The advantages of Usenet’s quoting conventions

This is the text of an article I posted to the newsgroup uk.religion.christian in early 2000, in response to someone who disliked the usual conventions of Usenet quotation. It’s rather rambling, but might be of some interest to someone ...

Philip Saunders dislikes the standard Usenet quoting conventions; he prefers to put what he has to say at the start of his articles and then quote the article he’s replying to, in toto and without further comments. He says:

> IMO it would be simpler and easier to follow a thread if you could read
> what is new first, rather than have to run through a snipped and
> therefore dubious history of the arguments every time.

That’s a plausible opinion, but I think it’s wrong. This article is an attempt to explain why.

It’s quite long, so here’s an executive summary.

  1. The standard convention fits better with the conversational tone of the medium.
  2. The standard convention results in articles that are more coherent and easier to read, because it keeps topics together.
  3. The standard convention does a better job of reminding readers of context when they need to be reminded of it.
  4. The standard convention is standard, and people are used to it; a non-standard convention has to be quite a lot better if its advantages are to outweigh the advantage of familiarity.
  5. That doesn’t mean that the standard convention is always best. but it’s usually best, and breaking with it needs a better reason than "I happen to prefer it some other way".

Numbers in the right margin indicate stuff relevant to these five points. [Actually, I forgot to put them in. -- gjm]


To begin at the beginning ...

Usenet is a funny medium. It’s rather conversational, but it’s not like an ordinary conversation because, whereas in an ordinary conversation each person usually speaks briefly and then yields the floor to someone else, in Usenet it’s usual for each article to have much more content in it. (Conversations in which each person gets in several paragraphs before the next speaks are pretty rare.)

But, although Usenet articles are much longer than typical conversational utterances, they’re usually not at all like essays or newspaper articles or monographs. As I said before, Usenet is a conversational medium; the "turnaround time" is much shorter than that in any printed medium, and the level of formality is much lower.


In a spoken conversation, everything happens rather quickly. A says something brief, B replies, C butts in, and so on, It’s not hard to keep track of what’s being said. Well, not very hard. Conversations have a way of losing their way, and when the subject is difficult or there are a lot of people involved it can be hard to follow what’s going on.

Usenet discussions are harder to follow than spoken conversations. They can involve a lot more people. They extend over longer time-scales, so that when C is replying to B, A’s earlier words probably aren’t clear in his memory. And there are lots of them going on at once. uk.r.c gets a couple of hundred articles every day. Oh, and because the articles are relatively long, the point you want to reply to may be submerged somewhere in the middle of the text.

There’s another problem, which is less serious in uk.r.c than in many newsgroups (for technical reasons that I shan’t go into): when you’re reading an article, you don’t necessarily have its predecessors available. They might have expired from your news server. They might not have arrived yet. Even if the predecessors are available, most newsreading software makes it fiddly -- not impossible, but fiddly -- to have your nascent reply, and the thing it’s a reply to, and a bunch of other articles, all visible at once. Therefore, each article needs to contain enough context that a reader can tell what’s going on in it without needing to trawl through lots of past articles.


OK, that’s enough background. We need a way of replying to articles that (1) provides enough context that readers don’t have to work too hard, and (2) fits well with the informal, sort-of-conversational tone of the medium.

There are two ways to provide context. You can quote, or you can summarise. (Of course, these aren’t mutually exclusive.) Summarising is hard to do well, and if it’s done badly then it can cause a lot of grief. Quoting is easy to do reasonably well. It’s also well supported by the tools we use: all news-reading software has some facilities for quoting the article you’re replying to. Therefore, quoting is the best general solution. (Those who are good with language and able to be impartial can summarise instead when it’s appropriate to do so.)

You can try to reply to individual points one by one, as is usually done, or you can try to reply to the whole of an article. Replying point by point is much more conversational; it makes the discussion much more like a face-to-face discussion between friends or colleagues. Replying to the article as a whole produces an atmosphere more like that of an academic journal ("In a previous article, Professor X argued that pseudomorphic maps are a good tool for the study of comparative epistemology. The purpose of this article is to suggest that actually hypermorphic maps are a better tool"). It may be worth mentioning that even in academic articles, block quotation from earlier articles is common.

The usual convention also results in individual topics being kept together. Suppose a group of articles deal with topics A, B, C and D; then the usual Usenet quoting techniques produce something like this

    > > A1
    > A2
    A3
    > > B1
    > B2
    B3
    > > C1
    > C2
    C3
    > > D1
    > D2
    D3

whereas if everyone treats the article they’re replying to as an individual whole we’ll get

    A3
    B3
    C3
    > A2
    > B2
    > C2
    > > A1
    > > B1
    > > C1

(or the reverse) which may look more elegant but splits each topic into fragments, making it harder to follow the thread of the discussion.

But, Philip claimed, it’s best to see what’s new first. Not so, in my opinion. As I’ve already said, context is of vital importance. Readers cannot be expected to remember everything that’s gone before, and they need to know what’s gone before if they’re to see the point of the new stuff. (This isn’t always true, but I think it’s true in most discussions in uk.r.c .) That suggests that the context should come before the new material.

The importance of context provides a reason for replying point by point, too. If I write an article (say) 40 lines long, and you reply to everything I’ve said in it but don’t match up your new material with the text in my article to which it’s replying, then the new reader has to work that much harder to see why what you’ve said is relevant. (Or, if they start from what I’ve written, to see what you’ve said in reply to each thing I’ve said.)


My final argument in favour of the standard conventions
is a rather obvious one. They’re standard. This isn’t
a defence of mere conventionalism; the point is that
people get used to things, and work better with what
they’re familiar with. There’s nothing fundamentally
better about reading from the top of a page to the
bottom instead of the other way around, but I bet you
find the following paragraph harder to read than this
one.

one.
found the previous paragraph easier to read than this
bottom instead of the other way around, but I bet you
better about reading from the top of a page to the
they’re familiar with. There’s nothing fundamentally
people get used to things, and work better with what
a defence of mere conventionalism; the point is that
is a rather obvious one. They’re standard. This isn’t
My final argument in favour of the standard conventions

In the same way, readers of Usenet newsgroups have generally become familiar with the conventions of the medium, and know how to work with them. Different conventions will make them work harder and feel less comfortable. If you’re going to do that, there had better be a good reason.


Other approaches have advantages. Giving no context at all saves disc space and network traffic; there are situations in which it’s appropriate, but Usenet isn’t one of them. Summarising and then replying to the whole article might be best when your issues aren’t with individual things someone has said but with some broader idea or ideas. Point-by-point replies can encourage argumentativeness, and they look less tidy on the page.

But, most of the time, these advantages aren’t enough to justify abandoning the replying conventions that have established themselves as the standard. Those standard conventions work well, and they should only be broken when you have a good, solid reason why they will not do.


It probably hasn’t escaped your notice that this very article defies the conventions it’s arguing for; it contains little context, summarises as well as quoting, and isn’t do anything like a point-by-point answer to anything. That’s because this is itself very different from most Usenet articles. It is deliberately more expository than conversational; it’s fairly formal in tone; it’s trying to provide fairly complete coverage of an issue that’s been only vaguely addressed before.

So I think the style of this article is appropriate. But the style has its dangers. I bet a lot of people will skip straight over it because it’s too long. (The separator lines and the executive summary are attempts to make its length less of a problem.) And, frankly, it’s a lot less fun to read than many articles in uk.r.c . In other words, even when a different posting style has a lot going for it, it’s marginal. Be warned.


This has been a public service announcement. Thank you for listening. Er, reading. If you ignored everything else here and just skipped to the last paragraph, too bad.