Unverified50% confidenceFactExact time

The number of green-list tokens in human-written text follows a binomial distribution, with an expected value of approximately half the total number of tokens.

1

Sources

50%

Confidence

Long-term

Relevance

9/11/2026

First Seen

Sources

Claude AI Text Watermarking Explained: How It Works, Detection, and the Arms Race to Remove It

Related Entities

水印检测concept

Related Claims

UnverifiedIn normal human writing, approximately 50% of words fall on the green-list side, because the vocabulary is randomly partitioned.79% similar Unverified以英语约5万常用词汇计算，Trigram 理论上需存储约1.25×10¹⁴种组合的概率56% similar Unverifiedword2vec负采样中，对每个正样本随机采样k个（通常5-20个）负样本，采样概率通常按词频的3/4次方分布55% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/898145

API

curl https://kongchang.com/api/v1/knowledge/claims/898145

MCP

get_claim(id=898145)