Estimated User #164
3 IP addresses
▶
An IP address relates to the location of a user's computer or device, and a single person may have multiple IP addresses if they use different locations or networks. These IP addresses are grouped because they each submitted a fiction prompt longer than 30 characters that ended up in the same DBSCAN cluster—meaning the prompts are near-duplicates—from the same state (Pennsylvania). Groups are merged transitively—if A and B share a near-duplicate prompt cluster, and B and C share another, all three are grouped even if A and C never shared a prompt directly.
Each row below is a DBSCAN cluster of near-duplicate prompts (cosine similarity ≥ 0.9) that two or more of these IPs each contributed at least one conversation to. Only the conversations in these shared clusters are "evidence" of grouping; the rest of each IP's 92 total fiction conversations live in clusters where only that IP participated (or didn't form near-duplicate clusters at all). Per-IP fiction totals:
e5fc6db5a3…×39,
ae8bf0aea4…×33,
0dfdcced39…×20
14983
0dfdcced39…×18 of 20 total,
ae8bf0aea4…×9 of 33 total
15011
e5fc6db5a3…×10 of 39 total,
ae8bf0aea4…×4 of 33 total
Explore Story Permutations (11)
▼
Story permutations group this user's prompts into themes based on how similar their wording or meaning is. Click a group to see how the user's prompts evolved: within a selected group, each prompt is shown as a word-level diff against the previous one — green marks text the user added, and red strike-through marks text they removed. This reveals how the user revised, expanded, or pruned their prompts as they iterated. Jump to methodology →