Dear James,
very well argued, thanks for the insightful post.
Saving drafts on the other hand could help avoid many conflicts on
less-trafficked pages.
Right now, on a page that is edited infrequently, this happens:
- User A starts an edit
- User A saves not to lose work, not quite done yet. Resumes the edit.
- User B (typically an editor) sees the edit by A, and sets to work
polishing it. Saves.
- User A saves -- conflict
The first edit by A woke up B, and led to the conflict.
If we allowed saving drafts, the following would be more likely:
- User A starts an edit
- User A saves a draft, and continues the edit.
- User A saves the edit.
- User B (typically an editor) sees the edit by A, and sets to work
polishing it. Saves.
The conflict would occur only if A had second-thoughts about the edit and
continues work after saving it, which might happen, but les frequently.
Of course saving drafts is also cumbersome to implement at scale (how long
would they persist? there would be clean up needed, etc; maybe they could
persist for one week then be mailed back to the author and deleted?).
Luca
On Fri, Sep 26, 2014 at 11:22 AM, James Forrester jforres...@wikimedia.org
wrote:
[Re-sending as it bounced first time.]
On 25 September 2014 22:45, Pine W wiki.p...@gmail.com wrote:
FWIW there were sessions at Wikimania about concurrent editing. I think
there is community support for the concept. If it helps us retain good
faith new editors then that is another good reason to press foward on this
subject. Perhaps James Forrester can provide an update on the outlook for
concurrent editing capability.
Hey.
[This is a bit off-topic for wiki-research-l, but I've been asked to
answer.]
First things first: There aren't any plans right now to try to roll this
out any time soon.
Collaborative real-time editing is an interesting task in terms of
engineering, but an exceptional challenge in terms of product. I think that
it's reasonable to talk about it as a possible solution to issues, but the
number of problems it raises is so great that people should be careful to
not talk of it as some magic pixie dust. :-)
For a couple of brief examples:
If the objective is to prevent all edit conflicts by making parallel edits
them impossible, this means either:
* everyone has to use the collaborative editor;
* people who can't use the collaborative editor (e.g. old computer, slow
network, no JavaScript, etc.) can't edit at all;
* people who don't like the collaborative editor are unable to edit ever
again; and
* bots can't edit at all (because they can't react to prompts from other
users)
… or:
* you have to choose to use the collaborative editor for each edit (how do
newbies know, or is it opt-out somehow?)
* as soon as someone wants to edit an article collaboratively, everyone
else's edits die and they're told so (or they all have to wait for the
collaborative edit session to end and then manually resolve the edit
conflict);
* for people who can't or don't want to use the collaborative editor, and
all bots, the article is essentially locked from their editing until the
collaborative edit is finished.
Neither of these are great options.
If instead we're happy to keep having edit conflicts, we can allow
parallel edits, but then the benefit for newbies (and, frankly, the rest of
us) goes away the second your collaborative edit conflicts with a
non-collaborative edit. Whoops.
Say that we've decided on a course of action for the above, maybe by
biting the bullet and denying people with older computers *etc.* the
ability to edit (which I think would be sad and a dereliction of
our values); what do you do when there are too many parallel editors of an
article?
When you're editing in a real-time collaborative editor, that means you
see the edits of each of the participants, alongside their
cursors/selections and comments in the chat system if there is one (which
there normally is). When there's two or three of these, it's relatively
easy to see what's happening. But what if there are 1,000 people trying to
edit the article at once (e.g. the article of a very famous individual just
after they've died unexpectedly; think Michael Jackson or Robin Williams).
Showing 1,000 cursors at once isn't just unhelpful – the level of traffic
would probably kill most users' browsers. Consequently, there needs to be a
limit somehow on the number of participants; maybe call it 10.
So, what happens when you click edit on an article where 10 people are
already editing?
* Do you just get told tough?
* Does the least-recently active editor get kicked out so you can join?
* Does this mean that all I need is 11 bots requesting to edit an article
to DoS it?
If you're a special user (e.g. a sysop), can you get into a
collaborative edit even if it's at the limit?
* If yes, doesn't this go against our values to place some editors above
others?
* If yes, do we just let
