Source-linked AI summary
BuyTheBy: A dataset of 18,710 text-based paper mill advertisements with 51,812 timestamped prices
Reese AK Richardson, Spencer S Hong, Anna Abalkina
TL;DR
Market-price evidence for paper-mill products has been sparse, anecdotal, or indirect. The authors assemble structured data from text-based advertisements and report a median first-author article price of 788 USD, while noting substantial variation across markets and contexts.
Problem
Evidence on paper-mill product prices has largely been sparse, anecdotal, or indirect, despite unknowns about paper-mill operations and product markets.
Source: Introduction · p. 1Method
The authors assembled BuyTheBy by collecting text-based paper-mill advertisements and structuring their metadata and price information.
Source: Introduction · p. 2, Data and methods · p. 2Results
788 USD is the median price for a first-author position on an academic article across the advertisements in the dataset, though prices vary considerably.
Source: Summary and analysis · p. 9, Discussion · p. 9Takeaways & Limitations
The dataset supports quantitative study of paper-mill operations and could inform research on pricing, product quality, academic discipline, and advertising strategies.
Source: Discussion · p. 9The dataset covers only text-based advertisements from a non-systematic selection of sources and has limited geographic and temporal coverage.
Source: Discussion · p. 9
Abstract
from arXiv · showhide
The study of paper mills and similar businesses operating in the market for academic and education fraud services is frustrated by the lack of market price data on their various offerings. Here, we assemble BuyTheBy, a large, annotated dataset of timestamped, text-based paper mill advertisements from seven businesses operating out of seven different countries. The dataset consists of 18,710 individual advertisements, of which 15,839 have prices listed. Among these there are 20,598 positions listed as for sale on 5,567 unique products in 14 different product categories with 51,812 timestamped price data points. We perform elementary analysis of this dataset to demonstrate its utility for quantitative understanding of markets for academic fraud services and suggest future use cases.
Introduction
Research on paper mills and related academic fraud services has advanced, but important aspects of their operations and markets remain unknown. BuyTheBy addresses the limited evidence on paper-mill product prices with structured data from advertisements.
Introduction
- Paper mills are part of a broader market for academic fraud services, while many aspects of paper-mill operations and product markets remain unknown.Paper mills are described as selling authorship on pre-written manuscripts and colluding with editors to bypass rigorous peer review.Source: Introduction · p. 1
- Available price evidence has been sparse, anecdotal, or indirect, with reported authorship-position prices ranging from 20 USD to 26,300 USD across investigations.Examples include a Chinese firm's 1,600–26,300 USD listings and a 2025 investigation reporting 20–1,500 USD.Source: Introduction · pp. 1–2
- BuyTheBy assembles structured price data from largely unstructured text-based advertisements for services related to publishing and credential fraud.Source: Introduction · p. 2
Data and methods
The authors collected text-based advertisements from Telegram channels and business websites, treating messages or listed topics as advertisements. They extracted and validated metadata and prices, redacted identifying information, and converted prices to USD.
Data and methods
- Three businesses' publicly accessible Telegram channels and four businesses' websites supplied the advertisements; attached PDF and image advertisements were not processed.Telegram channels were selected for largely English text and price data, to build a large dataset with minimal manual cleanup and redaction.Source: Data and methods · p. 2
- B1–B3 Telegram messages were treated as advertisements; B1 and B2 messages were in English, while B3 also included Uzbek, with prices in INR, USD, and UZS.The channels covered different collection periods, from B3's March 2020 start to B1's March 2022 start.Source: Data and methods · p. 2
- B4's searchable website listed 178 English article topics, each with up to eight priced authorship slots, and each topic counted as one advertisement.Source: Data and methods · p. 2
- B5's old-site 7,907 contracts were retained after excluding 17,763 zero-priced fake contracts; its newer site was archived across 13 dates.Each retained old-site page had one authorship slot; the newer site listed 85 topics in April 2024 and 491 in April 2026.Source: Data and methods · p. 3
- B6 advertised article, textbook, and abstract slots across archived language versions, while B7 advertised field-organized article slots; each listed topic counted as one advertisement.B6's abstract pages were understood to refer to abstracts in conference proceedings.Source: Data and methods · p. 3
- Metadata and prices were extracted and validated manually and automatically, identifying details were redacted, and prices were converted to USD using historical exchange rates.The processed dataset and data dictionary are available on Zenodo under CC BY 4.0, and code for figures and summary statistics is on GitHub.Source: Data and methods · pp. 3–4
Summary and analysis
BuyTheBy supports exploratory comparisons of advertised products, publication targets, and prices across businesses and over time. The analyses show substantial variation in prices between businesses and countries, alongside a first cross-dataset central estimate for an authorship position.
Summary and analysis
- B1 and B6 offered varied products, including textbook authorship, textbook editor positions, and patent inventorship, while the other five businesses advertised only academic-article authorship positions.Table 2 summarizes B1’s priced product types.Source: Summary and analysis · pp. 4–5, Table 2 · p. 5
- B1, B2, and B3 typically named target journals, whereas B4–B7 did not; B2 and B3 implied publication in indexed journals from major publishers, while B1 mainly advertised IEEE proceedings and small regional publishers, many likely hijacked.Tables 3 and 4 summarize venue publishers and references to indexing services or journal lists in advertisements.Source: Summary and analysis · pp. 5–6, Table 3 · p. 6, Table 4 · p. 6
- B1’s median advertised price for first authorship on an “international” textbook fell several times from about 15,000 INR to about 7,000 INR between June and December 2022, while IEEE proceeding authorship prices apparently rose twice in June/July 2024.The IEEE-advertising period captured in the dataset was a single month.Source: Summary and analysis · p. 7, Figure 1 · p. 7, Figure 2 · p. 8
- Prices for academic-article authorship positions varied considerably between businesses and countries, with differences likely related to the income levels of targeted clientele.Figure 3 shows distributions for authorship slots one through five; its y-axis is truncated at 2,500 USD, and B5’s maximum listed price was 5,631 USD.Source: Summary and analysis · p. 8, Figure 3 · p. 8
- 788 USD was the median listed price for a first-author position on an academic article across the seven businesses, using each position’s most recently advertised price.The estimate is reported across all advertisements in the dataset.Source: Table 5 · p. 9, Summary and analysis · p. 9
Discussion
BuyTheBy’s analysis finds substantial price variation over time and across markets, supporting its use for quantitative study of paper mill operations. The dataset also suggests further research uses while remaining limited in coverage and in linking advertisements to published products.
Discussion
- Prices can fluctuate considerably over time and vary across markets and contexts, even within one business’s portfolio.The authors use this finding to argue that rich datasets are important for quantitative understanding of paper mill operations.Source: Discussion · p. 9
- BuyTheBy could support analyses of prices by product quality, institutional reward, or academic discipline, as well as portfolio exposure and advertising strategies.The authors also identify qualitative study of paper mills’ advertising and communications as a potential use.Source: Discussion · p. 9
- The dataset excludes image- and video-based advertisements, uses a non-systematic selection of sources, and has limited geographic and temporal coverage.It also does not link advertisements to products eventually published.Source: Discussion · p. 9
- Advertisement-to-publication matching is complicated because paper mills may change article text and published products may appear in a different venue; advertised topics may model or plagiarize prior articles.Source: Discussion · pp. 9–10
Abbreviations
The listed abbreviations expand currency names and IEEE.
Abbreviations
- USD, EUR, INR, UAH, UZS, and RUB name the United States dollar, euro, Indian rupee, Ukrainian hryvnia, Uzbekistani sum, and Russian ruble; IEEE stands for Institute of Electrical and Electronics Engineers.Source: Abbreviations · p. 10