| |
It is clear from the debate raging further down the group (as NNTP
newsreaders see it :) ) that one of the fundamental flaws with ratings at
the moment is that there is no criteria for ratings posts. This in turn
means that the ratings themselves mean nothing at present. Therefore I hope
to start a debate that will hammer out the criteria on which messages are
rated and hopefully reduce acrimony among some members of Lugnet.
I have proposed that posts should be separately rated on TWO criteria,
although admittedly they are at present too general to be of any use to
anybody.
1 - The relevancy of the post to the message commencing the thread. This
would enable background noise to be filtered out in lengthy debates and also
to indicate when a thread is starting to move off-topic and should be
re-housed to another area of Lugnet. However, one cannot rate the relevancy
of the post commencing a thread as this would mean one would be rating the
relevancy of a message to itself :) I have named this the 'Relevancy
Rating' (I'm not good at originality)
2 - The quality of the concepts/ideas expressed within the posts. This is
more far-reaching than point one as it enables one to draw attention to the
more interesting posts. For example, down in the middle of a thread about
train chassis design someone may mention a brilliant idea for train
superstructure design but because of the relevancy rating it disappears into
the murky depths of Lugnet. With the 'Quality Rating' (as I have christened
it) it would come to the surface to be seen by all as an interesting post in
its own write, independent of its relevancy.
Thus, when one looks at the message in the web interface it has TWO ratings
at the top. One, its 'Relevancy' and two, its 'Quality'.
I hope that these ideas get the ball rolling in creating criteria that
results in a meaningful and useful ratings system.
Nicholas Allan
|
|
| |
In lugnet.admin.general, Nicholas Allan writes:
<consider using Relevancy and Quality as dimensions>
Agreed that these are two. I nominated
Newsworthiness
and
Long Term Usefulness
as well.
A post that told us all of a huge and very cheap sale that expires in 5 days
would have a high newsworthiness but a low long term usefulness. So it would
sort to the top of top-post lists, but only for a short while.
Gary Istok posting another magnum opus on the theory of window design isn't
very newsworthy. ( :-) We all know he knows his stuff backwards and forwards,
so it's not news that he comes out with another good one). However it has long
term relevance... what happened in 1968 in Hungarian dealer catalogs is likely
to matter just as much next year as it does tomorrow. So IT would sort to the
top of top-post lists for a long long time.
Here's a new idea:
I would allow the ORIGINATOR of a post that was rated by others as highly long
term relevant to OVERRIDE that rating in the downward direction (not upward,
but downward). Why? Because the originator may well have produced another more
accurate post that supplants it and may want to help newcomers find that one
instead.
In fact, come to think of it, I would allow the ORIGINATOR of posts to
downgrade (but not upgrade) a number of these dimensions. Consider it almost
like cancelling. It's not editing because content isn't being changed, and it's
better than cancelling because the post doesn't go away completely, it just
doesn't sort as high any more.
This notion of downgrading ratings by the originator drives home, conceptually,
that a multidimensional rating scheme is NOT a popularity contest. It's a
usefulness gauge.
Note that I suspect that implementing multidimensional ratings, while an
awesome idea, will require a lot of LDT, EIther the current numbers need to be
ripped out and replaced or something very crufty needs to be done.
++Lar
|
|
| |
Larry Pieniazek <lar@voyager.net> wrote:
> Note that I suspect that implementing multidimensional ratings, while an
> awesome idea, will require a lot of LDT, EIther the current numbers need to be
> ripped out and replaced or something very crufty needs to be done.
Good suggestions. I'd like to flog my personal dead horse here :) and note
that it'd be nice to just have ONE button/checkbox for each scale. You'd
still end up with continuum of scores, but based on the total number of
people who pressed the button.
I don't think that much is lost with this approach, and it radically
simplifies the interface -- especially if there are going to be multiple
scales.
--
Matthew Miller ---> mattdm@mattdm.org
Quotes 'R' Us ---> http://quotes-r-us.org/
Boston University Linux ---> http://linux.bu.edu/
|
|
| |
In lugnet.admin.general, Matthew Miller writes:
> Good suggestions. I'd like to flog my personal dead horse here :)
Hey.. there's a cue for dead-horse floggage in this group ;-)
> and note
> that it'd be nice to just have ONE button/checkbox for each scale. You'd
> still end up with continuum of scores, but based on the total number of
> people who pressed the button.
It does have some benefits - how would the "off-topic" button work though if
the message was cross-posted, and assuming that it wasn't off-topic in every
group? Also - that is a problem with any cross-posted message - it might be
great in one group, but not in another - there are quite a few examples of this
already!
Also - if you like the message, do you rate it? Or do you only rate messages
that you REALLY REALLY like? This one might actually work itself out - not
everybody will rate the latest pirate jungle, but a lot more people will rate
the latest news from LD, thereby giving it greater prominance. But it is
perhaps best not to assume that it would work that way, eithout further
consideration?
Richard
|
|
| |
Richard Franks <spontificus@yahoo.com> wrote:
> It does have some benefits - how would the "off-topic" button work though if
> the message was cross-posted, and assuming that it wasn't off-topic in every
> group? Also - that is a problem with any cross-posted message - it might be
> great in one group, but not in another - there are quite a few examples of this
> already!
Yes, that's a point I hadn't thought of. Hmm. Since having the off-topic
score is really important to this whole concept, it might have to be kept
seperately for each group the message was posted in. Yuck. (But that's also
a problem with the current system.)
> Also - if you like the message, do you rate it? Or do you only rate messages
> that you REALLY REALLY like? This one might actually work itself out - not
> everybody will rate the latest pirate jungle, but a lot more people will rate
> the latest news from LD, thereby giving it greater prominance. But it is
> perhaps best not to assume that it would work that way, eithout further
> consideration?
Oh, I think that's exactly how it'll work out. Additional points:
- having to put less thought into weighing exactly how much each message
is "worth" makes the process less tedious, and therefore something more
people are more likely to do.
- with the current system a there's not much of a way to tell the
difference between a message two people liked a lot and one 100 people
liked a little bit.
- I actually think that a large number of people wouldn't make use of the
scale anyway. Since scoring articles means going out of my way, I only
bother to do it if I think it's something people _really_ should see and
I rate it 100.
--
Matthew Miller ---> mattdm@mattdm.org
Quotes 'R' Us ---> http://quotes-r-us.org/
Boston University Linux ---> http://linux.bu.edu/
|
|
| |
In lugnet.admin.general, Matthew Miller writes:
> Richard Franks <spontificus@yahoo.com> wrote:
> > It does have some benefits - how would the "off-topic" button work though if
> > the message was cross-posted, and assuming that it wasn't off-topic in every
> > group? Also - that is a problem with any cross-posted message - it might be
> > great in one group, but not in another - there are quite a few examples of • this
> > already!
>
> Yes, that's a point I hadn't thought of. Hmm. Since having the off-topic
> score is really important to this whole concept, it might have to be kept
> seperately for each group the message was posted in. Yuck. (But that's also
> a problem with the current system.)
Well, I'm no computer whiz but it doesn't seem like it'd too hard. After all,
on the page after you post there are links to every group you posted in, and
those are created on the fly, isn't that so? ...so how hard is it to create
the o-t part of the rating on the fly, with a checkbox next to each group?
(I'm not trying to be annoying, I'm really asking how hard is it.)
> > Also - if you like the message, do you rate it? Or do you only rate messages
> > that you REALLY REALLY like? This one might actually work itself out - not
> > everybody will rate the latest pirate jungle, but a lot more people will rate
> > the latest news from LD, thereby giving it greater prominance. But it is
> > perhaps best not to assume that it would work that way, eithout further
> > consideration?
I agree with Matthew - that's definitely how it'd work out. I don't see any
other aspects that could come up.
> - having to put less thought into weighing exactly how much each message
> is "worth" makes the process less tedious, and therefore something more
> people are more likely to do.
Right!
> - with the current system a there's not much of a way to tell the
> difference between a message two people liked a lot and one 100 people
> liked a little bit.
> - I actually think that a large number of people wouldn't make use of the
> scale anyway. Since scoring articles means going out of my way, I only
> bother to do it if I think it's something people _really_ should see and
> I rate it 100.
I usually only rate 100s as well. And it seems like a lot of people do that.
I've seen so many posts that are slowly climbing to the maximum possible for
that number of ratings (meaning they only get 100 scores). So why not just
make it an "all or nothing" rating? Either it's newsworthy or it's not, either
off-topic or not. No need to say by how much it's off topic, or how newsworthy
it is; that can be decided by how many people rate it.
|
|
| |
In lugnet.admin.general, Shiri Dori writes:
> > Yes, that's a point I hadn't thought of. Hmm. Since having the off-topic
> > score is really important to this whole concept, it might have to be kept
> > seperately for each group the message was posted in. Yuck. (But that's also
> > a problem with the current system.)
>
> Well, I'm no computer whiz but it doesn't seem like it'd too hard. After all,
> on the page after you post there are links to every group you posted in, and
> those are created on the fly, isn't that so? ...so how hard is it to create
> the o-t part of the rating on the fly, with a checkbox next to each group?
> (I'm not trying to be annoying, I'm really asking how hard is it.)
the problem I think isn't in generating the page, but in storing the
information. Right now, each message has only one set of ratings to keep track
of... kinda like one rating file per message. But if you have to keep track of
the rating in each group, you end up with one file per group per message. Not
impossible to do, just not "clean". :/ Dunno what other way there is to do
though.
<snip>
> I usually only rate 100s as well. And it seems like a lot of people do that.
> I've seen so many posts that are slowly climbing to the maximum possible for
> that number of ratings (meaning they only get 100 scores). So why not just
> make it an "all or nothing" rating? Either it's newsworthy or it's not, either
> off-topic or not. No need to say by how much it's off topic, or how newsworthy
> it is; that can be decided by how many people rate it.
Yes, I think since we're hopeing for a large number or raters, this would be a
good way to go.
:)
Dan
|
|
| |
In lugnet.admin.general, Shiri Dori writes:
> > > Also - if you like the message, do you rate it? Or do you only rate messages
> > > that you REALLY REALLY like? This one might actually work itself out - not
> > > everybody will rate the latest pirate jungle, but a lot more people will
> > > rate the latest news from LD, thereby giving it greater prominance. But it
> > > is perhaps best not to assume that it would work that way, eithout further
> > > consideration?
>
> I agree with Matthew - that's definitely how it'd work out. I don't see any
> other aspects that could come up.
I'm just advocating objectivity and consideration, rather than optimism.
I trust that Todd put a lot of individual thought into the original system, but
that didn't stop confusion and peoples feelings getting hurt. The replacement
should be thought out *very* carefully.
Within minutes of the original system coming online, several people had
pinpointed potential problems - many of which came to life. Maybe LUGNET should
go back to its roots and have an 'experimental' test server or area which could
test out new features, and give an idea of the technical and social problems
that could be caused. No matter how smart Todd is, a group of indivuals using
LUGNET via a test-interface/server/whatever will always be able to come up with
more insights than one person.
This would also stop confusion and annoyance to casual users who (suprise!)
just want to talk about LEGO, and don't care that much about the latest fun
features.
I'm all for playing with new geeky toys, but from some private email recently..
I'm starting to understand that it **REALLY** bugs some of the casual users, as
features come and go. Testing features on the main server also means that they
can't be tweaked as often as would be productive - on a test interface they
could be overhauled as necessary - once you implement something on your main
server it's not as easy to just pull it down and start again.
Richard
|
|
| |
In lugnet.admin.general, Richard Franks writes:
<snip>
> I trust that Todd put a lot of individual thought into the original system, but
> that didn't stop confusion and peoples feelings getting hurt. The replacement
> should be thought out *very* carefully.
>
> Within minutes of the original system coming online, several people had
> pinpointed potential problems - many of which came to life. Maybe LUGNET should
> go back to its roots and have an 'experimental' test server or area which could
> test out new features, and give an idea of the technical and social problems
> that could be caused. No matter how smart Todd is, a group of indivuals using
> LUGNET via a test-interface/server/whatever will always be able to come up with
> more insights than one person.
>
> This would also stop confusion and annoyance to casual users who (suprise!)
> just want to talk about LEGO, and don't care that much about the latest fun
> features.
what if there was a group .admin.server to discuss and test new features? That
way, we'll move this traffic from .admin.general, (since I agree that it
doesn't really belong here), and have a playground to test new things, without
affecting the casual user... In this case, the rating system would have been
tested only in that group, before being released to all of lugnet...
:)
Dan
|
|
| |
In lugnet.admin.general, Matthew Miller writes:
> [...]
> - with the current system a there's not much of a way to tell the
> difference between a message two people liked a lot and one 100 people
> liked a little bit.
> [...]
There are two ways, currently -- but they're designed to work without getting
in the way (rather than to be obvious).
First way: If two people like a message a lot (and, say, each mark it 100),
then it has a lower composite score than, say, several dozen people all
marking it 100. This is because of the automatic softener-vote[1] by the
server. As an example, let's say the automatic vote is 0/100 rather than
50/100. Two votes of 100 gives (0+100+100)/3 = 67. Twenty votes of 100
gives (0+100*20)/21 = 95. On a 1-to-5 or 1-to-4 "star" scale, 67 maps to
fewer "stars" than 95.
Second way: The coloring of the rating symbols is two-dimensional. Its input
is (a) the composite rating (the average, including the softener) and (b) the
total number of human votes. The composite rating translates into a color-
brightness from black to red, and the number of votes translates into
a saturation level from gray to the color-brightness. Thus, two votes of 100
looks reddish-grayish and twenty votes of 100 looks very bright red. In other
words, the color saturation is how seriously to take the rating. It's subtle,
but it's like that on purpose so that, basically, the more red you see, the
more your eye is drawn to it. (Red, biologically, has the strongest draw of
all colors, for non-colorblind humans.)
--Todd
[1] http://www.lugnet.com/admin/general/?n=4854
http://www.lugnet.com/admin/general/?n=5487
|
|
| |
Todd Lehman <lehman@javanet.com> wrote:
> First way: If two people like a message a lot (and, say, each mark it 100),
> then it has a lower composite score than, say, several dozen people all
> marking it 100. This is because of the automatic softener-vote[1] by the
Yes, I understand that, but what about two people liking a message a lot vs.
several dozen saying that it's mediocre? Those are quite different.
> Second way: The coloring of the rating symbols is two-dimensional. Its input
Ah. Yes, that's much better. (Does MS Win still only use 20 colors for text,
or can you use anything now?)
> words, the color saturation is how seriously to take the rating. It's subtle,
> but it's like that on purpose so that, basically, the more red you see, the
> more your eye is drawn to it. (Red, biologically, has the strongest draw of
> all colors, for non-colorblind humans.)
Red also has connotations of "Warning", "Stop", and "Negative", doesn't it?
--
Matthew Miller ---> mattdm@mattdm.org
Quotes 'R' Us ---> http://quotes-r-us.org/
Boston University Linux ---> http://linux.bu.edu/
|
|
| |
In lugnet.admin.general, Matthew Miller writes:
> Todd Lehman <lehman@javanet.com> wrote:
> > First way: If two people like a message a lot (and, say, each mark it 100),
> > then it has a lower composite score than, say, several dozen people all
> > marking it 100. This is because of the automatic softener-vote[1] by the
>
> Yes, I understand that, but what about two people liking a message a lot vs.
> several dozen saying that it's mediocre? Those are quite different.
Two people liking a message a lot would have a composite score of 60 to 70.
Several dozen saying that it's mediocre would have a composite score of 0 to
30, depending on how you define "mediocre." A low score with a lot of input
shows up dark (far away from red and far away from pale gray) if it even shows
at all.
We need to get rid of the "having been marked down" stigma...even if it loses
functionality for those who have tough enough constitutions to weather any
perceived feedback. It seems from the input here today and yesterday that
it's far better (overall -- from a community-wide standpoint) not to be able
to mark something as mediocre or worthless, than to be able to do so at the
risk of hurting someone's feelings.
> > Second way: The coloring of the rating symbols is two-dimensional. Its
>
> Ah. Yes, that's much better.
It was like this before, but with the change of the default from 50 to 0, it
was possible to make the "starting" color pure white rather than mid-gray --
that helps make the gradations much more visible. I'm happy to see some of
the marks now showing up only _very_ faintly.
> (Does MS Win still only use 20 colors for text, or can you use anything
> now?)
Dunno. Only about 5 or 6 are really needed anyway.
> > words, the color saturation is how seriously to take the rating. It's
> > subtle, but it's like that on purpose so that, basically, the more red you
> > see, the more your eye is drawn to it. (Red, biologically, has the
> > strongest draw of all colors, for non-colorblind humans.)
>
> Red also has connotations of "Warning", "Stop", and "Negative", doesn't it?
Depending on the culture and, more importantly, the context, yes. But I
believe that in all cultures which use printed or visual materials, red is
an attention-getter (because it's hard-wired into our biology) -- and that's
the purpose here -- to bring attention.
--Todd
|
|
| |
In lugnet.admin.general, Todd Lehman writes:
> Depending on the culture and, more importantly, the context, yes. But I
> believe that in all cultures which use printed or visual materials, red is
> an attention-getter (because it's hard-wired into our biology) -- and that's
> the purpose here -- to bring attention.
Oh -- one other thing -- Orange also has this property. Red and Yellow are
the two most attention-getting colors to humans.
--Todd
|
|
| |
Todd Lehman <lehman@javanet.com> wrote:
> perceived feedback. It seems from the input here today and yesterday that
> it's far better (overall -- from a community-wide standpoint) not to be able
> to mark something as mediocre or worthless, than to be able to do so at the
> risk of hurting someone's feelings.
I agree. And (guess what I'm going to say next *grin*) I think the "one
button" idea lends itself to this.
> > (Does MS Win still only use 20 colors for text, or can you use anything
> > now?)
> Dunno. Only about 5 or 6 are really needed anyway.
The problem was that only 2 of those 20 were red....
--
Matthew Miller ---> mattdm@mattdm.org
Quotes 'R' Us ---> http://quotes-r-us.org/
Boston University Linux ---> http://linux.bu.edu/
|
|
| |
In lugnet.admin.general, Matthew Miller writes:
> > > (Does MS Win still only use 20 colors for text, or can you use anything
> > > now?)
> >
> > Dunno. Only about 5 or 6 are really needed anyway.
>
> The problem was that only 2 of those 20 were red....
Oh, I see what you're saying. You mean 20 colors total, not 20 shades of a
single color.
No, you have go all the way back to *really* early browsers and 4-bit color
before you get super-restricted color choices. (Anyone running something
that prehistoric is already hating life.)
MicroSoft Windows does properly display at least 216 different text colors
(giving 6 shades of red, if you count black as a shade of red :) starting at
Netscape 3.0 and Win95. Not sure about MSIE but I think it had improved color
choices before NN did.
--Todd
|
|
| |
Todd Lehman <lehman@javanet.com> wrote:
> MicroSoft Windows does properly display at least 216 different text colors
> (giving 6 shades of red, if you count black as a shade of red :) starting at
> Netscape 3.0 and Win95. Not sure about MSIE but I think it had improved color
> choices before NN did.
Ok. My memory of that must be from back when I was designing web pages that
had to work with NS 2.0. Wow, I must be getting old. :)
--
Matthew Miller ---> mattdm@mattdm.org
Quotes 'R' Us ---> http://quotes-r-us.org/
Boston University Linux ---> http://linux.bu.edu/
|
|
| |
In lugnet.admin.general, Matthew Miller writes:
> Todd Lehman <lehman@javanet.com> wrote:
> > perceived feedback. It seems from the input here today and yesterday that
> > it's far better (overall -- from a community-wide standpoint) not to be able
> > to mark something as mediocre or worthless, than to be able to do so at the
> > risk of hurting someone's feelings.
>
> I agree. And (guess what I'm going to say next *grin*) I think the "one
> button" idea lends itself to this.
OK, there are now two choices (down from 5, down from 11 originally) in a
drop-down list...
CHOICE ACTION
- - - No action (default choice).
° Highlight Recommend for highlighting with a ° symbol.
°° Spotlight Recommend for highlighting with a °° symbol and also
for inclusion in the LUGNET Spotlight page.
The first choice sends a "-" (meaning "undefined") to /news/rate.cgi as input.
The second choice sends a 75; the third choice sends a 100.
--Todd
|
|
| |
Todd Lehman <lehman@javanet.com> wrote:
> OK, there are now two choices (down from 5, down from 11 originally) in a
> drop-down list...
:) yay! thanks.
> The first choice sends a "-" (meaning "undefined") to /news/rate.cgi as
> input. The second choice sends a 75; the third choice sends a 100.
Has the interface for rate.cgi changed at all? (I assume not.) I'll probably
hack slrn this weekend to have rate keys...
PS -- choice of characters rates a °°. :)
--
Matthew Miller ---> mattdm@mattdm.org
Quotes 'R' Us ---> http://quotes-r-us.org/
Boston University Linux ---> http://linux.bu.edu/
|
|
| |
In lugnet.admin.general, Dan Boger writes:
> what if there was a group .admin.server to discuss and test new features?
> That way, we'll move this traffic from .admin.general, (since I agree that it
> doesn't really belong here), and have a playground to test new things, without
> affecting the casual user...
I think something like admin.server would be useful, at the moment normal
admin.general traffic is getting drowned. Maybe admin.playground would convey
the testing-ground message?
> In this case, the rating system would have been
> tested only in that group, before being released to all of lugnet...
At the moment, members have cookies that tell LUGNET to munge up a slightly
different version of the interface (ie the rating options). If there was a pool
of volunteer-testers then these testers could have cookies too to let them try
out new features - they wouldn't actually have to use a different server - and
they could still fully participate in LUGNET, just seeing a different version
of the web-interface.
The best thing would be that you could make whatever changes you wanted, but
only the testers would see them.. and the testers wouldn't cause a fuss as they
had signed up because they wanted to help - no casual LUGNET users would get
burned.
If this post has any salient points (it is 4am+), then I'd happily be a
volunteer tester.
Richard
|
|
| |
In lugnet.admin.general, Todd Lehman writes:
> OK, there are now two choices (down from 5, down from 11 originally) in a
> drop-down list...
>
> CHOICE ACTION
>
> - - - No action (default choice).
What about "No Opinion"? I thought it was heavy negative until I tried
submitting it and investigated why it had no effect.
> ° Highlight Recommend for highlighting with a ° symbol.
>
> °° Spotlight Recommend for highlighting with a °° symbol and also
> for inclusion in the LUGNET Spotlight page.
What if you really dig a message, but you don't think it's relevant to
the spotlight - ie - a message on this current topic? Would it be okay to use
'spotlight' then? Or is 'spotlight' only for those messages that you truely and
honestly believe should go into the spotlight?
Is 'Spotlight' saying "This message should be in the spotlight", or "This
message is so great it is spotlight-esque in its funkiness"? Those are two
different things to me.
Richard
|
|
| |
In lugnet.admin.general, Matthew Miller writes:
> > The first choice sends a "-" (meaning "undefined") to /news/rate.cgi as
> > input. The second choice sends a 75; the third choice sends a 100.
>
> Has the interface for rate.cgi changed at all? (I assume not.) I'll
> probably hack slrn this weekend to have rate keys...
It hasn't yet, but it will need to. It will need to reject input values of
0, for one thing -- in order to prevent abuse. (The HTML pages submit either
"-", "75", or "100", but that doesn't limit other clients.) Probably it will
need to only accept that set of three inputs -- or do something to snap to
the closest value in that set. :-)
> PS -- choice of characters rates a °°. :)
Huh?
--Todd
|
|
| |
Todd Lehman <lehman@javanet.com> wrote:
> 0, for one thing -- in order to prevent abuse. (The HTML pages submit either
> "-", "75", or "100", but that doesn't limit other clients.) Probably it will
> need to only accept that set of three inputs -- or do something to snap to
> the closest value in that set. :-)
Ok. I'll make my script just use those. (Or rather, just the last two; I
won't bother with the -.)
> > PS -- choice of characters rates a °°. :)
> Huh?
I like it, of course.
--
Matthew Miller ---> mattdm@mattdm.org
Quotes 'R' Us ---> http://quotes-r-us.org/
Boston University Linux ---> http://linux.bu.edu/
|
|
| |
In lugnet.admin.general, Todd Lehman writes:
> Probably it will need to only accept that set of three inputs -- or do
> something to snap to the closest value in that set. :-)
Does that mean the current 2-choice incarnation is the final one? If the score
display in the group-view is only limited to three choices
(bad/neutral,good,excellent), then there isn't the same danger of feelings
getting hurt - so why not reinstate the 11-point scale, and just display the
results (if at all) in the new gentler fashion? Wouldn't this be a lot better
from a table and search-sorting perspective? IMHO that is more important than
showing the results in the group-view :) (esp when the group-view indicators
are toned-down as to be almost meaningless)
Richard
|
|
| |
Richard Franks <spontificus@__nospam__yahoo.com> wrote:
> (bad/neutral,good,excellent), then there isn't the same danger of feelings
"Bad"? Where's this coming from?
One suggestion might be to change the input field from a dropdown to being a
checkbox. (Although this makes having two choices more complicated.) The
current dropdown is actually indended (if I'm understanding Todd correctly)
to be a multi-stage check box -- unchecked, checked, and double-checked.
The "- - -" choice isn't really a choice at all -- it's a passive
"I'm not going out of my way to highlight this" -- not a "bad", or really
even "neutral".
--
Matthew Miller ---> mattdm@mattdm.org
Quotes 'R' Us ---> http://quotes-r-us.org/
Boston University Linux ---> http://linux.bu.edu/
|
|
| |
Matthew Miller <mattdm@mattdm.org> wrote:
> current dropdown is actually indended (if I'm understanding Todd correctly)
^^^^^^^^
intended
--
Matthew Miller ---> mattdm@mattdm.org
Quotes 'R' Us ---> http://quotes-r-us.org/
Boston University Linux ---> http://linux.bu.edu/
|
|
| |
In lugnet.admin.general, Matthew Miller writes:
> Richard Franks <spontificus@__nospam__yahoo.com> wrote:
> > (bad/neutral,good,excellent), then there isn't the same danger of feelings
>
> "Bad"? Where's this coming from?
It's the category in which all the bad and neutral posts will live (unrated).
> One suggestion might be to change the input field from a dropdown to being a
> checkbox.
If there are going to only be three choices, then I'd prefer buttons - one
click instead of click,locate,scroll,unclick,locate,click.
Richard
|
|
| |
Richard Franks <spontificus@__nospam__yahoo.com> wrote:
> It's the category in which all the bad and neutral posts will live (unrated).
Do you think that this:
> If there are going to only be three choices, then I'd prefer buttons - one
> click instead of click,locate,scroll,unclick,locate,click.
would reduce the perception that unrated = bad? (Assuming that the button is
labelled something like "spotlight"?)
--
Matthew Miller ---> mattdm@mattdm.org
Quotes 'R' Us ---> http://quotes-r-us.org/
Boston University Linux ---> http://linux.bu.edu/
|
|
| |
In lugnet.admin.general, Richard Franks writes:
> In lugnet.admin.general, Todd Lehman writes:
>
> > OK, there are now two choices (down from 5, down from 11 originally) in a
> > drop-down list...
> >
> > CHOICE ACTION
> >
> > - - - No action (default choice).
>
> What about "No Opinion"? I thought it was heavy negative until I tried
> submitting it and investigated why it had no effect.
OK
> > ° Highlight Recommend for highlighting with a ° symbol.
> >
> > °° Spotlight Recommend for highlighting with a °° symbol and also
> > for inclusion in the LUGNET Spotlight page.
>
> What if you really dig a message, but you don't think it's relevant to
> the spotlight - ie - a message on this current topic? Would it be okay to
> use 'spotlight' then? Or is 'spotlight' only for those messages that you
> truely and honestly believe should go into the spotlight?
>
> Is 'Spotlight' saying "This message should be in the spotlight", or "This
> message is so great it is spotlight-esque in its funkiness"? Those are two
> different things to me.
I'll have to find a way to clarify that. It's supposed to mean, "I recommend
that this message be included in the LUGNET Spotlight page."
--Todd
|
|
| |
In lugnet.admin.general, Matthew Miller writes:
> Todd Lehman <lehman@javanet.com> wrote:
> > 0, for one thing -- in order to prevent abuse. (The HTML pages submit either
> > "-", "75", or "100", but that doesn't limit other clients.) Probably it will
> > need to only accept that set of three inputs -- or do something to snap to
> > the closest value in that set. :-)
>
> Ok. I'll make my script just use those. (Or rather, just the last two; I
> won't bother with the -.)
The "-" choice is actually helpful in case you hit the wrong key and
accidentally give input when you didn't mean to. (Or maybe, like LarryP,
you don't ever make misteaks :-) But I use it at least once a day! :)
> > > PS -- choice of characters rates a °°. :)
> > Huh?
> I like it, of course.
I thought you meant something like that, but I didn't (still don't) know what
you mean by "choice of characters rates." I assume you mean that you approve
of the choice of characters -- there's some meta-thing going on there. :)
--Todd
|
|
| |
In lugnet.admin.general, Richard Franks writes:
> In lugnet.admin.general, Todd Lehman writes:
> > Probably it will need to only accept that set of three inputs -- or do
> > something to snap to the closest value in that set. :-)
>
> Does that mean the current 2-choice incarnation is the final one?
Not neccessarily, no. It's conceivable (hopefully unlikely) that the whole
thing is still no good. Need to gather a couple days of feedback, I think,
and switch tasks.
> If the score
> display in the group-view is only limited to three choices
> (bad/neutral,good,excellent),
Three choices -- neutral, excellent, way excellent -- no bad.
> then there isn't the same danger of feelings
> getting hurt - so why not reinstate the 11-point scale, and just display the
> results (if at all) in the new gentler fashion? Wouldn't this be a lot better
> from a table and search-sorting perspective? IMHO that is more important than
> showing the results in the group-view :) (esp when the group-view indicators
> are toned-down as to be almost meaningless)
OK, I think I see what you're saying. Well, it might be safe to allow inputs
of, say, 50 through 100, but allowing 10 or 5 as an input (as it currently
does) is, for all practical purposes, like allowing 0.
--Todd
|
|
| |
Todd Lehman <lehman@javanet.com> wrote:
> you mean by "choice of characters rates." I assume you mean that you approve
> of the choice of characters -- there's some meta-thing going on there. :)
Yes, that's what I was going for. Obviously not too successfully. :)
--
Matthew Miller ---> mattdm@mattdm.org
Quotes 'R' Us ---> http://quotes-r-us.org/
Boston University Linux ---> http://linux.bu.edu/
|
|
| |
In lugnet.admin.general, Richard Franks writes:
> In lugnet.admin.general, Matthew Miller writes:
> > Richard Franks <spontificus@__nospam__yahoo.com> wrote:
> > > (bad/neutral,good,excellent), then there isn't the same danger of feelings
> >
> > "Bad"? Where's this coming from?
>
> It's the category in which all the bad and neutral posts will live (unrated).
Neutral is only bad when 4 out of 5 messages are higher than non-neutral and
the odd neutral stands out for not having been uprated. This new arrangement
(hopefully) will have the opposite -- 4 out of 5 messages neutral and only 1
out 5 non-neutral. And (roughly) 1 out of 100 as Spotlight.
Some munging of old data will be needed of course.
> > One suggestion might be to change the input field from a dropdown to being
> > a checkbox.
>
> If there are going to only be three choices, then I'd prefer buttons - one
> click instead of click,locate,scroll,unclick,locate,click.
That could work for single-article displays. Still needs to be (at the very
least) radio buttons for multiple-article displays (like threads). Chose a
drop-down list because it saves room for other things like "grab a copy of
this article and store it in my personal 'Xyzpdq' folder/channel."
--Todd
|
|
| |
In lugnet.admin.general, Todd Lehman writes:
> In lugnet.admin.general, Richard Franks writes:
> > In lugnet.admin.general, Todd Lehman writes:
> >
> > > OK, there are now two choices (down from 5, down from 11 originally) in a
> > > drop-down list...
> > >
> > > CHOICE ACTION
> > >
> > > - - - No action (default choice).
> >
> > What about "No Opinion"? I thought it was heavy negative until I tried
> > submitting it and investigated why it had no effect.
>
> OK
Clarification: that means "OK, I hear you" -- not necessarily "OK, I'll
change that right away." It's not quite accurate that the default choice
is "no opinion" because it just may be the case that your opinion is not
to highlight it (rather than not caring one way or the other). Probably
something incorporating the word "no" or something equivalent but sounding
like a default is what should be used?
--Todd
|
|
| |
In lugnet.admin.general, Todd Lehman writes:
> It's subtle, but it's like that on purpose so that, basically, the more
> red you see, the more your eye is drawn to it. (Red, biologically, has
> the strongest draw of all colors, for non-colorblind humans.)
I think you should take colorblindness into consideration:
LUGNET is predominantly male, and 1 in 5 males is colorblind.
At a glance, I can't tell red text from black, which is why I suggested you
change the background color:
http://www.lugnet.com/admin/general/?n=6385
The more there is of a color, the more likely I'll notice.
On the other hand, I think the change in font weight for the first post in a
thread works very well. I don't know how well browsers and humans detect
gradual changes in font weight, but I do know that it is possible to vary the
weight using CSS...
Just my .02
--Bram
|
|
| |
In lugnet.admin.general, Bram Lambrecht writes:
> In lugnet.admin.general, Todd Lehman writes:
> > It's subtle, but it's like that on purpose so that, basically, the more
> > red you see, the more your eye is drawn to it. (Red, biologically, has
> > the strongest draw of all colors, for non-colorblind humans.)
>
> I think you should take colorblindness into consideration:
Hmm, doesn't it now? That's why it uses shades of one color (red) rather than
two colors (red & green, or red & blue, for example).
> LUGNET is predominantly male,
Well, it seems that way, but it's not necessarily so. I believe it to be
probably correct, but note that that's without real evidence -- all we
typically ever see is the names of people who have posted messages, which
doesn't of course include lurkers (which, it's known, make up by far the
total number of casual users of the system).
> and 1 in 5 males is colorblind.
> At a glance, I can't tell red text from black, which is why I suggested you
> change the background color:
> http://www.lugnet.com/admin/general/?n=6385
> The more there is of a color, the more likely I'll notice.
Are having difficulty distinguishing the presence of red from the absense of
red, or having difficulty distinguishing red from other colors? The colors
on the symbols actually start out grayish and get redder the more input is
given by people. Something marked very highly from 15 people, for example,
is bright bright bright red. I can't tell red (pink, almost) from black
(gray) when there are only a couple of inputs either -- but that's how it's
intended. That is, the less gray the highlight symbol is, the more seriously
you can take the symbol.
How about this page?--
http://www.lugnet.com/lego/?n=*1,40&v=c
Do you see red or black in the highlight column there?
> On the other hand, I think the change in font weight for the first post in a
> thread works very well. I don't know how well browsers and humans detect
> gradual changes in font weight, but I do know that it is possible to vary the
> weight using CSS...
So the degree symbols at that small point size are too pixel-deficient? Maybe
if they were shown in boldface?
--Todd
|
|
| |
In lugnet.admin.general, Todd Lehman writes:
> How about this page?--
>
> http://www.lugnet.com/lego/?n=*1,40&v=c
>
> Do you see red or black in the highlight column there?
If I look hard, I can see that some are redder than others, right? But I
shouldn't have to look hard. Boldface might help.
--Bram
|
|
| |
In lugnet.admin.general, Richard Franks writes:
> In lugnet.admin.general, Todd Lehman writes:
>
> > OK, there are now two choices (down from 5, down from 11 originally) in a
> > drop-down list...
> >
> > CHOICE ACTION
> >
> > - - - No action (default choice).
As I've said before and as I believe others are saying, there is a possibility
that a linear scale, no matter if it has 100, 10, 5, 2 or only a binary
setting, doesn't work, as article attributes are not one dimensional.
Therefore all the tinkering with how many choices there are and what symbols to
use and so forth may not be addressing the root problem, if there is one.
++Lar
|
|
| |
In lugnet.admin.general, Todd Lehman writes:
> The "-" choice is actually helpful in case you hit the wrong key and
> accidentally give input when you didn't mean to. (Or maybe, like LarryP,
> you don't ever make misteaks :-) But I use it at least once a day! :)
If we're down to just three choices, suggest that each choice get its own
submit button. That's faster even than a radio button, and WAY faster than the
current drop down list. And that improves usability from a human factors
perspective.
Of course I realise that ease of use may not be your highest sort criterion
when deciding what your LDT priorities are, and none of the rest of us have
visibility to either the criteria or the priority list, which is fine.
++Lar
|
|
| |
In lugnet.admin.general, Larry Pieniazek writes:
> If we're down to just three choices, suggest that each choice get its own
> submit button. That's faster even than a radio button, and WAY faster than the
> current drop down list. And that improves usability from a human factors
> perspective.
That's already been asked and answered (in the depths of this thread!) - reason
for not doing it is because when you use the list view (>1 post on same page)
it's more efficient just to have one submit button.
> Of course I realise that ease of use may not be your highest sort criterion
> when deciding what your LDT priorities are, and none of the rest of us have
> visibility to either the criteria or the priority list, which is fine.
True, but if we did it might quell the frequent "When is X likely to be
implememted?" questions? Although it wouldn't help stop the repeated "what
about doing this?" suggestions which might only be quelled by some sort of
feature/suggestion area, rateable or not :)
Richard
|
|
| |
In lugnet.admin.general, Richard Franks writes:
> In lugnet.admin.general, Larry Pieniazek writes:
>
> > If we're down to just three choices, suggest that each choice get its own
> > submit button. That's faster even than a radio button, and WAY faster than • the
> > current drop down list. And that improves usability from a human factors
> > perspective.
>
> That's already been asked and answered (in the depths of this thread!) - • reason
> for not doing it is because when you use the list view (>1 post on same page)
> it's more efficient just to have one submit button.
Easy fix... Different views, different UI.
In the list or tree view, use radio buttons and a single submit. In the single
article view, use multisubmit.
++Lar
|
|
| |
Larry Pieniazek <lar@voyager.net> wrote:
> Easy fix... Different views, different UI.
I think the UI should stay as consistant as possible....
--
Matthew Miller ---> mattdm@mattdm.org
Quotes 'R' Us ---> http://quotes-r-us.org/
Boston University Linux ---> http://linux.bu.edu/
|
|
| |
In lugnet.admin.general, Richard Franks writes:
> That's already been asked and answered (in the depths of this thread!) -
> reason for not doing it is because when you use the list view (>1 post on
> same page) it's more efficient just to have one submit button.
That's half the reason. The other half is keeping it small because it won't
always be the only thing in the article footer.
--Todd
|
|
| |
In lugnet.admin.general, Larry Pieniazek writes:
> Easy fix... Different views, different UI.
> In the list or tree view, use radio buttons and a single submit. In the
> single article view, use multisubmit.
Tradeoffs of that nature (increased speed in exchange for decreased
consistency) can certainly be an improvement from an HCI POV when the UI item
in question is or should be used frequently. But that's not the case here.
--Todd
|
|
| |
In lugnet.admin.general, Todd Lehman writes:
> In lugnet.admin.general, Larry Pieniazek writes:
> > Easy fix... Different views, different UI.
> > In the list or tree view, use radio buttons and a single submit. In the
> > single article view, use multisubmit.
>
> Tradeoffs of that nature (increased speed in exchange for decreased
> consistency) can certainly be an improvement from an HCI POV when the UI item
> in question is or should be used frequently. But that's not the case here.
You're right. Now that we're only using one watered down dimension with a
meaning such that only 1 in 10 articles deserves highlighting and one in 100
deserves spotlighting, you're absolutely right, "recommending" won't be used
all that often... I take back what I said, don't bother dumping any more LDT
into it. It's fine the way it is.
++Lar
|
|
| |
In lugnet.admin.general, Todd Lehman writes:
> Three choices -- neutral, excellent, way excellent -- no bad.
That's another way you could arrange the form:
Comments? * No comment * great post * excellent post [Submit]
where * is a radio button.
Just my .02
--Bram
|
|
| |
In lugnet.admin.general, Bram Lambrecht writes:
> In lugnet.admin.general, Todd Lehman writes:
> > Three choices -- neutral, excellent, way excellent -- no bad.
>
> That's another way you could arrange the form:
>
> Comments? * No comment * great post * excellent post [Submit]
>
> where * is a radio button.
I was just saying "neutral, excellent, way excellent" in reply to something as
a means of clarification. We'd actually like to avoid any kind of qualitative
judgment type words in the choices.
By saying "Highlight" and "Spotlight" (assuming these are clear -- and maybe
they aren't quite clear enough yet) it's not really asking someone to make
any kind of "good" / "great" / "excellent" judgment, but simply to ask whether
or not the post ought to be highlighted or included in the Spotlight page.
It's still subjective as to _why_ or _when_ to suggest Highlight or Spotlight
(that's everyone's personal preference, naturally) but it's objective about
what's being suggested.
--Todd
|
|
| |
> °° Spotlight Recommend for highlighting with a °° symbol and also
> for inclusion in the LUGNET Spotlight page.
Should all this be automatic now? If so, it does not appear to be working for
me. The spotlight page I see is stuck on last Thursday???? Is it just that
there is nothing worth "Spotlighting" - either via rating or by Todd's own
opinion?
Scott A
|
|
|
