Monday, September 14, 2009

Public Service Announcement

*cough, cough*
[Microphone feedback]
Is this thing on?
[tap, tap, tap]
(Oh, good.  Ahem.)

Ladies and Gentlemen, can we please stop pretending that the word "data" is plural?

Thank you.  Good night everyone, drive safely.


  1. This comment has been removed by the author.

  2. Oh, all right, I'll elaborate.

    Look. We don't use the word "data" like a plural anymore. At best it is plural in the sense that "sand" is plural: a large contiguous mass of material, any piece of which is almost infinitely divisible. In other words, it has become a mass noun in the common usage, and very sensibly so.

    Nobody uses "datum". Very few people who have data could actually tell you what one "datum" of it is. I collect reams of data, and I have no idea what a "datum" would be: a single sensor record? An element in that sensor record? If I break down one sensor record into individual parts, most of them make no sense on their own - but some of them do. If I break any of them down into bits, those bits lose their meaning instantly: they are no longer data, they are noise... unless it's a field of flags, in which case it's still data. It doesn't break down cleanly, you see? It's like "chickens": when you start dealing with giblets and thighs and hooves and whatever, you're not talking about "chickens" anymore, you're talking about "chicken".

    So please, let's be grown-ups about this.

    (This rant has been brought to you by: The Economist Magazine, an excellent publication with sometimes inscrutable editorial standards)

  3. All right, all right: I mean the English word "data". I will concede that, in Latin, "data" is plural.