Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Strategies & Market Trends : The Art of Investing -- Ignore unavailable to you. Want to Upgrade?

To: Sun Tzu who wrote (10689)	12/11/2025 8:32:27 PM
From: Sun Tzu	Respond to of 10698

More OT stuff - the pain never ends! Debug trace below (one of many). 2) The big “AI mistake class” behind this (and how it showed up again) 2.1 Substring classification trap: "ineligible" contains "eligible" This is the exact kind of bug that causes practice-killing errors: I inferred a boolean eligible=True from tags containing ineligible, because my logic tested ("eligible" in tag) without excluding ("ineligible" in tag). That produced internally inconsistent fields (eligible=True + summary says ineligible), which then poisons any “experience proxy” stats. This is not just a one-off. This is a class of failures: label inference by naive string contains silent contradictions between fields downstream stats become wrong LLM then “reasons” from corrupted signals . Sigh!