How Should the SIGCHI Community Distribute the Cost of Publication?

tldr;

• Each published paper (e.g. Main Proceedings and Extended Abstracts for CHI2020) costs approximately $260 to produce. If we calculate this based on submitted papers, each submission costs approximately $70 to support.
• Edit: SIGCHI publications depend on the support and infrastructure of ACM. This cost is best represented in the current APC for Gold Open Access ($700), which should be included in the overall cost of publication.
• Currently, about half of this is covered by conference revenues (mainly conference registration), and the other half is covered by SIGCHI revenues (mainly publication revenue).
• Separating publication finances from conference finances mitigates risks and distributes costs more fairly and transparently. We could also reduce our dependence on publication revenue and explore open access options.

Recent conference cancellations and the resulting loss of revenues that would recover the costs invested in producing publications adds more urgency to the debate around the cost of publication. The fundamental problem is still the same: publishing has material costs, but how should we fund this?

This blog post analyses the publication finances of SIGCHI, the parent organisation that sponsors the CHI flagship conference and 24 specialised conferences. We use CHI2020 as a case study for the cost of producing large scale proceedings, but specialised conferences incur the same expenses and have widely varying approaches to producing conference proceedings.

Conference and Publication Finances

ACM, SIGCHI’s parent organisation, has traditionally funded publications by placing published content behind a paywall. SIGCHI has typically combined the costs of producing publications with the cost of attending conferences to avoid “publication fees.” This distributes the cost of publishing to conference attendees, and authors do not pay at the points where costs are incurred (e.g. during peer review¹, camera ready, producing the programme and proceedings, and archiving content). SIGCHI conferences also benefit from substantial investment in centralised publication infrastructure like PCS, which is funded primarily from our publication revenue².

Problems with this approach were less visible when conference attendance was less closely coupled to conference presentation, but changes in the ratio of attendees to presenters makes this more problematic. At SIGCHI conferences, a greater number of our attendees are also publishing content. A conference fee is not a publication fee, but as a greater proportion of attendees become authors the conference fee becomes a de facto publication fee. New publication models like Proceedings of the ACM (PACM), which publish journal issues independently of conference events, also highlight the need to separate these finances when publication is no longer coupled to conference attendance.

One of the benefits of separating content from presentation is the opportunity to untangle these finances. This would make it easier to be transparent with these costs, mitigate the financial risks of underwriting conferences, and distribute costs in a fairer way. This also creates an opportunity to explore new models that lessen our dependence on publication revenues as we move towards universal open access.

SIGCHI, Conference Cancellations, and Publication Finances

When publication costs are coupled to a conference, SIGCHI must be able to absorb these costs when conference revenues are lost due to conference cancellation. SIGCHI is able to absorb financial losses in case of emergencies such as a global pandemic because the organisation maintains a cash fund balance as a proportion of annual operating expenses³.

In these extraordinary circumstances, SIGCHI can absorb significant financial losses. Authors will not be required to pay the fixed costs of publication that would usually be included in conference registration fees, but this is not sustainable. We need a viable and transparent way to support the costs of publication, fairly distribute these costs, and mitigate the financial risks associated with running large events and exploring virtual events.

The Cost of Publishing CHI2020

Even though the in-person event for CHI2020 has been cancelled, the proceedings will be published. With a large technical programme, CHI2020 gives a good example of the costs related to publication when separated from the in-person event.

Technical Content Processed by CHI2020

• Papers: 3,126 submitted, 647 accepted
• Extended Abstracts: 1,724 submitted, 606 accepted
• Totals: 4,850 submitted, 1,253 accepted¹¹

CHI2020 Conference Budget

The conference budget includes direct costs that support the technical programme and proceedings.

• Programme Committee Meeting: $58k
• Proceedings Production: $90k
• Video Capture and Captioning for Archiving: $16k (based on CHI2019 spending)
• Total Direct Cost to CHI Budget: $164k

SIGCHI Publication and Technical Programme Support

SIGCHI provides centralised support, infrastructure, and covers associated costs to all 24 sponsored conferences for many core technical components. This includes PCS, QOALA (program scheduling software), progressive web app (formally native mobile apps), video tools for capture and upload (6 bespoke professional video capture stations) and archiving (to YouTube and the ACM DL). The CHI conference produces about 30% of SIGCHI publications portfolio annually⁴, we will estimate SIGCHI’s spending on infrastructure for operations and publications as part of indirect costs of publishing at CHI.

SIGCHI Executive Committee Costs

• Publications: CHI2020 Template Transition Support⁵, Direct Cost $16k
• Conferences: PCS for CHI2020, Direct Cost $33k
• Total Direct Costs, $49k
• Operations: Automatic Program Scheduler (QOALA), Progressive Web App (PWA), SIG Website, and other technology support. Total cost annually, $375k
• Total Indirect Costs for CHI2020 (30% * $375k): $112k

The Financial Reality

The total cost of producing proceedings of the technical programme at CHI2020 is $325k ($213k direct costs and $112k indirect costs). 50% of this is covered by the conference budget and the remainder is paid for by SIGCHI. Each published paper, including the Main Proceedings and Extended Abstracts, costs approximately $260⁶ to produce. If we consider this based on submissions, each submission costs approximately $70⁷ to support. These costs reflect the expenses of the PC meeting, running PCS, providing scheduling support and creating the technical programme, collating the proceedings, and archiving the work in the ACM Digital Library.

Edit: These costs represent the costs to SIGCHI, but we also depend on the support and infrastructure of ACM. The cost of this support is best represented in the current APC for Gold Open Access ($700), which should be included in the overall cost of publication.

Sustainable Publication Finances

Should we continue to combine conference and publication finances, or should we decouple these and distribute costs differently? Should authors pay to disseminate their work, and if so, when and how should fees be organised? Should we continue to offset these costs using publication revenue, or develop new financial models and move towards universal open access?

One option is distributing costs to the authors that contribute to and benefit from creating, reviewing, and disseminating their work. In all cases, SIGCHI would ensure that any fees applied can be waived so that authors unable to pay would not be denied access to publishing with SIGCHI⁸.

Upon introducing submission and/or publication fees, SIGCHI could also expect a 30% reduction in submissions. Lowered submission numbers would also create an opportunity to either raise the acceptance rate or cut down on simultaneous sessions at CHI, both of which are potential visions for future CHI events.

Submission Fee

We could adopt a model where all submitters pay a fee to submit. This has the advantage of recognising the value one gets from peer review and distributing costs at this point. Authors unable to pay can apply for a fee waiver. The model could also have variants, such as a single free submission before a fee for additional submissions.

Publication Fee

We could only require authors to pay a fee after acceptance. This has the benefit of removing barriers for authors to participate in submitting, but distributes costs to a smaller number of authors. This approach also fails to recognise the costs incurred during peer review. Authors do not pay to submit, but can apply for a fee waiver if accepted.

Combined Fee

In a combined fee model, all submitters pay a smaller fee at submission time and a larger fee after acceptance. This distributes the cost evenly across all authors, but those who publish their work pay a larger portion of the costs. Authors unable to pay can apply for a fee waiver.

Open Access

SIGCHI could publish all articles⁹ open access by including “Gold OA” article processing charges in addition to the fees described above. The current article processing charge for ACM Proceedings (SIG and ACM Members) is $700¹⁰. This amount pays for ACM’s publication support and infrastructure, without which SIGCHI publications would not be possible. Although revenue models are changing, for example ACM’s transformative agreements, these APCs are the primary route to open access at this time.

What Now?

• The in-person event for CHI2020 is cancelled and attendees have been offered full refunds.
• The CHI2020 proceedings will continue on schedule and be published before 25 April 2020.
• SIGCHI will absorb the financial losses of the in-person events cancelled to date, but we need to identify a plan moving forward in these uncertain times.
• The immediate plan may be different from what we want to do long-term, but having a vision of where we want to go may help us craft a short-term plan, so we should take this opportunity to reconsider how we financially support our publications.

After reading this and discussing it amongst your peers, please submit your thoughts to the SIGCHI EC.

Footnotes

1. There are material costs for supporting peer review software on top of the immense in-kind costs provided by the community as volunteers.
2. SIGCHI Revenue for FY2020 (excluding conference revenue, which does not generate income) is $1,170k, of which 74% is from publication revenue. This is primarily from library subscriptions to the ACM Digital Library.
3. For FY2020, SIGCHI has a projected closing balance of $2,886k.
4. In 2018, SIGCHI published 4,014 articles, of which 1,214 were from CHI’18. This does not include two articles from CHI’18 which were not published in the final proceedings.
5. The CHI2020 template is incompatible with the XML archive format required by ACM. As a short term solution, SIGCHI is funding manual translation of the CHI2020 format to the required XML format.
6. A total cost of $325k divided by the 1,253 published papers results in $259 per paper.
7. A total cost of $325k divided by 4,850 submissions results in $67 per paper.
8. How exactly this would function is out of the scope of this post, but SIGCHI can commit substantial resources to supporting authors unable to pay fees.
9. Non-archival content like Late Breaking Works, Panels, and Workshop abstracts fall through the cracks here: there is no APC process for making these articles open. As the largest SIG producing these publications, we should develop solutions to this open problem.
10. APCs at for-profit publishers vary widely, but represent a very different economic model. For example, Elsevier has an average APC of $2,600, with titles charging up to $5,900 per article.
11. Edit: Accepted article numbers vary slightly from final programme after shepherded papers, withdrawals, and other changes to numbers are finalised closer to publication. The final number of accepted papers is 1,365.