TL;DR
This paper investigates the costs associated with academic publishing in Computer Science, examining both author and reader expenses in relation to venue influence (measured by h5-index). The study reveals a surprising lack of correlation between publication costs and venue impact, suggesting that high fees may not necessarily equate to higher quality or influence.
What Problem Does This Paper Address?
Academic publishing is a cornerstone of research dissemination, but concerns have long been raised about the escalating costs associated with it. Researchers and institutions face significant financial burdens for accessing published work (reader costs) and for publishing their own findings (author costs). This paper aims to quantify these costs within the field of Computer Science and explore whether there's a relationship between these expenses and the influence or prestige of the publication venue. The authors are motivated by historical critiques of publishers creating unnecessary cost burdens and the ongoing debate around open access models.
Methods
The researchers compiled a dataset of top Computer Science venues, ranked by their h5-index, which measures the impact of publications over the last five years. They collected data on author costs (minimum cost for a non-student to publish, including conference registration or publication fees) and reader costs (price to purchase a single article from a paywall). This data was gathered through surveys sent to editors/organizers and manual information gathering from venue websites. The analysis involved plotting costs against the h5-index to identify correlations and patterns, and also examining the distribution of free vs. paid access venues across different subfields of Computer Science.
Datasets
The primary dataset consists of top Computer Science venues, identified using Google Scholar's rankings for Top Venues in Computer Science. This list was further refined by considering the top 20 venues from each subfield of Computer Science, resulting in 288 unique venues after removing duplicates. Data on author costs, reader costs, and venue influence (h5-index) were collected for each venue.
Key Insights and Ideas
- Cost Doesn't Necessarily Equal Influence: One of the most striking findings is the absence of a significant correlation between a venue's h5-index (influence) and its reader or author costs. This challenges the assumption that higher fees translate to greater impact or prestige.
- Publisher Clustering: Major publishers like ACM and IEEE tend to charge consistent rates for articles, creating clusters in reader costs. However, author costs are more variable, often influenced by conference location.
- Outliers in the System: Several venues, such as JMLR, SWJ, and arXiv, offer free access to readers and authors, often due to different funding models or editorial philosophies (e.g., authors retaining copyright, volunteer efforts).
- Free Access and Influence: In many subfields of Computer Science, the proportion of free access venues is lower than their representation in terms of overall h5-index. This suggests that venues with free access might gain higher influence due to increased accessibility.
- Sponsorships Don't Guarantee Free Access: The analysis of venue sponsors revealed no clear relationship between sponsorship and whether a venue offers free access to its papers.
Key Results
- No significant correlation was found between a venue's h5-index and its reader or author costs.
- Major publishers like IEEE and ACM have relatively consistent pricing structures for reader access.
- In most subfields of Computer Science, the ratio of free access venues to paid access venues is lower than the ratio of their total h5-index.
- Fields like Databases, Theory, Graphics, Security, and Operating Systems show a notable difference in the proportion of free access conferences compared to paid ones.
- IEEE and Google were identified as major sponsors of the analyzed venues.
Limitations and Future Work
The authors acknowledge that their study is focused solely on Computer Science and that future work could expand to other academic fields. They also note that the data collection involved manual efforts and surveys, which can introduce limitations. The study serves as a starting point, and more comprehensive data collection across different years would be beneficial to observe trends in publishing costs. The authors hope their findings will contribute to discussions about the progression of costs in academic publishing and help ensure research remains accessible without monetary restrictions.
Summary updated