In a comment on Geoff Nunberg's "The data are" post, Jo wryly reminds us that the data-is-plural-dammit peevers need to consider their position on the word agenda. The OED's (historically) first sense of agenda is
1. With pl. concord. Things to be done, viewed collectively; matters of practice, as distinguished from belief or theory. Sometimes opposed to credenda. Obs.
with citations like this:
1860 M. F. Maury Physical Geogr. Sea (ed. 8) i. §67 But notwithstanding all that has been done..for human progress, there still remain many agenda. There is both room and need for further research.
Plural agenda is of course etymologically correct:
< classical Latin agenda (neuter plural) business, affairs, in post-classical Latin also divine office (4th cent.), legal proceedings (12th cent. in British sources), plural of agendum thing which is to be done (usually in plural), neuter gerundive of agere to do
Over the past century or two, other uses of agenda have come to dominate, starting with
4. A list of items to be discussed at a formal meeting, typically circulated to attendees in advance. on the agenda: scheduled for discussion at a meeting. Originally as collective plural; now always treated as singular.
1832 Ld. Dover Let. in Proc. Rec. Comm. (1833) 36, I see among the Agenda of the meeting a question for removing certain Historical Documents..from the Chapter House to the State Paper Office.
1957 E. Hyams Into Dream ii. ii. 101 It's a short agenda, by the way, only two items.
and continuing to
5. In extended use (orig. U.S.).
a. A (notional) list of things to be done, problems to be addressed, or events likely to happen.
b. A campaign, programme, or plan of action arising from a set of underlying principles or motives. Hence: the underlying intentions or motives of a particular person or group.
If you feel that the word data must evermore remain plural, you should also feel compelled to protest the rampant abuse of agenda.
I should add a brief note on the state of play in the field of computer science, in the form of some counts of reasonable (though imperfect) proxy search patterns from the web site of the Association for Computing Machinery. There are 428 instances of "this data is" vs. 179 instances of "these data are". There are 40 instances of "some data is" vs. 6 instances of "some data are" (and no instances of "many data are").
Things like "one data is" and "a data was" do not occur, nor (apparently) do any higher numbers as modifiers (e.g. "three data were discarded" or "seventeen data were found in the couch cushions").
So the dominant pattern is to treat data as a singular mass noun, with a minority usage treating it as a grammatically plural mass noun.