## The Open Access Landscape: 7. Computer Science

Computer Science includes software, data processing, AI, robotics and portions of what might be considered information science. This topic includes 338 journals, which published a total of 23,281 articles in 2013 and essentially the same number (not allowing for new journals), 23,153, in 2014.

## Grades

 Grade Journals %J Articles %A A/J A 188 56% 10,667 46% 57 Free 116 62% 4,061 38% 35 Pay 72 38% 6,606 62% 92 A\$ pay 11 3% 1,533 7% 139 B 60 18% 7,100 30% 118 Free 16 27% 1,057 15% 66 Pay 44 73% 6,043 85% 137 C 19 6% 3,394 15% 179 Free 1 5% 15 0% 15 Pay 6 32% 2,361 70% 394 Unk 12 63% 1,018 30% 85 D 60 18% 587 3% 10 Free 45 75% 361 61% 8 Pay 15 25% 226 39% 15

Table 7.1. Journals and articles by grade

Table 7.1 shows the number of journals and 2013 articles for each grade, the free, pay and unknown numbers, and average articles per journal. Boldface percentages (grades) are percentages of all computer science journals or articles, while others (free, pay, unknown) are percentages of the particular grade. So, for example, 18% of the journals are grade B, and 27% of that 18% are free.

Since A\$ means an APC of \$1,000 or more, all A\$ journals are Pay, so that line doesn’t appear. As is fairly typical, those journals average many more articles per journal than other A journals—but, unusually, they average fewer articles per journal than the highly questionable C journals and just more than the large number of APC-charging B journals. Across the board and as usual, however, journals with APCs publish more articles—on average—than journals without APCs.

The D journals, which as usual include relatively few articles, include these subgroups: C: five journals with 33 articles in 2013; D: ten journals, 37 articles; E: 12 journals, 41 articles; H: 11 journals, 393 articles; S: 22 journals, 83 articles.

## Article Volume (including all of 2014)

 2014 2013 2012 2011 Journals 284 316 304 262 %Free 53% 54% 55% 58% Articles 22,314 22,263 20,111 12,562 %Free 22% 25% 27% 31%

Table 7.2. Journals and articles by date

Table 7.2 shows the number of free and APC-charging journals that published articles in each year, including all of 2014, how many articles those journals published and what percentage were free.

The 12 “unknown” journals (with 1,018 articles in 2013, a fairly large number for journals that conceal their APCs) are omitted. The numbers still don’t add up to 338 because some journals don’t publish articles in any give year—ten of them in 2013, for example.

Although most computer science OA journals don’t charge APCs, the percentages here are lower than for OA as a whole or STEM, and show a slow decline (that is, increase in APC-charging journals) over recent years. The article percentages are distinctly low even for STEM, and the percentage of free articles has been steadily declining.

Without the “unknown” journals, total OA articles increased marginally in 2014, after a modest increase from 2012 and a huge increase from 2011. It’s quite possible that OA activity in computer science fields has plateaued, although new journals may change that picture.

Looked at on a journal-by-journal basis, 120 journals published more articles in 2014 than in 2013; 20 stayed the same; 198 published fewer articles in 2014. In terms of significant change, 111 journals (33%) published at least 10% more articles in 2014 than in 2013; 58 (17%) were relatively unchanged; and 169 (precisely half) published at least 10% fewer articles in 2014—including 42 that, so far, have not published any articles in 2014 (but a few of those are tricky cases, because one publisher’s archival controls seem to be malfunctioning).

 Journals No-Fee % Articles No-Fee % Prolific 4 0% 3,316 0% Large 26 8% 8,691 8% Medium 86 29% 7,193 30% Small 124 62% 3,284 64% Sparse 98 76% 797 69%

Table 7.3. Journals by peak article volume

Table 7.3 shows the number of journals in each size category, 2013 articles for journals in that group, and what percentage is free or in no-fee journals. The pattern here is not terribly unusual: the prolific journals all charge APCs, nearly all of the large ones also do, and two-thirds of the small and sparse ones don’t.

## Fees (APCs)

 APC Jour. %Fee %All Art. %Fee %All High 5 3% 2% 719 4% 3% Medium 21 14% 6% 1,273 8% 5% Low 43 29% 13% 5,964 36% 26% Nominal 79 53% 24% 8,813 53% 38% None 178 55% 5,494 24%

Table 7.4. Journals and articles by fee range

Table 7.4 shows the number of journals in each fee range and the number of 2013 articles for those journals. “Unknowns”—journals with APCs that aren’t stated—are left out of these calculations.

Since fee ranges for the OA universe were established based on actual quartiles (that is, 25% of fee-charging journals are in each range from High through Nominal), deviations from 25% represent differences between computer science OA journals and OA as a while. Here’ the differences are fairly clear: computer science journals are far less likely to charge high or even medium APCs than fee-charging OA journals as a whole—and the journals with relatively high fees don’t publish a large percentage of articles.

There is no statistical correlation (-0.06) between APC level and volume of articles; given the broad figures, a negative correlation might be expected.

## Starting Dates and the Gold Rush

 Year Total Free% 1980-89 2 50% 1990-91 0 1992-93 2 100% 1994-95 2 100% 1996-97 6 83% 1998-99 7 100% 2000-01 11 91% 2002-03 19 79% 2004-05 19 79% 2006-07 35 57% 2008-09 66 53% 2010-11 111 43% 2012-13 57 30%

Table 7.5. Starting dates for computer science OA journals

Table 7.5 shows computer science journals by starting date, including the percentage of journals started in a given date range that currently do not charge APCs.

For DOAJ as a whole, I get a sense of a gold rush of new APC-charging journals from 2006 through 2011, diminishing somewhat since then. Not surprisingly, there are no very early computer science OA journals: before 1980, there just wasn’t much of a field there. The gold rush seems clear enough: from 1992 through 2005, at least three-quarters of OA journals do not charge APCs—but that percentage drops sharply in later years as the number of new journals rises sharply. Figure 7.1 shows essentially the same information as Table 7.5, but as a graph with lines for free and APC-charging journals, including markers so that certain dates show up. I think the graph is fairly clear: almost no APC-charging journals as free journals started rising—then a huge surge in APC-charging journals through 2010-2011.

Figure 7.1. Computer science OA journals by starting date

 Year Journals Articles Art/Jrnl 1980-89 2 23 12 1990-91 0 0 1992-93 2 65 33 1994-95 2 166 83 1996-97 6 303 51 1998-99 6 334 56 2000-01 11 375 34 2002-03 18 1,279 71 2004-05 18 2,165 120 2006-07 33 2,125 64 2008-09 64 4,184 65 2010-11 109 8,434 77 2012-13 57 3,828 67

Table 7.6. Articles per journal by starting date

Table 7.6 shows journals that published articles in 2013, when they started, and average articles per journal. Two time periods stand out: journals that began in 2004-2005 have considerably more articles per journal than others, with 1994-95 not too far behind.

## Comments

Computer science is generally a newer field than most other broad topical divisions. While the emergence of hundreds of OA journals, most of them charging APCs, suggests a gold rush, most of those journals charge relatively modest fees—and the ones with four-digit APCs don’t publish a high percentage of articles.

### Definitions and notes

See The Open Access Landscape: 1. Background for definitions and notes

If you’re interested in a book-form version of this material (with an additional bonus graph added in each chapter), let me know, either in a comment or by email to waltcrawford at gmail dot com

Comments are closed.