Related info about quoting, in German and Dutch:
There is a growing controversy in Internet discussion forums about whether to put reactions before or after the quoted text. It is my opinion that a reply should always be below the quoted material, and I'll try to explain why.
Replying below is consistent with the flow of time, and with the sequence in a normal, spoken discussion: when people talk, you normally hear "question 1, answer 1, question 2, answer 2", and not usually "answer 1, answer 2, question 1, question 2".
Replying before the quote seems to have the advantage that the most recent text is seen first. You don't have to wade through old text before being able to read what the current participant has to say. However, this argument is only valid if too much text is quoted, which is a bad idea anyway. Replying above is the wrong solution of an unnecessary problem.
In a discussion thread, text is quoted so every message is understandable in its own right. Without quoting anything, the new text appears out of thin air. But quoting all the text of the previous message in a thread is too much: most people follow a thread either from beginning to end, or not at all. So all they need is a brief reminder of what the reaction refers too, not all the other remarks that are often not directly related, not all the details all previous authors put in their signatures, etc.
So if quotes are short, new text below can be seen at a glance just as easy as new text above.
For someone who hasn't followed a thread so far, and wants to join now, a brief reminder may be insufficient. But mail lists often have searchable archives on the web, and old messages in newsgroups are kept online for quite some time, and much longer by services such as what used to be DéjàNews, now Google, which eliminates the need for any complete quotations in the discussion thread itself.
A single short quote with a single reaction above it is just as clear as a short quote with text below it. However, when posting that single reaction, you never know if this simple exchange grows into a real thread in which several persons participate. That means that mixing styles is not a good idea, everybody has to follow the same style, within the same forum (Usenet newsgroup, or mail list), but preferably throughout the Internet, because many people participate in several, and it is easier for everybody to get used to one and the same style. It is my opinion that this is even true in a one-to-one email exchange, because adding even one CC-address turns it into a small forum.
Look what happens when reply styles are mixed:
>>> Reaction to reaction to original statement > Original contribution >> Reaction to original statement
This, often in combination with different ways of distinguishing quoted and new text (see below), makes it very hard to see who said what in reaction to what.
Note that this is only a problem when not only the text that your reply directly refers to is quoted, but also what was written before that. This shouldn't be done unless absolutely necessary, because in general, one should quote no more than is essential to follow the line of the discussion. Yet, in a long and complicated discussion with many participants, a three or four level quote (each quote being short of course) often adds to clarity.
Many discussion threads have, or gradually develop, several different aspects. Good writers separate them into paragraphs (or split the thread, but that often complicates matters, when done inconsistently, by several at the same time). Replying below the quote means you can also separate your reactions, and still make clear exactly what previous remark you are referring to. By quoting below, this becomes very difficult, especially when more than two people participate.
Here too, at the start you don't know whether this one question and one answer will develop into a thread with multiple, but still closely related sub-threads. Therefore a guideline that says "quote below when one-to-one, quote above when many one-to-one's" will lead to confusion through mixed styles within the same thread.
It is enough if someone suggests an answer to the original question, and comments on the suggestion made by somebody else, to split the thread into different aspects. The relation between them is often still so close that giving each a thread of its own is a step too far.
Software for email and newsgroups encourages a certain style of reacting, and as I see it, newer software tends to encourage the wrong style. So when I push a reply button, I don't want to see something like this:
CURSOR POSITION ---- Original message ----------- From: Someone <email@example.com> To: Somebody Else <firstname.lastname@example.org> Date: Thursday, January 1, 1999, 00:01 Subject: A boring little thread for the new year Non-wrapped text of message, paragraph continued in same line
Someone <email@example.com> wrote: > Line-wrapped text of message > paragraph continued in next line CURSOR POSITION
This second style makes it easy to weed out what is unnecessary, keep what is essential, and quickly add what I have to say. A quote of a quote is still easily identifiable. The other style encourages lazy quoting, and pages and pages of irrelevant text. A quote of a quote becomes a mess.
This second style was already used by UNIX Internet software many years before Internet dawned upon the masses, and has proven useful in practice. The newer method tends to come from one software maker who insisted that Internet was no good until very recently, and from another one that helped make it popular a few years before that.
The ASCII character '>' which is often used to mark quoted text, is probably an imitation of a vertical bar in the left margin. It doesn't look as good as a real bar, but it works on any system that supports simple unformatted text. It has the further advantage that second and third level, or even deeper quotations can easily be recognized simply because there's an extra '>' for every level. A program needs only add the '>'s when the reply function is executed (or reply button is hit, for those who like buttons), and a text that was already a quote is automatically turned into a deeper level quote.
Where only the most recent author is mentioned before the quoted text, which is easy with most software, it is still clear (for those who know the convention) when text is not a direct quotation of something written by that author, but text that was already a quotation in the previous message of the thread.
With a horizontal marker/separator, such as the dreaded
---- Original message -----------
this is much harder to see: you can see who wrote what, but not in what order, and what was written in response to what.
One problem with using '>' as a would-be vertical bar is that it requires text with hard line-breaks. In modern text-processing software this is avoided, hard returns should only be used at the end of a paragraph, and the line breaks are added dynamically by software, depending on page or screen width, font size etc. The advantage of course is that you don't need to reformat when a word is added or deleted.
This is true of text in an electronic discussion too, except that in quoted text you don't normally add or delete words. You only delete complete paragraphs to trim to what is essential. So even if the original message was written without hard line-breaks, and only paragraph breaks, it is all right if the software, in generating the quotation for you, fixes the position of the line-breaks (using some reasonable not too wide line length that all receiving software can probably cope with), and adds the '>' before each line.