[biblio] Bug in acm.bst

[Peter Selinger]

Hi Karl,

I disagree with Oren's explanation for one important reason: the
Association for Computing Machinery (ACM), after whom the style file
is presumably named, does not actually use the comma in this way.

I sampled a number of their journals and magazines. See

http://www.mathstat.dal.ca/~selinger/downloads/ACM/

for some sample pages from recent journal issues. (I selected the
journals, and the articles within the journals, randomly, but I chose
the most recent issue I could get my hands on in each case). 

* Communications of the ACM: This style is closest to what the acm.bst
  produces. It shows authors as "last name, first initials", but there
  are *no* commas before "and" (neither for 2 authors, nor in the case
  of 3 or more).

* Journal of the ACM: shows names as "first names last name", but uses
  *no* commas before "and" in the case of 2 authors. It uses a comma
  when there are 3 or more authors.

* Transactions on Algorithms: shows names as "first initials last
  name", but uses *no* commas before "and" in the case of 2
  authors. It uses a comma when there are 3 or more authors.

I have not been able to find any ACM journal or magazine that uses a
comma before the "and" in the case of 2 authors.

Moreover, the ACM publishes it own BST file, available here:

http://www.acm.org/publications/latex_style/

(All three zip files contain the same BST file named
ACM-Reference-Format-Journals.bst). Their own style file produces the
"first names last name" author order, and it uses *no* comma before
"and" when there are 2 authors, but uses a comma before "and" when
there are 3 or more.

Finally, I recently submitted a paper to Communications of the ACM
using the acm.bst style, and the editor actually asked me to remove
the offending comma.

So even if Oren is right and this was intended as a feature when
acm.bst was last updated in 1988, I think there is plenty of evidence
that it is a bug now.

Thanks, -- Peter


Karl Berry wrote:
> 
> Hi Peter,
> 
> Oren looked into the situation and sent this reply (and gave the ok to
> post it here).
> 
> Best,
> Karl
> 
> --------------------
> From: Oren Patashnik
> 
> I saw in biblio at tug.org that Peter Selinger had asked to fix an
> alleged bug in the ACM bibliography style acm.bst.  Please don't do
> that, as the "bug" is actually a feature.  Here's an explanation:
> 
> It's completely clear from looking at both acm.bst and btxbst.doc (the
> template file for the standard styles and for acm.bst) that the
> behavior Peter describes is intended.  To review, Peter says:
> 
>   The style inserts a comma before "and" in a list of authors,
>   even if there are only two authors. Example: [...]
> 
>     Meter, R. V., and Horsman, C.
> 
> So it's the comma after the "V." that's in question.
> 
> The reason why acm.bst inserts this comma probably requires more
> explanation than is appropriate for this list -- short answer: it's
> due to the inverted-order name "Meter, R. V." -- but suffice it to say
> that *The Chicago Manual of Style* (page 531 of the 14th edition, the
> one I have on my shelf) also recommends the analogous comma, albeit in
> a slightly different style:
> 
>    TWO AUTHORS
>    15.85 [...] In the bibliography the name of the first author is
>    inverted, the second is given in its natural order, and the
>    conjunction joining the two names is preceded by a comma:
> 
>      Unwin, Liam P., and Joseph Galloway
> 
> The comma after the "P." here is analogous to the comma after the "V."
> that Peter objects to in his acm.bst example.
> 
> Bottom line: As I say, the acm.bst comma is a feature, not a bug.
> 
> -- Oren
>