Re: Advertising xmpp

2008/1/5, Samuel Penn <samglendale.org.uk>:
> On Saturday 05 January 2008 12:48:14 Sander Devrieze
wrote:
> > 2008/1/5, Samuel Penn <samglendale.org.uk>:
> > > A small list would be better than a big list
> >
> > Why?
>
> Is there an advantage to a big list?
>
>   * It's harder to maintain.

Harder is not impossible

>   * If you list lots of small unreliable servers,
people who
>     choose them are likely to get a bad impression of
XMPP.

Unreliable servers should not be on the list; it's not
impossible to
add such measures, only hard 

>   * It may be confusing as to what the differences are
between
>     the servers, when really there isn't any.

How can this be confusing and how can this confusion be
fixed?

> Giving a user a big list of options, without any
sensible way of
> determining which they should choose, seems pointless.
They will
> most likely just choose the first in the list.

Is that a problem? Would it help to randomize the list?

> I can't think of an advantage of having a big list (it
doesn't
> mean there isn't one, just that I can't think of one
and haven't
> seen any mentioned here).

It's fair? It's compatible with the goal of openess (no hugh
entry
barriers to get on a list).

> Email clients don't present a user with a big list of
email servers.

You can't register an email account with an email client.

> On Saturday 05 January 2008 12:55:29 Tomasz Sterna
wrote:
> > And who will be the judge which servers are
"more equal" to be included
> > on the list, and which are "less equal"
and won't be included?
>
> Does it matter if you're not on the list?

Yes, it may be an entry barrier and entry barriers are not
compatible
with the mantra of openness of the XMPP community.

<snip>
> On Saturday 05 January 2008 12:48:14 Sander Devrieze
wrote:
> > My main email address does not indicate it is an
email address. People
> > just call it "email address". We
currently lack some similar naming
> > like "email address" and this is a *big*
issue (ping to XSF again).
>
> Yes, I agree. We're in the unfortunate situation that
the open protocol
> came along after the walled gardens had been set up.
People can just say
> email address, or web address, without caring about the
server
> implementation because SMTP and HTTP were open before
any of the big
> companies realised the importance of email or the web
and tried to
> create their own versions.

Not true, there were similar issues with email: there also
were walled
gardens for electronic mail.

<snip>
> Both Gaim and Kopete (I'm on Linux) present a number of
options when
> creating an account, like AIM, Jabber, MSN etc. Very
simply, they
> could present GTalk as a separate option. Technically,
it's the same
> as Jabber, but most people won't know this.
Alternatively, they
> could be presented as the same option (Jabber/Google
Mail). In either
> case, people who use Google mail might be more inclined
to use the
> talk facility.

Clients like Pidgin and Adium already do this but I don't
think this
is a good idea at all: in this way it creates even more
fraction.

> Many programs have a hints and tips dialog that pops
up, and the link
> between GTalk and XMPP/Jabber could be mentioned
there.

Could be a possibility indeed, but maybe there is a better
way?

-- 
Mvg, Sander Devrieze.
_______________________________________________
JAdmin mailing list
JAdminjabber.org
http:/
/mail.jabber.org/mailman/listinfo/jadmin
FAQ: http://ww
w.jabber.org/about/jadminfaq.shtml
_______________________________________________

On Saturday 05 January 2008 14:29:36 Sander Devrieze wrote:
> 2008/1/5, Samuel Penn <samglendale.org.uk>:
> > On Saturday 05 January 2008 12:48:14 Sander
Devrieze wrote:
> > > 2008/1/5, Samuel Penn <samglendale.org.uk>:
> > > > A small list would be better than a big
list
> > >
> > > Why?
> >
> > Is there an advantage to a big list?
> >
> >   * It's harder to maintain.
>
> Harder is not impossible

Agreed, but if resources are scarce it may be undesirable
unless
the advantages are large.

> >   * If you list lots of small unreliable servers,
people who
> >     choose them are likely to get a bad impression
of XMPP.
>
> Unreliable servers should not be on the list; it's not
impossible to
> add such measures, only hard 

I seem to recall seeing uptime stats for servers somewhere,
but can't
find it now.

It's not just one list that needs to be maintained though -
it's many.
When a version of a jabber client is packaged, the
developers need to
ensure that the list they use is uptodate. A mature client
may be
stable for a while, and get chosen for inclusion in an OS
distribution,
which possibly means that the list actually used by users
may be a
year or two out of date. That's assuming the developers
remembered to
grab the latest version of the list before building a
release of their
client.

The obvious solution is to have a dynamic list stored on a
server, which
a client can request, with a smaller backup list built into
the client
in case the server is not available. This requires getting
agreement
from client developers to make use of the list however.

> >   * It may be confusing as to what the differences
are between
> >     the servers, when really there isn't any.
>
> How can this be confusing and how can this confusion be
fixed?

If I'm presented with a list of a 1000 servers (say) to
choose from,
I'd want to have some idea as to how to make a choice. I'd
like
information on which are the most reliable and support the
most
services. However, if I know nothing about XMPP, then I may
not
even be aware that some servers might support more
facilities than
others. An uninformed decision isn't really providing any
benefit
to the user.

Admins and developers are actually in a better position to
decide
why one server is better than another than most users are.
Only
list those servers with a very high uptime, good performance
and
a wide range of features. Possibly limit it to servers local
to the
user (if that is possible to figure out).

If servers can dynamically report that they are lightly
loaded and
would be happy to have lots more users, list them as
recommended.
Spreading the load across lots of servers would be an
advantage,
but the user would have to be guided towards lightly loaded
servers
for this to work.

This is potentially a lot of work however.

> > I can't think of an advantage of having a big list
(it doesn't
> > mean there isn't one, just that I can't think of
one and haven't
> > seen any mentioned here).
>
> It's fair? It's compatible with the goal of openess (no
hugh entry
> barriers to get on a list).

I think that is only an advantage if server admins want more
users
on their server (for my server is bigger than yours bragging
rights).
If a server has been set up for a group of friends, a
project or an
organisation, they may not mind other people signing up, but
possibly
won't want *lots* of people signing up.

> > Email clients don't present a user with a big list
of email servers.
>
> You can't register an email account with an email
client.

True, I hadn't considered that.

> > On Saturday 05 January 2008 12:55:29 Tomasz Sterna
wrote:
> > > And who will be the judge which servers are
"more equal" to be included
> > > on the list, and which are "less
equal" and won't be included?
> >
> > Does it matter if you're not on the list?
>
> Yes, it may be an entry barrier and entry barriers are
not compatible
> with the mantra of openness of the XMPP community.

It's only an entry barrier if having lots of users on your
server is a
desirable goal. As long as people on my server can talk to
people on
your server, it makes no difference whether the split is
500:500 or
10:990 (ignoring problems caused by server load).

Because XMPP is open, the network effect applies equally to
all servers
regardless of how many users are on any particular server.
The only
figure that matters is the total number of users across all
servers.

This is different from IRC where though the protocol is
open, different
networks can't talk to each other. Here, signing up to
server A instead
of server B provides an advantage to server A.

<snip>
> Not true, there were similar issues with email: there
also were walled
> gardens for electronic mail.

Okay. I've never encountered any, though I didn't start
using the
internet and email until 1991. 

> > Both Gaim and Kopete (I'm on Linux) present a
number of options when
> > creating an account, like AIM, Jabber, MSN etc.
Very simply, they
> > could present GTalk as a separate option.
Technically, it's the same
> > as Jabber, but most people won't know this.
Alternatively, they
> > could be presented as the same option
(Jabber/Google Mail). In either
> > case, people who use Google mail might be more
inclined to use the
> > talk facility.
>
> Clients like Pidgin and Adium already do this but I
don't think this
> is a good idea at all: in this way it creates even more
fraction.
>
> > Many programs have a hints and tips dialog that
pops up, and the link
> > between GTalk and XMPP/Jabber could be mentioned
there.
>
> Could be a possibility indeed, but maybe there is a
better way?

I'll try and have a think.

-- 
Be seeing you,                         http://www.glendale.org.uk

Sam.                        Mail/IM (Jabber): samglendale.org.uk 
_______________________________________________
JAdmin mailing list
JAdminjabber.org
http:/
/mail.jabber.org/mailman/listinfo/jadmin
FAQ: http://ww
w.jabber.org/about/jadminfaq.shtml
_______________________________________________