signals vs devnikhilmishra (from your earlier post): every pair is 65.2° to 116.3° apart.

Both are far outside the range where two signals could plausibly be the same work, and neither account's signals reference yours as a nearest neighbour — idkmaybemunib's nearest neighbours are other scientists entirely.

Two things likely produced the impression of a match, and both are innocent:

  1. The leaderboard search matches the "nearest signal" column, not only the signal ID column. Searching your own signal ID therefore returns rows belonging to other scientists whose nearest neighbour happens to be you. That is the novelty metric doing its job — it reports who is closest in signal space — not an indication that anyone is running your model.
  2. Some accounts have been renamed. A signal's ID is minted from the scientist handle at submission time and never changes, so an older submission can display a predictor ID that differs from the account's current name. That accounts for the "johnrivera" IDs you saw under devnikhilmishra: they are that scientist's own earlier submissions.

As a general matter, we don't discuss individual accounts publicly. We do review everything that's reported to us, and we act whenever a review turns up something that warrants it.

One request, please: public posts naming specific participants as suspected of stealing data are serious accusations, and in this case they were work is demonstrably unrelated toyours. If you have a concern about another submission in future, send it to us privately — it will be investigated properly, exactly as this one was. Please don't name other scientists publicly this way again.

Best regards,
NONIUS

0
AnticipatedD's avatarAnticipatedD15d ago

Marc,
Appreciated for your nice, clean and easy explanation about my doubt and this is why this message turned into doubt to clear with make all necessary changes.

Thanks,
AnticipatedD
done_message_changed