|
|
| Author |
Message |
| 8 new of 16 responses total. |
jadecat
|
|
response 9 of 16:
|
Jun 27 14:44 UTC 2007 |
One of the things I like about Livejournal, and dislike about Vox (yeah,
I have accounts all over) is that Livejournal allows for threaded
'conversations' and Vox doesn't. Would it be possible to make the Grex
Blogs work in a threaded fashion, so a person could reply specifically
to one comment rather than the way conferences currently work?
|
remmers
|
|
response 10 of 16:
|
Jun 27 16:03 UTC 2007 |
In a sense, threading is already possible in Backtalk - you can say, for
instance "re resp:6" and Backtalk will create a link back to response
#6. I guess that's not really full-blown threading in the sense it's
usually meant on mailing lists and usenet groups, though. Not sure I'd
want that - when discussions can branch into sub-discussions and
sub-sub-discussions, following the whole thing becomes exponentially
cumbersome. I like the strictly linear response structure that one has
in conferences here, and also in most blogging software.
Grex blogs provide an opportunity to make the rest of Grex more visible.
I'd like to see sidebar links to the Grex home page and Backtalk
entrance page on every blog page, for example.
Again on the subject of linking, can the Backtalk blogging system avoid
the auto-linking flaw that currently exists for conferences? What I'm
talking about is this. Consider the statement:
Be sure to check out item:agora,14 for the latest Happy Hour news.
Backtalk will link this to item 14 of the current Agora, which as I type
this is indeed about the Happy Hour. Three months from now, Agora will
have been restarted, item 14 is likely to be about something completely
different, and the above statement will be wrong, making me look like an
idiot. :)
I guess Backtalk is doing the best it can with that, given that there is
no built-in system of "unique identifiers" for items that is guaranteed
not to change over time. I'm just wondering if this can be avoided for
the blogs, or even some day remedied for the conferences too.
|
mary
|
|
response 11 of 16:
|
Jun 27 16:17 UTC 2007 |
Hey, and what's the problem with being an idiot? Some of my best
friends are idiots! ;-)
By the way, I'm really happy this blogging bit is going forward.
Thanks, Jan.
|
janc
|
|
response 12 of 16:
|
Jun 27 18:50 UTC 2007 |
Controls on linking:
I'm currently writing a new access control scheme for Backtalk. This
will give very fine grained control of access in a conference.
Depending on the conference configuration, these permissions will be
editable either only by cfadm or by fairwitnesses for the conference.
On Grex the fancy configuration settings will only be used for blogs,
where they will be settable by the blog owners. You'll be able to
control exactly who may post, who may edit, erase, etc. You'll also
be able to control who can link items out of the conference, what
conferences they can be linked to, and which items can be linked.
The default will probably simple be not to allow linking out of
blog items.
Of course, this wouldn't in general, prevent Picospan from linking
an item from a blog to a conference, since Picospan won't pay
any attention to my access control settings. But Picospan won't
even have the blogs in it's conflist, so it won't know those
conferences exist, and won't create any links from them.
The problem with the item:4 references in linked items, really goes back
to the fundamental half-wittedness of Picospan's data structures. I
could solve the problem by expanding out the links at post time, so that
an links like item:14 would be automatically convertedinto
item:coop,14. But even that wouldn't suffice, because if coop restarts we'd
want the" coop" will no longer point to the same place. Really we want ite
m:coop9,14 (or whatever). Here "coop9" is something we hope will be a
uniquename for the conference, good forever. But, in fact, given the Picospan
data structures, there is no way to find such a uniquename for a conference, or
to ensure that it remains unique. To really solve this problem requires
creating some new data structures, and leaving picospan compatibility in the
dust. Probably a good idea, on the whole.
Having trees of responses instead of a linear list of responses is sort
of the other major model of conferencing. Backtalk inherits some
low-level infrastructure for that from Yapp. In yapp, when you respond
to an item from the "respond or pass prompt" they you can say "resp 13"
to say that you are responding to response 13 of the item. Yapp still
displays the responses in chronological order, but responses entered
that way will say <responding to response 13> or something on that, so
the "parent" of the response is stored.
I could write an interface that would actually display these as a tree.
Then you'd be able to have a dual view of the same item. From one
interface, it would look like a tree of responses, from another
interface it would look like linear list of response, with a lot of them
having "Re: response 23" kinds of headers. I have never actually
implemented a hierarchical interface though, probably mostly because I
was never really very fond of them. It'd be nice to do in an AJAX
interface, where you can expand out arbitrary subtrees though.
|
unicorn
|
|
response 13 of 16:
|
Jun 27 22:42 UTC 2007 |
I, too, prefer a linear interface. I agree with remmers that with a
"threaded" interface, "following the whole thing becomes exponentially
cumbersome." A linear interface is more conversational. The discussion
flows the way a live conversation with all participants in the same room
at the same time would.
For that reason, I'd rather not see a "dual view" of the same item
implemented, because it would encourage some participants (those using
the threaded interface) to veer off onto multiple tangents, creating
the "sub-discussions and sub-sub-discussions" remmers mentioned, making
it difficult to follow no matter which interface you chose. A linear
interface tends to discourage that. Anyone who wants to veer off on a
tangent can always start a new item.
|
gelinas
|
|
response 14 of 16:
|
Jun 28 02:01 UTC 2007 |
Not in a blog: only the blogger can start new items. I think some will
find the 'threaded' view more useful. If the programmer wants to write it,
let's see how it works.
|
jadecat
|
|
response 15 of 16:
|
Jun 28 12:59 UTC 2007 |
I find the tree interface more useful- in blogs. You can keep a running
conversation between a couple of people going without necessarily
cluttering up everything.
The way Livejournal does it- you get a page of responses to a posting,
if the responses reach a certain number than the entire response to a
response will not be shown, just the header and who posted it. Plus the
non-blog owner can have an easier time of it. Vox doesn't do this and
often people will post a comment and then have to remember to come back
and check to see if their comment got responded to and there isn't a
quick visual way to do that.
From my personal organizational preferences it's a neater way of keeping
conversations together. Sure, a posting is kind of like a discussion at
a party, however, even at a party you're more likely to talk to one
group of people, then move to another group and so on.
|
unicorn
|
|
response 16 of 16:
|
Jun 30 00:21 UTC 2007 |
Re #14: Maybe the blogger should be able to choose whether the linear,
threaded, or both views are available to his/her readers.
Re #15: Yes, at a party you may move between groups and take part in
multiple discussions, but this is more like multiple items than it is
like multiple threads within an item. Multiple threads is something
that would be impossible to do live. It would mean taking part in
multiple discussions simultaneously, as well as starting new discussions
on the fly, and continuing the old ones as well. The human mind is
incapable of the level of multitasking that would require.
|