Template talk:Talk quote inline/Archive 1

This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Archive 1

Archive 2

Similarity to example font

The following discussion is closed. Please do not modify it. Subsequent comments should be made in a new section.

Why is the default appearance of {{tq}} (like this) ostensibly identcal to that used for examples with {{xt}} (like this), as distinct from that for incorrect examples with {{!xt}} (not like this). I see {{tq}} so rarely, when I first saw it, I thought it was someone misusing the {{xt}} template. It is confusing when discussing on talk pages examples of correct and incorrect uses and someone quotes someone else.

Can {{tq}} be changed to another colour and/or font? —sroc (talk) 08:03, 23 June 2013 (UTC)

I've tried adding the following to my my style sheet:

.inline-quote-talk { color: #FF6600; }

but it does not work. Actual use of the {{tq}} template produces this HTML source:

 … 

Because the style parameter is hard coded into the span tag, does this effectively override the class set in the stylesheet, therefore making it impossible to customise? I can customise the font size, for example, presumably because this is not hard coded. —sroc 💬 11:15, 21 July 2013 (UTC)

Use !important; in your CSS to override the inline styles. — Edokter (talk) — 11:39, 21 July 2013 (UTC)

Egads! Thank you! —sroc 💬 12:17, 21 July 2013 (UTC)

FYI, I've adapted my style sheet as follows:

.inline-quote-talk, .inline-quote-talk2 { color: #000000 !important; font-family: Arial, sans-serif !important; border: solid 1px #1E90FF; border-left:solid 3px #1E90FF; padding: 0px 3px; background-color: #DFEFFF; }

Thus, a short quote would look like this, which reflects a similar style to {{talkquote}} used for block quotes of comments which look like this:

Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.

Perhaps this could be considered for amending the style of {{tq}} generally? —sroc 💬 15:22, 21 July 2013 (UTC)

We generally tend to avoid inline boxes, as they are quite distracting. — Edokter (talk) — 16:06, 21 July 2013 (UTC)

That's fine; it works for me! Still, I think the font colour could be changed to distinguish from {{xt}}. —sroc 💬 17:26, 21 July 2013 (UTC)

I'm all for distinguishing this template from the example templates, thought the orange (#FF6600) is too outstanding for me. — Edokter (talk) — 17:11, 12 October 2013 (UTC)

What about this shade of blue (#1060C0) then? Maybe we can update the colour of the text in the {{talkquote}} format while we're at it?

I quite like the idea of the {{talkquote}} and {{tq}} templates matching.

Then again, it does look a bit like link text. Maybe a lighter shade of orange (#FF8020) would be better?

I quite like the idea of the {{talkquote}} and {{tq}} templates matching.

I say lighter orange to more clearly distinguish from the {{!xt}} template, which uses red (#8B0000). —sroc 💬 05:23, 26 February 2014 (UTC)

RfC: Change the TQ template font colour

There appears to be a consensus for changing the colour of both the TQ and the Talkquote template. The colour that seems to have been settled on is teal (#008560).. Number 5 7 11:19, 15 May 2014 (UTC)

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Extending the RfC for another month in the hope of generating more comments. —sroc 💬 15:20, 15 April 2014 (UTC)]

I propose to change the font colour used in the {{tq}} template used for quoting comments on talk pages, which is currently green and so similar to the green used by the {{xt}} template for example text that these templates are sometimes used interchangeably. It has been suggested to change it to blue (mimicking the colour used by the corresponding {{talkquote}} template) or orange (provided that it is distinguishable from the red used by the {{!xt}} template).

Should the colour of the {{tq}} template be changed?
If so, what colour should it be changed to?
Should the colour of the text in the {{talkquote}} template be changed to match the result, since they have similar purposes?

Please comment below. —sroc 💬 14:14, 16 March 2014 (UTC)

Should the colour of the TQ template be changed?

Yes to avoid confusion with the XT template. —sroc 💬 14:14, 16 March 2014 (UTC)
Yes. I didn't even know about the two templates, or I would have used them from time to time. The green color is very confusing, and TQ should be changed to be consistent with Talkquote, since they have related functions. Reify-tech (talk) 12:39, 9 April 2014 (UTC)
No. Why is it necessary for this template's output to differ greatly from that of {{xt}}? In what scenario would the current similarity cause an actual problem? It's important for {{tq}}'s text to stand out from the text that typically appears alongside it (talk page messages and links), not text appearing in a different context.
Note that the proposed change from green to teal would reduce the text's contrast ratio (when appearing on a standard #f8fcff talk page) from 4.98:1 to 4.63:1. —David Levy 00:32, 17 April 2014 (UTC)

Sorry, I implemented the changes before I saw your comment just now, thinking the discussion may not attract more comments unless people see the changes in action. Feel free to revert them while the discussion continues.

The problem is two-fold. Firstly, editors frequently use {{tq}} instead of {{xt}}, or vice versa, because they see the green serif font used for both cases and don't realise they're different templates. Secondly, quoted text may include an example which is barely perceptible from the surrounding text: "Do I need to give you an example?". Additionally, it would be convenient for the colours used by {{tq}} and {{talkquote}} to match since they serve the same purpose. —sroc 💬 02:10, 17 April 2014 (UTC)

Firstly, editors frequently use {{tq}} instead of {{xt}}, or vice versa, because they see the green serif font used for both cases and don't realise they're different templates.

I've seen that occur on occasion, but how is it substantially detrimental?

Note that the color change also applied to the template's italic mode, activated via the "i=y" parameter or through the {{gi}} ("green italic") wrapper template, whose appearance already differed from that of {{xt}} significantly.

Secondly, quoted text may include an example which is barely perceptible from the surrounding text

In such an instance, a better solution is to use the {{talkquote}} template. But I've never actually encountered that circumstance (in my messages or anyone else's), which surely is relatively uncommon. Conversely, {{tq}} almost always appears alongside default text (and is intended to distinguish quoted material from it), so I'm confused as to why it would make more sense for the template to share that font style.

Additionally, it would be convenient for the colours used by {{tq}} and {{talkquote}} to match since they serve the same purpose.

That sounds fine, but it could be accomplished by simply adjusting the coloration of {{talkquote}} (instead of both templates). —David Levy 03:02, 17 April 2014 (UTC)

"I've seen that occur on occasion, but how is it substantially detrimental?" It is inherently confusing to have text formatted in a particular stand-out format for contrary purposes. For example:

He said "I think it's fine" in relation to the example "I don't like it!"

In any case, as people become conditioned to associate green text with good examples, it seems odd to use virtually the same format for something else entirely, especially given that both often appear in the same discussions. Better to have distinctly different formats to avoid such confusion.

"Note that the color change also applied to the template's italic mode, activated via the "i=y" parameter..." That's fine, but the default is the serif font. I think it would be preferable to have either no font specified or the italic sans-serif as the default (and allow serif as an optional parameter) than using the same font as examples by default. You are meant to use quotation marks anyway when using the {{tq}} template, which already marks the text visually, so no meaning is lost by not changing the font.

"In such an instance, a better solution is to use the {{talkquote}} template." That's not a bad idea. I've generally seen {{tq}} for short quotes (although sometimes used for text spanning several lines) and {{talkquote}} for longer quotes (several lines, a paragraph, or more), so editors may prefer {{tq}} for their shorter quotes that happen to be used in conversations that also include examples.

"...it could be accomplished by simply adjusting the coloration of {{talkquote}} (instead of both templates)." The problem is that the colour for the border of {{talkquote}} was so light that it would not provide a good contrast ratio if it were used for text in {{tq}}. —sroc 💬 04:12, 17 April 2014 (UTC)

It is inherently confusing to have text formatted in a particular stand-out format for contrary purposes.

I'm just not seeing any evidence of non-minor consequences. Occasionally, someone uses use {{xt}} instead of {{tq}} (or vice versa). But what major harm does this cause?

You've cited hypothetical examples of the two templates colliding (and failing to stand out from each other). As I noted, such an occurrence is both relatively uncommon and better addressed via use of the {{talkquote}} template.

In any case, as people become conditioned to associate green text with good examples, it seems odd to use virtually the same format for something else entirely, especially given that both often appear in the same discussions. Better to have distinctly different formats to avoid such confusion.

I've yet to encounter the semantic confusion that you envision. In other words, people might mix up the templates, but I've never seen someone appear to mistake a discussion quotation for example text or example text for a discussion quotation.

That's fine, but the default is the serif font. I think it would be preferable to have either no font specified or the italic sans-serif as the default (and allow serif as an optional parameter) than using the same font as examples by default.

The absence of special styling would greatly reduce the extent to which the quoted text is set apart from the text surrounding it (the template's purpose).

I prefer the italic styling in most instances (as you can see) and don't object to making it the default if there's consensus for that. But it would be advisable to deploy a bot to update preexisting transclusions lacking the "i=y" parameter, some of which depend on the current styling (in particular, when they contain italicized text).

You are meant to use quotation marks anyway when using the {{tq}} template, which already marks the text visually, so no meaning is lost by not changing the font.

Quotation marks often are omitted (hence their absence from the template itself), particularly when including line breaks (as I've done) instead of quoting inline.

The problem is that the colour for the border of {{talkquote}} was so light that it would not provide a good contrast ratio if it were used for text in {{tq}}.

I'm saying that if such consistency is desired, {{tq}}'s green coloration can be applied to {{talkquote}}. —David Levy 05:09, 17 April 2014 (UTC)

Why does it need to be "major harm"? A minor inconvenience is still an inconvenience and if we can avoid it, why shouldn't we? Some uses of {{tq}} might be better off using {{talkquote}}, but: (1) I'm not the template police and I don't go about telling people which template I think they should use; (2) the suggestion to use a different template indicates that the current template is not serving its intended purpose; (3) short quotes are still better off using {{tq}} and these cases can still be confused with {{xt}}.

I'm glad we can agree on making the sans-serif italics the default. I'm not averse to having a bot deploy the proposed s=y parameter to past posts ~~(not technically transclusions)~~ to ensure the current serif format is preserved, but I will leave this to others (it's in the nature of templates that they change, so I'm not sure this is necessary).

Changing the colour of {{talkquote}} but not {{tq}} could promote consistency but would undermine the whole point of the genesis of this change (namely, to avoid confusion with {{xt}}). Sorry I misread your comment on this, as I assumed this was understood and though you were proposing to change {{tq}} but not {{talkquote}} to resolve both issues. —sroc 💬 06:26, 17 April 2014 (UTC)

Why does it need to be "major harm"?

It doesn't. I also used the term "non-minor", and I was tempted to write "non-trivial" (but I was concerned that this might come across as dismissive, which isn't my intention).

I'll be more specific. If someone uses {{xt}} instead of {{tq}} (or vice versa), what consequence beyond "the wrong template was used" is likely to arise?

A minor inconvenience is still an inconvenience and if we can avoid it, why shouldn't we?

If the current setup is causing a minor inconvenience, I would support changes addressing the problem, provided that they don't weaken the template in some other respect.

The template's purpose is to set off quotations from the normal talk page text surrounding them, accomplished via the use of styling and coloration distinct therefrom. Replacing #008000 with #006464 would reduce the color contrast ratio between the quotations and black text from 4.09:1 to 3.01:1. This doesn't pertain to foreground/background contrast (so it isn't the same issue discussed below), but it does relate to readability.

Some uses of {{tq}} might be better off using {{talkquote}}, but: (1) I'm not the template police and I don't go about telling people which template I think they should use;

Nor do I. My point is that a viable solution exists for those who wish to avail themselves of one, so there's no need to adversely impact the majority of {{tq}}'s usage for the sake of accommodating a relatively uncommon application.

(2) the suggestion to use a different template indicates that the current template is not serving its intended purpose;

I was addressing a hypothetical scenario that you cited. As I noted, I've never actually encountered it.

(3) short quotes are still better off using {{tq}} and these cases can still be confused with {{xt}}.

And that brings me back to the question of "what consequence beyond 'the wrong template was used' is likely to arise". If you're aware of instances in which semantic confusion (i.e. someone mistaking a discussion quotation for example text or example text for a discussion quotation) arose, please provide links.

I'm glad we can agree on making the sans-serif italics the default.

And isn't that a sufficient distinction from {{xt}}?

I'm not averse to having a bot deploy the proposed s=y parameter to past posts (not technically transclusions)

How are they not transclusions?

to ensure the current serif format is preserved, but I will leave this to others (it's in the nature of templates that they change, so I'm not sure this is necessary).

I wouldn't describe it as "necessary", but it seems advisable. As I noted, some instances contain italicized text. (I've used {{tq}}'s default style on occasion, specifically for that reason.) —David Levy 08:05, 17 April 2014 (UTC)

Teal (#008080) has a colour contrast with black (#000000) of 4.40:1, which is even stronger than the current green (4.09:1), so would you not support this change? Or #007B7B (4.12:1)? Or #007878 (3.96:1)?

It does happen that editors quote text in {{tq}} that contain {{xt}} (not just me), whether you think they should or not, and editors should be able to do this without having to resort to some other template because of confusing formatting (I especially don't want to be forced into using {{talkquote}} if only quoting a few words or a line just because {{tq}} is confusing). I don't see how changing the colour/font will necessarily "adversely impact the majority of {{tq}}'s usage" if we do it well.

Sorry, I mistook "transclusions" for "substitutions".

By the way, I used a custom CSS to format {{tq}} and {{xt}} differently precisely because I was frustrated with this situation; consequently, if someone uses the wrong template, it stands out like a sore thumb. If we ignore this issue and editors continue to use these templates interchangably because "who cares?", we deprive all users the ability to customise CSS sheets effectively in this way if they choose. It may seem like a trivial matter to you, but every so often someone needs to stand up for the seemingly insignificant. —sroc 💬 08:59, 17 April 2014 (UTC)

Teal (#008080) has a colour contrast with black (#000000) of 4.40:1, which is even stronger than the current green (4.09:1), so would you not support this change? Or #007B7B (4.12:1)? Or #007878 (3.96:1)?

I've suggested another option below.

I don't see how changing the colour/font will necessarily "adversely impact the majority of {{tq}}'s usage" if we do it well.

Agreed. It's worth pursuing changes that everyone finds acceptable.

By the way, I used a custom CSS to format {{tq}} and {{xt}} differently precisely because I was frustrated with this situation; consequently, if someone uses the wrong template, it stands out like a sore thumb.

As a fellow user of custom CSS, I sympathize. But this borders on the topic that I intended to address when writing this essay. (As noted above, however, a change that doesn't adversely impact others is fine.)

It may seem like a trivial matter to you, but every so often someone needs to stand up for the seemingly insignificant.

I'm usually on the other end of this exchange. :-) —David Levy 23:04, 19 April 2014 (UTC)

Yes, greyish, dark blue-green/green/olive, something easy to read but not so attention-grabbing. — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼ 04:25, 23 April 2014 (UTC)

What colour should it be changed to?

Blue, similar to the Talkquote template. —sroc 💬 14:14, 16 March 2014 (UTC)

There don't appear to be many colors available which are suitable for (potentially) large chunks of text which need to remain readable. Perhaps Teal ?? Reify-tech (talk) 12:39, 9 April 2014 (UTC)

Like this? Compared with examples.

And the {{talkquote}} template updated to match and examples set apart.

Not bad. Although it has me wondering whether we need the different font for tq or can just stick with the default font so the examples stand out more. —sroc 💬 05:44, 15 April 2014 (UTC)

Reify-tech has persuaded me to abandon blue due to similarity with the link colour. Teal is better, although still somewhat close to the {{xt}} colour. I think it would be better in this colour but with the default font, which would also match the {{talkquote}} format:

He said "An example of the {{tq}} template with an example included" above.

An example of the {{talkquote}} template with an example included (using teal text).

An example of the {{talkquote}} template with an example included (using black text).

Since there has not been much response to date, should we just apply this change and see who notices? —sroc 💬 15:36, 15 April 2014 (UTC)

The default font should be used, so the examples stand out more, even for color-blind readers. The most important distinctions should be visible regardless of color, which should be used as a supplementary visual aid for those blessed with full color vision. Reify-tech (talk) 16:10, 15 April 2014 (UTC)

Colors I think should be avoided are blue (similar to links), dark blue/purple (similar to visited internal and external links respectively), red (similar to redlinks and {{!xt}}), and anything too light such as yellow. Teal is a good option; it's similar to the color of redirects in my user style, but that shouldn't matter for anyone else since it isn't the default color for redirects in Anomie's script (which is light green). Text marked with background colors is also possible, but may distract too much, especially when using bordered bright-colored boxes like the short quote above. SiBr₄ (talk) 17:05, 15 April 2014 (UTC)
Greyed-out darkish teal/cyan/olive. Should be less distinguished from plain text than output of {{xt}}, but more than that of {{xtn}}. Should not (as current appearance does) attract more attention than the response to it. It should be less contrasty than plain text, so greyish. — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼ 04:23, 23 April 2014 (UTC)

@SMcCandlish: You will see from the discussion below that various shades of teal/blue-green have been considered, such as the following, and I would be interested in your views:

#007878: He asked "What do you think?"
#008560: He asked "What do you think?"

—sroc 💬 05:59, 23 April 2014 (UTC)

@Sroc:: The #007878 one looked better to me. I use an app (F.lux for Mac OS X) that after dusk starts modifying my screen output to be progressively softer and more ruddy (like sunset light, fire/candle light, etc.), and when that's in full swing around midnight, the greenish #008560 is almost grey, but the #007878 is still bluish and distinct. — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼ 02:32, 5 May 2014 (UTC)

Should the colour of the text in the Talkquote template be changed to match the result?

No, as blue could be confused with the link colour, but yes if a different colour is chosen which provides a sufficient contrast with the background. —sroc 💬 14:14, 16 March 2014 (UTC) [edited 14:52, 19 April 2014 (UTC)]
Readability should be paramount, with non-confusion with other uses of color next in priority. The use of blue, green, or red should be avoided, since these colors are so strongly identified with another use already. If a suitable color for the text cannot be found, then it is better to leave the text in Talkquote black (as it is now), and use the color of the box to make a distinction. Reify-tech (talk) 12:39, 9 April 2014 (UTC)
Question: has anyone tested the contrast between the proposed text and background colours, and the colours used in the other templates, against the WCAG 2.0 guidelines? (I'm using a borrowed PC, so don't have my usual tools for doing so to hand). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:24, 15 April 2014 (UTC)
- No. How does one do this? —sroc 💬 18:32, 15 April 2014 (UTC)
  - http://juicystudio.com/services/luminositycontrastratio.php Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:07, 16 April 2014 (UTC)

Thanks on the protip, Pigsonthewing! —sroc 💬 00:30, 17 April 2014 (UTC)

Background colour: #dfefef. Foreground colour: #008080. Result:
The contrast ratio is: 4.03:1
Passed at Level AA for large text only: If the text is large text (at least 18 point or 14 point bold), the luminosity contrast ratio is sufficient for the chosen colours (#dfefef and #008080).
Background colour: #dfefef. Foreground colour: #000000. Result:
The contrast ratio is: 17.73:1
Passed at Level AAA: The luminosity contrast ratio is very good for the chosen colours (#dfefef and #000000).

So black text it is!

{{xt}} text: #006400. Result:
The contrast ratio is: 6.28:1
Passed at Level AA for regular text, and pass at Level AAA for large text: If the text is large text (at least 18 point or 14 point bold), the luminosity contrast ratio is sufficient for the chosen colours at Level AAA; otherwise, Level AA (#dfefef and #006400).
{{!xt}} text: #8B0000. Result:
The contrast ratio is: 8.45:1
Passed at Level AAA: The luminosity contrast ratio is very good for the chosen colours (#dfefef and #8B0000).

What are the link colours? —sroc 💬 23:57, 16 April 2014 (UTC)

According to the monobook.css:

{{xtn}} text (for deprecated text): #696969. Result:
The contrast ratio is: 4.64:1
Passed at Level AA for regular text, and pass at Level AAA for large text: If the text is large text (at least 18 point or 14 point bold), the luminosity contrast ratio is sufficient for the chosen colours at Level AAA; otherwise, Level AA (#dfefef and #696969).
Default link (unvisited) colour: #002bb8. Result:
The contrast ratio is: 8.70:1
Passed at Level AAA: The luminosity contrast ratio is very good for the chosen colours (#dfefef and #002bb8).
Visited link colour: #5a3696. Result:
The contrast ratio is: 7.38:1
Passed at Level AAA: The luminosity contrast ratio is very good for the chosen colours (#dfefef and #5a3696).
"New" link (does not exist) colour: #cc2200. Result:
The contrast ratio is: 4.67:1
Passed at Level AA for regular text, and pass at Level AAA for large text: If the text is large text (at least 18 point or 14 point bold), the luminosity contrast ratio is sufficient for the chosen colours at Level AAA; otherwise, Level AA (#dfefef and #cc2200).

I think we've got a winner, using the light teal (#dfefef) background contrasting well with black foreground text and all the standard link colours and the example template colours. —sroc 💬 00:19, 17 April 2014 (UTC)

Interwiki link colour: #36b. Result:
The contrast ratio is: 4.72:1
Passed at Level AA for regular text, and pass at Level AAA for large text: If the text is large text (at least 18 point or 14 point bold), the luminosity contrast ratio is sufficient for the chosen colours at Level AAA; otherwise, Level AA (#dfefef and #36b).

—sroc 💬 00:27, 17 April 2014 (UTC)

Oh, and of course teal (#008080) on a white background checks out, too:

The contrast ratio is: 4.77:1
Passed at Level AA for regular text, and pass at Level AAA for large text: If the text is large text (at least 18 point or 14 point bold), the luminosity contrast ratio is sufficient for the chosen colours at Level AAA; otherwise, Level AA (#fff and #008080).

—sroc 💬 00:35, 17 April 2014 (UTC)

And, more significantly, on Wikipedia's off-white background colour (#f9f9f9):

The contrast ratio is: 4.53:1
Passed at Level AA for regular text, and pass at Level AAA for large text: If the text is large text (at least 18 point or 14 point bold), the luminosity contrast ratio is sufficient for the chosen colours at Level AAA; otherwise, Level AA (#f9f9f9 and #008080).

—sroc 💬 00:41, 17 April 2014 (UTC)

The text is slightly to bright, so I suggest #006464:

The contrast ratio is: 6.98:1
Passed at Level AA for regular text, and pass at Level AAA for large text: If the text is large text (at least 18 point or 14 point bold), the luminosity contrast ratio is sufficient for the chosen colours at Level AAA; otherwise, Level AA (#fff and #006464).

— Edokter (talk) — 01:37, 17 April 2014 (UTC)

The darker colour (#006464) looks good, but is even closer to the {{xt}} colour (#006400) than the proposed teal (#008080). I think it would be OK though if the default font for {{tq}} is changed to a sans-serif font or, better yet, invoked the sans-serif italics by default:

He said "I don't like it" in relation to the example "I hate it!"

We could switch it to make italics the default and use s=y as an option to implement the serif font, which would still be (slightly) more distinguishable from the example font than it is now:

He said "I don't like it" in relation to the example "I hate it!"

—sroc 💬 05:22, 17 April 2014 (UTC)

Also, as David Levy noted above: "Replacing #008000 with #006464 would reduce the color contrast ratio between the quotations and black text from 4.09:1 to 3.01:1." #008080 actually fares better with a colour contrast of 4.40:1. —sroc 💬 09:20, 17 April 2014 (UTC)

I am more concerned about the contrast with the background color. The contrast with the text color should not be that high anyway, as that hurts readability. I already think the #000080 is too high. — Edokter (talk) — 10:41, 17 April 2014 (UTC)

See my reply below. —David Levy 23:04, 19 April 2014 (UTC)

Yes, i.e. they should be synchronized. — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼ 04:26, 23 April 2014 (UTC)

Consensus building

If I may summarise the common ground on the discussion so far:

The {{tq}} template is intended for use "in talk page discussions to quote, in a visually and CSS-class distinct way, material quoted, e.g. from a guideline or from another editor's previous post."
It is related to the {{talkquote}} template, which is intended "for quoting other editors (or guidelines, etc.) on talk pages, in a block". Since these these templates serve related purposes, it is desirable that they adopt similar formats.
The default format for {{tq}} is serif font in green (example), which is virtually identical to the format used for examples with the {{xt}} template (example). Since these templates are unrelated, there is no reason for them to adopt similar formats that can be confused with each other (and they often are, as some editors mistakenly use them interchangeably).
{{tq}} takes an optional parameter (i=y) which uses italics in the default font (example). It would be preferable to adopt this as the default and allow the current serif font as an alternate option (e.g., s=y where "s" stands for "serif"). However, this alone would not distinguish from the {{xt}} template (without also changing the colour) because: (a) the output would still be similar where the serif option of {{tq}} is invoked; (b) quoted text which uses bold/italic formatting may be confused with {{bxt}}.
The {{tq}} template "does not automatically add quotations marks, which must be appended manually if desired (recommended when inline quotation is used)." Quotation marks provide a visual means of marking the text without relying on the formatting (which avoids accessibility issues), although some editors omit them when using the template.
The colour of the text displayed using the {{tq}} template should contrast well with the background and be distinct from other text (e.g., links, example templates, etc.). There is some concern to ensure that the text is sufficiently distinct from the plain text (David Levy: "The template's purpose is to set off quotations from the normal talk page text surrounding them, accomplished via the use of styling and coloration distinct therefrom"), although it has been suggested that this is not necessary (Edokter: "The contrast with the text color should not be that high anyway, as that hurts readability").

It is therefore proposed that:

The colour of the {{tq}} template be changed to be distinct from that of the {{xt}} and {{!xt}} templates and link colours;
The colour of the border of the {{talkquote}} template be changed to match the same colour (but the text remain in black);
The sans-serif italic font become the default of the {{tq}} template, with serif font as an option.

Colours suggested for the {{tq}} template include:

Colour	Example	Hex code	Contrast with background[1]				Colour difference with surrounding text[2]				Notes
Colour	Example	Hex code	Monobook (`#f8fcff`)	Vector / Modern (`#ffffff`)	Cologne Blue (`#ffffec`)	{{Archive top}} (`#edeaff`)	Black (`#000000`)	{{xt}} (`#006400`)	Link (unfollowed) (`#002bb8`)	Link (followed) (`#5a3696`)	Notes
Green	Example	`#008000`	4.98:1	5.14:1	5.08:1	4.36:1	128	28	269	314	Current colour
Teal (light)	Example	`#008080`	4.63:1	4.77:1	4.72:1	4.05:1	256	156	141	186	Proposed by Reify-tech
Teal (medium)	Example	`#007B7B`	4.94:1	5.10:1	5.04:1	4.32:1	246	146	141	186	Proposed by sroc^{[Note 1]}
Teal (medium-dark)	Example	`#007878`	5.14:1	5.31:1	5.25:1	4.50:1	240	140	141	186	Proposed by sroc^{[Note 2]}
Teal (dark)	Example	`#006464`	6.77:1	6.98:1	6.91:1	5.92:1	200	100	141	186	Proposed by Edokter
Custom shade	Example	`#008560`	4.50:1	4.64:1	4.59:1	3.94:1	~~100~~ 229	129	241	286	Proposed by David Levy

^ as a compromise approximating the contrast levels of the current green
^ revised compromise to reach AA level on #edeaff background

Can we reach any consensus on the precise colour? —sroc 💬 14:52, 19 April 2014 (UTC)

[Talk page background colour (and resultant contrast levels) corrected by David Levy, 19:09, 19 April 2014 (UTC).]

[table expanded by sroc 💬 09:28, 20 April 2014 (UTC)] [colour differences included by sroc 💬 10:00, 20 April 2014 (UTC)] [add contrast with {{Archive top}} by sroc 💬 10:13, 20 April 2014 (UTC)] [added medium-dark teal by sroc 💬 11:15, 20 April 2014 (UTC)] [added shading for values where contrast < 4.50:1 (below AA rating) or difference is < 140 by sroc 💬 11:21, 20 April 2014 (UTC)]

Where did you get the background colour as #f8fcff? That colour is not specified anywhere in cologneblue.css (unspecified), modern.css (specifies #f0f0f0), monobook.css (#f9f9f9) or vector.css (#f3f3f3). —sroc 💬 02:43, 20 April 2014 (UTC)

I'm a bit taken aback. It's part of the English Wikipedia's custom MediaWiki:Monobook.css (and has been since 3 July 2004). I had no idea that this never made it into MediaWiki:Common.css or the other skins. My apologies.

I'm also confused. I was about to revert my changes to the chart and remove my suggestion from consideration (because the color's contrast with #f9f9f9 of #f3f3f3 falls below the "AA" level), but when I view a talk page under Vector, its background color is #ffffff (or #fff), which doesn't appear in MediaWiki:Vector.css. (The same goes for Modern and MediaWiki:Modern.css.) Meanwhile, I'm seeing the color as #ffffec under Cologne Blue (despite its absence from MediaWiki:Cologneblue.css).

So for whatever reason, #f8fcff seems to be the most chromatic talk page background color currently in use here. We really ought to implement a single talk page background color across all of the skins (ideally white or something close to it). —David Levy 04:09, 20 April 2014 (UTC)

How very curious, and frankly baffling. I've no idea why MediaWiki:Monobook.css is wholly different from the link for "monobook/main.css" at Help:Cascading Style Sheets#Levels of CSS settings. —sroc 💬 08:57, 20 April 2014 (UTC)

Following some experimentation, I've added my suggestion (#008560) to the above list. It's a blue-green shade that I regard as sufficiently distinct from #006400 (the shade of green used in {{xt}}), #0645ad (the shade of blue used for internal links) and #3366bb (the shade of blue used for external and interwiki links). Its contrast with the our talk page background color is 4:50:1, which maintains compliance with the WCAG "AA" standard (and the "AAA" level with large text). Its contrast with black text is 4:52:1, making it the most evenly balanced of the options presented.

I'm baffled by Edokter's assertion that a high level of contrast with black text "hurts readability". The template's very purpose is to give quoted text a visual appearance distinct from the black text that surrounds it. Certainly, it's important to meet our normal accessibility standards by ensuring that the quoted text's contrast with the page's background is sufficient, but it also is important that the quoted text stand out, thereby assisting in the comprehension of its contextual meaning. Both enhance readability. —David Levy 23:04, 19 April 2014 (UTC)

I'm not convinced that your suggestion (#008560) is "sufficiently distinct from" {{xt}} (#006400). Take this example:

He said "I didn't propose I didn't do it! but I'll go with it."

The lighter shades of teal provide better contrast:

#008080: He said "I didn't propose I didn't do it! but I'll go with it."
#007B7B: He said "I didn't propose I didn't do it! but I'll go with it."
#006464: He said "I didn't propose I didn't do it! but I'll go with it."

—sroc 💬 02:26, 20 April 2014 (UTC)

I perceive plenty of contrast in the #008560 example, despite the smaller text size. —David Levy 02:39, 20 April 2014 (UTC)

I don't. Maybe my eyesight isn't as good as yours. Maybe my monitor isn't as good as yours. Maybe someone with mild colourblindness would struggle moreso. The point here is to make it easier to distinguish from {{xt}}'s green, which teal clearly does better than blue-green. —sroc 💬 02:48, 20 April 2014 (UTC)

I understand your goal, but I'm reluctant to support a shift so far into the blue range. Green is well established in this context, and it's even been referenced in talk page messages (which suddenly would be rendered nonsensical or misleading). I attempted to find the bluest possible shade that could reasonably be considered "green" (despite being much less green than the shade used in {{xt}}).

However, if white truly is the default skin's talk page background color, that arguably opens up a wider range of options. But before I pursue that avenue, I want to make sure that I'm not somehow seeing weird things on my end. What talk page background color do you get under Vector? —David Levy 04:09, 20 April 2014 (UTC)

How can I find this out? MediaWiki:Vector.css doesn't specify background colours and MediaWiki:Common.css only seems to specify background colours for specific objects/classes, not the whole page. It looks white but might be faintly grey or some other slightly-off-white colour. If I put this text on a white background, I can't see any difference. —sroc 💬 08:57, 20 April 2014 (UTC)

I usually paste screenshots into Adobe Photoshop and check colors via the eyedropper tool.

I don't know what operating system you're running, but there are programs for this purpose. (Here is an example for Windows.) —David Levy 09:51, 20 April 2014 (UTC)

Inkscape can do the same thing, by pasting a screenshot, selecting the "Choose colors" eyedropper and hovering over the page background. The selected color is then shown in RGB at the bottom of the screen. That's what I use. SiBr₄ (talk) 10:05, 20 April 2014 (UTC)

I appreciate that sudden changes to templates could potentially have some impact on existing statements, but that's inherent in using templates: they can, and do, change over time. I'm sure any reference to "green text" that suddenly becomes teal will still be understood and such references will soon fall into the archive pages and be superseded by uses of the revised template with greater clarity between the templates which are currently practically identical. The change is for the greater good.

Another possibility would be to make the {{xt}} green brighter while making the {{tq}} a darker blue-green, but this would involve changing another template which may have an even more widespread impact (I dare say {{xt}} is probably much more widely used than {{tq}}) and may receive stronger opposition (a brighter green may be harder to read). —sroc 💬 09:09, 20 April 2014 (UTC)

I appreciate that sudden changes to templates could potentially have some impact on existing statements, but that's inherent in using templates: they can, and do, change over time.

That's true, but it's reasonable to weigh the pros and cons.

I'm sure any reference to "green text" that suddenly becomes teal will still be understood and such references will soon fall into the archive pages

Many instances already are archived. I still want them to make sense.

I'm reminded of when you wrote that a related issue "may seem like a trivial matter to [me], but every so often someone needs to stand up for the seemingly insignificant." In response, I noted that "I'm usually on the other end of this exchange". Well, here I am.

and be superseded by uses of the revised template with greater clarity between the templates which are currently practically identical.

Keep in mind that I've already endorsed changing {{tq}}'s default typeface. —David Levy 09:51, 20 April 2014 (UTC)

As I said, I appreciate what you're trying to preserve; I believe change is necessary for the greater good, but care should be taken to get it right. That's why I started this RfC in the first place and why we're having these discussions, weighing up the alternatives to find a consensus. —sroc 💬 10:16, 20 April 2014 (UTC)

Likewise, I appreciate your efforts (and the care with which you're approached the endeavor). —David Levy 20:53, 20 April 2014 (UTC)

The background color is just plain #fff in Vector. Though you should also consider the background colors of the various archive section templates, such as #edeaff for {{Archive top}}. SiBr₄ (talk) 09:05, 20 April 2014 (UTC)

Oh, that's a good point. This is getting complicated. —David Levy 09:51, 20 April 2014 (UTC)

Monobook, Modern and Cologne Blue have talk page background colors of #f8fcff, #fff and #ffffec respectively. SiBr₄ (talk) 09:14, 20 April 2014 (UTC)

Thanks, SiBr4!

We have a tool for calculating the contrast of luminosity, but do we have a tool for contrasting different colours (i.e., how easily distinguishable specific shade of "green" and "teal"/"blue-green" are, even if they have similar luminosity)? —sroc 💬 09:19, 20 April 2014 (UTC)

That's a good question. It certainly would be handy. —David Levy 09:51, 20 April 2014 (UTC)

Thanks from me too, SiBr4. I have the same colors, so this confirms that there isn't something peculiar happening on my end. —David Levy 09:51, 20 April 2014 (UTC)

I've found this one and added the results in the table above. —sroc 💬 10:00, 20 April 2014 (UTC)

The results indicate that all of the shades of teal are equally different from the link colours, but the distinction from black and "example green" improve as the shade gets lighter. David's custom shade is more distinguishable from the link colours, but fares badly compared with black and "example green". On the whole, the lighter shades of teal offer the best contrast with all the other text colours (at least 141 across the board) whereas the dark teal and custom shade have mixed results (in some cases dipping as low as 100). —sroc 💬 10:05, 20 April 2014 (UTC)

Teal also contrasts better with {{Archive top}} than does David's custom shade. —sroc 💬 10:13, 20 April 2014 (UTC)

Based on the above testing, I believe medium-dark teal (#007878) is optimal amongst the proposed options because:

It provides a good contrast with the various backgrounds (complies with AA on all backgrounds tested);
It contrasts well with the other text colours (lowest value is 140, well above the current green (with 28 and 128) and David's custom (with 100 and 129) and on par with the others);
It is equidistant in terms of colour difference between "example green" (#006400) and "link blue" (#002bb8), so it is as far as we can move away from green before becoming too blue.

—sroc 💬 11:15, 20 April 2014 (UTC) [added third point 17:12, 20 April 2014 (UTC)]

Note that Template:Archive top was merely an example. Having consulted Category:Archival templates and Category:Deletion archival templates, I've determined that at least fifteen additional colors are widely used for this purpose:

#99ff99 (Template:Sfp create/Template:Sfp top), #bff9fc (Template:Cfd top), #e0e0e0 (Template:AfC-c), #e3d2fb (Template:Mfd top), #e3f9df (Template:Closed/Template:Tfd top), #e5ecf5 (Template:Puf top), #e6f2ff (Template:FAR top), #eeffee (Template:Poll top/Template:RM top/Template:Road top), #f3f9ff (Template:Afd top/Template:Archived WikiProject Proposal top/Template:Cem top/Template:Ffd top/Template:Rfc top), #f5f3ef (Template:Discussion centralize top/Template:Discussion top/Template:Election top/Template:Report top), #f5fff5 (Template:Fpo1 top/Template:Rfap/Template:Rfbagp/Template:Rfbp), #ff9999 (Template:Sfp nocreate), #ffeedd (Template:Rfd top), #fff2e6 (Template:Flr top/Template:Fl top) and #fff5f5 (Template:Fpo top/Template:Rfaf/Template:Rfbagf/Template:Rfbf)

The following colors appear in FPo and RfA templates that apparently receive relatively little use:

#e6f2ff (Template:FPR top) and #ffffe0 (Template:Rfah/Template:Rfbh)

Apart from those with many transclusions and/or association with familiar processes, the extent to which such templates are used can be tricky to gauge (because they're commonly substituted instead of transcluded). Other colors include these:

#eaffea (Template:Archive top green), #efefef (Template:Sfdisc top), #ffe4b5 (Template:Rfcn top) and #ffeaea (Template:Archive top red).

I probably missed some templates (but note that those containing white backgrounds have been omitted purposely). —David Levy 20:53, 20 April 2014 (UTC)

That's great research, but as you said: "This is getting complicated." I doubt whether any of the proposed colours would comply with all of the above and still provide the desired stand-out effect. Take the first example, a bright green (#99ff99), which already clashes with the current "TQ green" (4.19:1); your proposed #008560 is worse (3.78:1); my proposed #007878 is an improvement but still fails "AA" grade (4.33:1); the darker #006464 passes "AA" (5.69:1) but has reduced contrast with plain black text and "example green". I suspect that at some point we will need to make a compromise between contrasting with every possible background versus contrasting with the common text colours. It's a shame that there are so many bright background colours used in various templates which make choosing any safe text colour such a headache! —sroc 💬 21:20, 20 April 2014 (UTC)

Indeed, accommodating all of those colors seems unfeasible. My point is more that edeaff is only one of many, so we shouldn't focus on it in particular.

I agree that far too many colors are in use for this purpose (and some choices, such as the one that you mentioned, seem inherently impractical). I think that consolidation is advisable. We needn't settle on one color, but there's no need to maintain twenty-two. —David Levy 21:56, 20 April 2014 (UTC)

Agreed. I still maintain that #007878 is probably the best compromise because, putting aside the cornucopia of colour in the above templates, it represents the mid-point between "example green" and "link blue" and it contrasts well with the other main text colours and main backgrounds. It also passes against {{archive top}}, which is probably the most common of all these templates, and presumably will cope better than average with most of the others, so it's a good compromise. Certainly an improvement on the current green, which fails on several counts at least on the more common colour combinations (black text, "example green", {{archive top}}, as well as {{sfp create}} and no doubt others). —sroc 💬 03:05, 21 April 2014 (UTC)

Comments such as "I perceive plenty of contrast" and "I don't" are unhelpful; we don't know what hardware, software or settings you are using, nor what lighting conditions you are experiencing, and how your eyes function. I have indicated above tool for objectively measuring contrast, and guidelines for the desired levels of contrast. Please use those. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:59, 20 April 2014 (UTC)

Thanks, Pigsonthewing. My conclusion immediately above is based on the objective data as set out in the table above using the tools provided. —sroc 💬 16:23, 20 April 2014 (UTC)

I'm well aware that personal perception is of limited significance. That's why I wrote "I perceive plenty of contrast." instead of "There's plenty of contrast." —David Levy 20:53, 20 April 2014 (UTC)

In case anyone is wondering why the above table uses "Colour difference" instead of "Contrast" when comparing different font colours, here's an example: khaki brown (#5D4A00) and bright blue (#0000FF) are clearly different colours and unlikely to be confused; they have a colour difference of 422 (which is high) but a luminosity contrast ratio of 1:1 (identical), so clearly colour difference is the better indication for comparing font colours (but they would be horrible as a foreground/background combination: like this). —sroc 💬 17:32, 20 April 2014 (UTC)

Expanded table including some additional archive templates:

Colour	Example	Hex code	Luminosity contrast ratio with background						Colour difference with surrounding text								Notes
			Page background			Common archive templates			Plain text		Template	Links
			Monobook (`#f8fcff`)	Vector / Modern (`#ffffff`)	Cologne Blue (`#ffffec`)	{{Archive top}} (`#edeaff`)	{{Sfp top}} (`#99ff99`)	{{Mfd top}} (`#e3d2fb`)	Vector skin (`#252525`)	Other skins (`#000000`)	{{xt}} (`#006400`)	Internal (unvisited) (`#002bb8`)	Internal (visited) (`#5a3696`)	External / interwiki (`#3366bb`)	Nonexistent (unvisited) (`#cc2200`)	Nonexistent (visited) (`#a55858`)
Green	Example	`#008000`	4.98:1	5.14:1	5.08:1	4.36:1	4.19:1	3.64:1	165	128	28	269	314	264	298	293	Current colour
Teal (light)	Example	`#008080`	4.63:1	4.77:1	4.72:1	4.05:1	3.89:1	3.38:1	219	256	156	141	186	136	426	245	Proposed by Reify-tech
Teal (medium)	Example	`#007B7B`	4.94:1	5.10:1	5.04:1	4.32:1	4.16:1	3.61:1	209	246	146	141	186	136	416	235	Proposed by sroc (#1)
Teal (medium-dark)	Example	`#007878`	5.14:1	5.31:1	5.25:1	4.50:1	4.33:1	3.76:1	203	240	140	141	186	136	410	229	Proposed by sroc (#2)
Teal (quite dark)	Example	`#006A6A`	6.23:1	6.42:1	6.35:1	5.45:1	5.24:1	4.55:1	175	212	112	141	186	136	382	201	Suggested by sroc (#3)
Teal (dark)	Example	`#006464`	6.77:1	6.98:1	6.91:1	5.92:1	5.69:1	4.95:1	163	200	100	141	186	140	370	189	Proposed by Edokter
Custom shade	Example	`#008560`	4.50:1	4.64:1	4.59:1	3.94:1	3.78:1	3.29:1	192	~~100~~ 229	129	241	286	173	399	218	Proposed by David Levy
Legend			< 375:1 < 400:1 < 425:1 < 450:1						< 75 < 100 < 120 < 140

[External/interwiki link colour added 02:22, 23 April 2014 (UTC). Vector skin plain text colour and nonexistent page link colors added 21:26, 23 April 2014 (UTC).]

The more background colours we compare against, the more attractive darker text looks to distinguish, but this simultaneously diminishes contrast with surrounding colours. I've added another shade of teal for comparison, although I don't necessarily endorse it as I think it is more important to provide contrast with {{xt}} (where confusion is more likely to arise) than "AA" compliance with every possible template (where the discussion is closed and the text will nonetheless be readable if a little bit of a strain):

#006A6A: He said "I didn't propose I didn't do it! but I'll go with it."

My preferred option remains #007878 (but I would be happy with a lighter version of teal or a slightly darker version if that's where the consensus lies):

#007878: He said "I didn't propose I didn't do it! but I'll go with it."

—sroc 💬 05:59, 21 April 2014 (UTC) [corrected error in table —sroc 💬 14:45, 22 April 2014 (UTC)]

I agree that we shouldn't place too much emphasis on the multitude of archive template background colors in use, whose needless diversity should be reduced greatly.

However, I disagree with the premise that we should seek the "mid-point between 'example green' and 'link blue'". The latter is part of Wikipedia's native interface, while the former is merely another template's output, so I don't see why they should have equal standing.

You initiated this discussion, in good faith, because you regard {{tq}}'s default styling as too similar to that of {{xt}}. This is a valid concern, and it's reasonable to consider tweaking {{tq}}'s color. But you essentially propose that it be changed from the traditional green (selected specifically because of its absence from Wikipedia's native interface) to blue, thereby making it less distinct from links and more distinct from {{xt}}. But what about the countless other templates in use? Do you know for certain this won't result in greater similarity to others? (I'm not aware of any in particular, but I never would have guessed that we have at least twenty-two discussion archive colors in use.)

As I noted, I'm willing to switch to a much bluer shade of green. But moving to blue outright requires stronger justification than a clash with another template that you happen to use.

Incidentally, what's the "< 140" cutoff's basis? Is a minimum color difference of 140 recommended somewhere? —David Levy 14:19, 21 April 2014 (UTC)

I've only noticed any issues with the {{xt}} template which (along with {{!xt}}) is the one I see used the most. There may well be other templates that invoke colour formatting, but if we can't think of them and they do not arise (often) in practice, then they are unlikely to present (as much) of a problem (compared with the present state of affairs, at least – for the greater good).

I'm not aware of any particular "standard" for basing a comparison between text colours; I arbitrarily chose 140 since the comparison between all varieties of teal being considered with the unvisited link colour was 141, so it seemed like as good a place as any the draw a line in the sand. The shading is really just a visual aid to help see where the problem areas lie. —sroc 💬 15:05, 21 April 2014 (UTC)

You arbitrarily chose a number that places your preferred color's difference from #006400 (140, a 400% increase) inside the recommended range and my suggested color's difference (129, a 360% increase) in the "problem area".

We could just as easily define the "problem area" as 240 and below, thereby placing your preferred color in pink across all four categories and mine only in the two categories that the template's current color fails. —David Levy 16:09, 21 April 2014 (UTC)

As I said, the shading is merely a visual aid. The numbers are more telling and the categories where your proposal fails fail quite badly. As I have also said, I believe it is necessary to strike a balance between distinguishing from common backgrounds and distinguishing from other colours, hence my attempts to find a middle-ground. —sroc 💬 23:47, 21 April 2014 (UTC)

As I said, the shading is merely a visual aid.

It aids in identifying thresholds that you've acknowledged are arbitrary and intentionally selected to "draw a line in the sand" just below the numerical result in a particular category when your preferred change — a switch to teal — is made. (If the minimum were 142 instead of 140, every color except the current one and my suggestion would fail in that category.)

This is potentially misleading, particularly given your use of exactly the same color-coding to indicate noncompliance with the WCAG "AA" standard. Until you explained otherwise, I thought that perhaps you'd obtained the "140" figure from the same source or another authority.

The numbers are more telling and the categories where your proposal fails fail quite badly.

They fail under the formula that you invented (with the teal shades' preexisting results in mind). I'm no more an authority on the subject than you are, but "100" strikes me as a more reasonable minimum. By that equally arbitrary criterion, every shade except the current one passes in all four categories.

Regarding the comparison with #006400, I find it rather odd that you regard a color difference of 140 (which happens to be the exact minimum that you unilaterally set) as sufficient, while 129 "fails quite badly".

Or were you referring to the "Luminosity contrast ratio with background" portion of the chart, wherein my suggested color fails when paired with the three archive template background colors that you selected out of at least sixteen in wide use? (To be clear, in my previous message, I was referring specifically to the "Colour difference with surrounding text" portion.)

As I have also said, I believe it is necessary to strike a balance between distinguishing from common backgrounds

If you're going by commonness, you should note that #99ff99 and #e3d2fb certainly aren't among the three most widely used archive template background colors. (I don't know which three are, but #edeaff is the only realistic possibility of the three that you selected. Out of the sixteen whose use I know to be substantial, #99ff99 probably is near the bottom of list.)

and distinguishing from other colours, hence my attempts to find a middle-ground.

I've explained that I disagree with the idea of "striking a balance" in which Wikipedia's link coloration and that of a template that you happen to use are assigned equal standing.

Nonetheless, I've consented to a change resulting in a 360% increase in color difference between {{tq}} and {{xt}} (with the former barely retaining some semblance of its traditional color), but you're holding out for 400%. —David Levy 01:23, 22 April 2014 (UTC)

Sorry, when I said "fail quite badly", I was indeed referring to the "AA" standard of the luminosity contrast ratio (which is not arbitrary). The three templates I chose were the ones that seemed to be the boldest in the blue–green spectrum and therefore seemingly the most likely to clash with all of the proposed text colours and hence provide the best indication of where problems lie.

I assume by 360% increase, you refer to the change from 28 to 129. Expressing this as a percentage is misleading since the original 28 is very bad so any change can seem significant by comparison. If the difference was originally 12, you could claim a 1075% increase, but that still doesn't mean the new version is really good enough. We have the opportunity to make the difference much more distinct, yet you seem to be holding out. The only reason I see for avoiding a shift from green towards blue–green is to avoid conflicting with blue link colours, and this is why I'm trying to find a happy medium. There's a range of shades from #007B7B to #006A6A (with some leeway on either side) that I would be happy with, but I think that #008560 is still too close to #006400 for the difference to stand out. Of course, this isn't about what you or I want, but about reaching consensus, and I'm at least happy that we're moving in the right direction. —sroc 💬 01:52, 22 April 2014 (UTC)

I assume by 360% increase, you refer to the change from 28 to 129.

That's correct.

Expressing this as a percentage is misleading since the original 28 is very bad so any change can seem significant by comparison.

My comparison was between your preferred color's difference increase and my suggested color's difference increase, so this wasn't misleading at all.

If the difference was originally 12, you could claim a 1075% increase,

That would be a 975% increase, actually. (You didn't subtract the original 12 from 129.) And in such a scenario, your preferred color would provide a 1067% increase. The relevant comparison is between the two percentages.

In a direct comparison (without regard for {{tq}}'s current color), your preferred color's difference from that used in {{xt}} is 8.53% greater than my suggested color's difference is. But you've unilaterally deemed 140 "passing" and anything below (including 129) "failing".

but that still doesn't mean the new version is really good enough.

But your preferred color, which provides 8.53% greater difference, is good enough because its score is the exact minimum that you selected arbitrarily.

Oddly, you've also proposed an alternative (not your first choice, but another with your support) whose score in that category is only 112 (so it actually provides less difference from {{xt}}'s color than my suggested color does).

We have the opportunity to make the difference much more distinct, yet you seem to be holding out.

I strongly disagree that 8.53% is "much more".

The only reason I see for avoiding a shift from green towards blue–green is to avoid conflicting with blue link colours, and this is why I'm trying to find a happy medium.

I've explained why I disagree with the premise that we should seek the precise middle ground between Wikipedia's links and a template that you happen to use.

But I've agreed to a compromise that would result in 90% of the color difference increase that you seek.

There's a range of shades from #007B7B to #006A6A (with some leeway on either side) that I would be happy with, but I think that #008560 is still too close to #006400 for the difference to stand out.

Based on the aforementioned numbers, can you see why I find this difficult to understand? —David Levy 03:07, 22 April 2014 (UTC)

Sorry, you are of course right that I should have said a 975% increase, not 1075%. My point is that such comparisons are unhelpful when the starting point is so skewed. We should be looking at the end result, not the comparison with what was. I haven't designated anything as "pass" or "fail" on the "colour difference side". As I've said, I proposed #007878 because it is as far as we can go from "example green" towards blue–green before becoming more like "link blue". I added #006A6A to the table as a further compromise for the sake of comparing another shade trying to balance the luminosity contrast ratio and the colour difference level, but I haven't proposed this colour as such.

I'm not sure the best way to objectively demonstrate this, divorced of system settings, monitor settings, visual perception, etc., but to my eyes all the shades of teal look different from "example green" whereas your custom shade looks distinctly green and very similar to "example green". Admittedly, the different font is doing most of the heavy lifting in the examples, which is why I think the switch to italics by default is necessary. Perhaps we should consider altogether different colours? —sroc 💬 04:27, 22 April 2014 (UTC)

My point is that such comparisons are unhelpful when the starting point is so skewed.

And my point is that you're ignoring essential context and attempting to refute a claim of significance that I haven't made.

Again, I'm not citing these percentages in isolation. The intent is not to demonstrate the extent to which any particular option improves upon the status quo, but to compare your color's improvement with my color's improvement. Whether this is expressed as "400% vs. 360%", "1067% vs. 975%" or "2700% vs. 2480%" is immaterial. The point is that the two numbers in the pair (your preferred color's difference increase and my suggested color's difference increase) are not far apart.

I've also compared the color difference increases directly (thereby setting aside the numbers that you regard as misleading) and determined that my suggested color's use would result in 90% of the color difference increase that your preferred color's use would.

We should be looking at the end result, not the comparison with what was.

And the end result, as noted above, is that your preferred color would provide 8.53% greater difference from #006400 than my suggested color would. This has nothing to do with {{tq}}'s current color. It's simply a comparison between your preferred color's difference from {{xt}}'s color and my suggested color's difference from {{xt}}'s color.

I haven't designated anything as "pass" or "fail" on the "colour difference side".

And you don't see how this is implied by your use of color-coding identical to that indicating noncompliance with the WCAG "AA" standard?

As I've said, I proposed #007878 because it is as far as we can go from "example green" towards blue–green before becoming more like "link blue".

And as I've said, I disagree with the premise that seeking the exact middle ground is desirable.

I'm not sure the best way to objectively demonstrate this, divorced of system settings, monitor settings, visual perception, etc., but to my eyes all the shades of teal look different from "example green" whereas your custom shade looks distinctly green and very similar to "example green".

And to my eyes, the difference between #006400 and #008560 is clear as day. I thought that we'd agreed to set aside these conflicting personal perceptions and rely upon the the tool that you found. —David Levy 05:42, 22 April 2014 (UTC)

We did agree on using the tool; the disagreement surrounds how we use it, since there is no "redline" for a "pass" or "fail" on a particular standard, as far as we are aware. In absence of an objective standard: it's my view that the optimal position is to find a mid-point between potentially conflicting colours in order to provide as much distance as possible; it's your view that this is unnecessary, but I'm not sure why you arrived at the arbitrary colour that you have, which provides less distinction with black and "example green" than my proposal and fails the "AA" test against any background other than the default skin backgrounds. —sroc 💬 06:34, 22 April 2014 (UTC)

We did agree on using the tool; the disagreement surrounds how we use it,

My statement that we agreed to use the tool was made in response to a comment regarding how the various options look "to [your] eyes".

since there is no "redline" for a "pass" or "fail" on a particular standard, as far as we are aware.

So you arbitrarily drew a "line in the sand" at 140, your preferred color's exact difference from #006400. 139 falls in the pink "problem area", but 140 is the magic number. And you acknowledge that you purposely selected a cutoff just below the teal shades' difference from #002bb8. In other words, you performed tests and reverse-engineered a criterion precisely calibrated to your desired outcome.

It just occurred to me that you omitted the external/interwiki link color (#3366bb) from your charts. Your preferred color's difference therefrom is 136. (The same goes for all of the other teal shades suggested, excepting Edokter's, which scores 140.) My suggested color's difference from #3366bb is 173.

Also, I just noticed that you have my suggested color's difference from #000000 misstated as "100". (The actual number is 229.)

In absence of an objective standard: it's my view that the optimal position is to find a mid-point between potentially conflicting colours in order to provide as much distance as possible;

Yes, you've reiterated this several times. And you apparently define "potentially conflicting colours" as those belonging to Wikipedia's native interface, along with templates that you personally use and archive templates that you've handpicked from a large list. And all of these have equal standing.

it's your view that this is unnecessary,

That makes my position seem dismissive. It's my view that certain considerations carry more weight than others, not that it's unnecessary to consider such matters.

but I'm not sure why you arrived at the arbitrary colour that you have,

As I explained, "I attempted to find the bluest possible shade that could reasonably be considered 'green' (despite being much less green than the shade used in {{xt}})", while adhering to a minimum contrast ratio of 4:50 on the background color #f8fcff.

which provides less distinction with black and "example green" than my proposal

Indeed, compared with your preferred color, my suggested color provides 95.4% of the difference from #000000 (240. vs 229) and 92.1% of the difference from #006400 (140 vs. 129). Meanwhile, compared with my suggested color, your preferred color provides 78.6% of the difference from #3366bb (173 vs. 136), 65% of the difference from #5a3696 (286 vs. 186) and 58.5% of the difference from #002bb8 (241 vs. 141).

and fails the "AA" test against any background other than the default skin backgrounds.

You tested all of them (not just the three that you handpicked for inclusion in the chart, two of which your preferred color fails against too)? —David Levy 08:47, 22 April 2014 (UTC)

Firstly, my apologies for any errors in the table; I assure you they are genuinely errors, probably from the copy-pasting in all the testing; I have corrected the error you mentioned, but please feel free to correct any others. I also apologise if you feel that I have mischaracterised your position in my summary; I am trying to understand your point of view better. I did not mean to imply that I had tested your colour against every template background colour that exists; my point was that your custom shade is right on the border with the darkest of the skin backgrounds, which itself is very light #f8fcff and barely imperceptible from white, so virtually any shading would fail "AA" standard against your text (other than very light shading which would not interfere with any other text colours).

What concerns me is your comment: "I attempted to find the bluest possible shade that could reasonably be considered 'green'". If I understand you correctly, you have come at this from the perspective that any substituted colour must be a shade of green; hence, any shade of teal would fail your test as being too blue, irrespective of the objective test results. The whole point of this exercise (i.e., to avoid confusion with "example green") is undermined if a criterion is that the new colour must be green. —sroc 💬 14:45, 22 April 2014 (UTC)

Firstly, my apologies for any errors in the table; I assure you they are genuinely errors, probably from the copy-pasting in all the testing;

No assurance is needed; the idea that this was anything other than an honest mistake never crossed my mind.

I have corrected the error you mentioned, but please feel free to correct any others.

I just corrected the same error in the original chart. One of your messages contains a couple of mentions of the "100" score (with one referring to both my suggested color and Edokter's), which I don't feel comfortable editing myself. (The statements would no longer make sense.)

Do you want me to add the external/interwiki link color to the charts? (That's an omission, not an outright error.)

I also apologise if you feel that I have mischaracterised your position in my summary; I am trying to understand your point of view better.

I'm not sure that "mischaracterised" is the best description, but I think that you might have an inaccurate impression of my attitude.

You seem to perceive my disagreement as dismissive and obstructionist, as though I'm stubbornly blocking obvious improvement because I view it as unnecessary and unimportant. From my perspective, I'm attempting to find a compromise that addresses your concern without introducing new problems. My personal preference would be to retain the shade of green currently in use, but I'm willing to sacrifice that "for the greater good".

I did not mean to imply that I had tested your colour against every template background colour that exists; my point was that your custom shade is right on the border with the darkest of the skin backgrounds, which itself is very light #f8fcff

As I've explained, that's no coincidence. I approached the endeavor with the understanding that a minimum contrast ratio of 4.50:1 on the background color #f8fcff was required. So as I experimented, if a color failed that test, I tweaked it until it passed.

and barely imperceptible from white,

They're night and day on my end, but let's set aside these conflicting personal perceptions and focus on the test results.

so virtually any shading would fail "AA" standard against your text (other than very light shading which would not interfere with any other text colours).

On the chart, only the two dark shades of teal fare well in combination with the archive template background colors, with your preferred color barely passing the #edeaff test and failing the other two (while the two lighter shades of teal fail across the board).

As we've discussed, many of the archive template background colors are simply poor selections (and nothing that we decide here can fix that). They probably shouldn't be much darker than #f8fcff, a color chosen specifically because it provided reasonable contrast with text.

But as I noted 2.5 days ago, I'd like to experiment further. I just don't want to waste my time.

What concerns me is your comment: "I attempted to find the bluest possible shade that could reasonably be considered 'green'". If I understand you correctly, you have come at this from the perspective that any substituted colour must be a shade of green; hence, any shade of teal would fail your test as being too blue, irrespective of the objective test results.

At no point have I asserted (nor do I believe) that we "must" use a particular color. I've stated unambiguously that I favor retaining some semblance of the traditional green coloration, so I've experimented with that objective in mind.

The whole point of this exercise (i.e., to avoid confusion with "example green") is undermined if a criterion is that the new colour must be green.

Setting aside the word "must" again, you seem to be saying that no shade of green is acceptable, even if its difference from the shade used in {{xt}} is 90% as great as that of your preferred teal shade.

You've stated that #007878 is "is equidistant in terms of colour difference between 'example green' (#006400) and 'link blue' (#002bb8), so it is as far as we can move away from green before becoming too blue." This relies on the assumption that our goal is to find a color that's as different from both "link blue" and "example green" as possible. As I've explained, I disagree with that premise. The former is part of Wikipedia's native interface, affecting every page on the site. The latter relates to a template that you happen to use, which isn't elevated to equal standing simply because avoiding confusion with it is the impetus behind your proposal.

Green was selected for use in {{tq}} because it doesn't clash with Wikipedia's native interface. Blue, conversely, does. So I believe that the sensible course of action is to add as much blue as we can to {{tq}}'s color before it ceases to be green. This reduces the difference from the various link colors somewhat, but the trade-off is justifiable.

Compared with my suggested color, your preferred color provides 8.53% greater difference from {{tq}}'s color and 4.8% greater difference from black.

Compared with your preferred color, my suggested color provides 27.2% greater difference from the external/interwiki link color, 53.76% greater difference from the visited internal link color, and 70.92% greater difference from the unvisited internal link color.

You're arguing that 70.92% greater difference from unvisited internal links is outweighed by 8.53% greater difference from a template that you use. You're placing the pursuit of parity (based on the premise that the site's UI and Template:Xt have equal standing) before the overall benefits. —David Levy 20:24, 22 April 2014 (UTC)

Thanks for your patience, constructive criticism and rational explanations (even if I don't agree with them all). Please do not edit my previous comments; I intentionally used markup to show the correction so any subsequent comments would still make sense historically. Please feel free to update the table to include the interwiki link colour, which would be great, and let's revisit the colour options once we have that data.

As I mentioned below, in order for this change to be effective in distinguishing {{tq}} and {{xt}}, I think that it is best both: (a) to tweak the colour of this template; and (b) to make the italics option the default. I believe you agree on this, we just haven't settled on the exact colour yet. If only the colour changes, however, I'm not convinced that this will be enough to avoid confusion between these templates. I trust you understand where I am coming from, even if you disagree. —sroc 💬 00:40, 23 April 2014 (UTC)

Please do not edit my previous comments;

As I noted, I would't feel comfortable editing them (even if you wanted me to).

Please feel free to update the table to include the interwiki link colour, which would be great, and let's revisit the colour options once we have that data.

Done. I included your color-coding, the validity of which I contest.

As I mentioned below, in order for this change to be effective in distinguishing {{tq}} and {{xt}}, I think that it is best both: (a) to tweak the colour of this template; and (b) to make the italics option the default. I believe you agree on this, we just haven't settled on the exact colour yet.

As I noted on the 17th, I "don't object to making it the default if there's consensus for that."

If only the colour changes, however, I'm not convinced that this will be enough to avoid confusion between these templates.

Are you prepared to settle for mitigating the problem (with the understanding that other factors outweigh the importance of "avoid[ing] confusion between these templates")?

I trust you understand where I am coming from, even if you disagree.

Yes, I understand (and I hope that we can reach a solution that satisfies everyone involved). —David Levy 02:22, 23 April 2014 (UTC)

Consensus building (break)

Sorry, I hope my "Please do not edit my previous comments" line didn't come across as suggesting you would; I noted that you had said you wouldn't in your earlier comment and meant only to acknowledge my appreciation for that.

Thank you for adding the external/interwiki link colour to the table. I note that this colour (#3366bb) is closer to all shades of teal (colour difference: 136) than your custom shade (colour difference: 173). Even so, this is not much different than the result for the internal link colour (#002bb8), which is only slightly further from teal (colour difference: 140). I would still prefer medium-dark teal (#007878) to your custom shade (#008560) on the basis that:

it provides a more even difference from all of the text colours in consideration, including {{xt}} (I acknowledge that you disagree with this reasoning);
it provides greater contrast with backgrounds with some shading beyond the default skin backgrounds (although I agree that this is not a paramount consideration and we would be better off dispensing with dark backgrounds in templates altogether).

Despite this, upon further consideration, I think I could justify settling on your custom shade for {{tq}} provided that the italics option (or some formatting other than the current serif) is set as the default, which should hopefully be enough to avoid confusion with {{xt}}. If that change doesn't happen, then I fear all of this will have been in vain (so, no, I would be disappointed with merely "mitigating the problem" rather than resolving it).

Change colour only	Change colour and font style
Whereas I previously said "I'm not convinced", I can compare this with examples like this and see that they could still be confused.	Whereas I previously said "I'm not convinced", I can contrast this with examples like this and see that they are probably far enough apart.

If you would support a change to Levy custom shade italics, would you also agree with updating the border of {{talkquote}} to match, since the templates are related? —sroc 💬 04:25, 23 April 2014 (UTC)

I note that this colour (#3366bb) is closer to all shades of teal (colour difference: 136) than your custom shade (colour difference: 173). Even so, this is not much different than the result for the internal link colour (#002bb8), which is only slightly further from teal (colour difference: 140).

And my suggested color's difference from #006400 (129) is only slightly lower than that (as well as your preferred color's difference therefrom, 140), but you've cited this as a deficiency.

You unilaterally defined the "problem area" as "< 140" (which you based on the teal shades' preexisting test results), and now that you've learned that your preferred color's difference from the one used for our external/interwiki links is 136, you seek to downplay the distinction.

Well, you needn't convince me. I contested the color-coding's validity before, and I continue to contest it. (I won't suddenly change my mind simply because your preferred color now has has a pink mark too.) Had you taken #3366bb into account at the time, you'd have selected a cut-off of 135 or 130. The arbitrary color-coding serves no meaningful purpose in this discussion. Can we please remove it?

it provides a more even difference from all of the text colours in consideration, including {{xt}} (I acknowledge that you disagree with this reasoning);

I certainly do. In comparisons with {{xt}}'s color and black, your preferred color beats my suggested color by an average of 6.67%. In comparisons with the three link colors, my suggested color beats your preferred color by an average of 50.62%.

So yes, my suggested color's differences from the other text colors are significantly less even, but that's mainly because three of the five are much greater.

it provides greater contrast with backgrounds with some shading beyond the default skin backgrounds (although I agree that this is not a paramount consideration and we would be better off dispensing with dark backgrounds in templates altogether).

Indeed. Both your preferred color and my suggested color fail against many of the archive templates' background colors. A problem exists, but neither of us created it.

Despite this, upon further consideration, I think I could justify settling on your custom shade for {{tq}} provided that the italics option (or some formatting other than the current serif) is set as the default, which should hopefully be enough to avoid confusion with {{xt}}.

Given the arguments advanced by RexxS, I think that we should begin considering alternative formatting options. A different font might work.

If that change doesn't happen, then I fear all of this will have been in vain (so, no, I would be disappointed with merely "mitigating the problem" rather than resolving it).

I hope that such an outcome can be averted. If not, coping with your disappointment (and accepting that an improvement is better than nothing) would be much more constructive than declaring that "all of this [has] been in vain" would be.

If you would support a change to Levy custom shade italics, would you also agree with updating the border of {{talkquote}} to match, since the templates are related?

If you're asking me, I'd be fine with that. —David Levy 06:07, 23 April 2014 (UTC)

Regarding the colour, I dispute that I "seek to downplay the distinction" regarding the shading. I admitted that the numbers for each shaded range were arbitrary, but not baseless. The numbers are what's important here, but the colours serve their purpose, too, providing a visual cue of where problems lie. You can change the colours and gradients and fill them throughout the table like a temperature chart if you like, but I'd prefer we didn't abandon the shading altogether. I was rather hoping that we had agreed on a colour – the custom shade you had proposed – that this would be redundant now, anyway.

Regarding the font formatting, I still think the simplest way to go is make italics the default, since this is already an option that many editors will be familiar with and indeed you prefer yourself; if the loss of emphasis is an issue, the template can be adjusted so that wiki markup can switch italics to non-italics within the template. I'm open to exploring other alternatives, but I'm not sure adding another font into the mix is ideal nor how many (if any) other font-family options there are that we can be reasonably confident users will have on their systems and which provide the desired contrast between both plain text and {{xt}}. Actually, what's the serif font that's recently been introduced for headings? —sroc 💬 10:24, 23 April 2014 (UTC)

Regarding the colour, I dispute that I "seek to downplay the distinction" regarding the shading.

In your previous message, you stressed that your preferred color's difference from the external/interwiki link color (136) is "not much different than the result for the internal link colour" (140). And I agree. 136 is close to 140.

But you drew a "line in the sand" at 140, which you purposely selected to accommodate the teal shades (and with no other basis). You did so with the knowledge that 140 was your preferred option's exact lowest score (at the time) and that my suggested color scored 129 in one category (and 100 in another, but that was an error). In other words, the formula was tailored to deliver your desired outcome; the "line in the sand" would separate the colors that you liked from those that you didn't.

To be clear, I'm not accusing you of anything underhanded. I don't believe that you set out to mislead anyone or gain an unfair advantage. I think that you were so confident in the light/medium teal shades' superiority that you genuinely perceived a tangible distinction. I'm imagining a thought process along the lines of "These colors seem good and those seem bad, so this must be where we should draw the line." But your good intentions don't change the fact that the color-coding has absolutely no objective basis.

I admitted that the numbers for each shaded range were arbitrary, but not baseless.

Indeed. You based them on your preferred colors' preexisting scores.

The numbers are what's important here, but the colours serve their purpose, too, providing a visual cue of where problems lie.

They indicate where you decided the problems lie (by determining your preferred color's lowest scores and deeming anything lower a "problem").

The four non-Edokter teal shades share a #3366bb score of 136, which appears in pink only because that comparison was omitted at the time. Otherwise, you'd have drawn the "line in the sand" at at 135 or 130. So how is this "visual cue" of any value?

You can change the colours and gradients and fill them throughout the table like a temperature chart if you like, but I'd prefer we didn't abandon the shading altogether.

I'm not interested in replacing arbitrary color-coding with different arbitrary color-coding. We have no guidelines regarding what minimum color difference is desirable in this context, and neither of us is qualified to invent them.

I was rather hoping that we had agreed on a colour – the custom shade you had proposed – that this would be redundant now, anyway.

If that's so, I don't see why removing the color-coding would be a problem.

But please keep in mind that consensus requires more than your agreement and mine.

Regarding the font formatting, I still think the simplest way to go is make italics the default, since this is already an option that many editors will be familiar with and indeed you prefer yourself;

Indeed, I prefer it in most instances. But other editors' opinions (including those of RexxS) matter too. Certainly, we should at least test some alternatives.

if the loss of emphasis is an issue, the template can be adjusted so that wiki markup can switch italics to non-italics within the template.

RexxS has explained why that's problematic. Do you honestly believe that the importance of preventing {{tq}}'s confusion with {{xt}} outweighs the importance of delivering standards-compliant markup?

Actually, what's the serif font that's recently been introduced for headings?

Per mw:Typography refresh, my understanding is that Linux Libertine, Georgia and Times (in that order) are specified. Linux Libertine is relatively uncommon, so most users see Georgia. —David Levy 12:38, 23 April 2014 (UTC)

I don't want to dwell on the shading issue, but I will address this one more time. I first wrote: "I arbitrarily chose 140 since the comparison between all varieties of teal being considered with the unvisited link colour was 141, so it seemed like as good a place as any the draw a line in the sand. The shading is really just a visual aid to help see where the problem areas lie." Then: "The numbers are more telling..." Then: "In absence of an objective standard: it's my view that the optimal position is to find a mid-point between potentially conflicting colours in order to provide as much distance as possible" so the shading marks that mid-point. It's not about which colours "seem good" or "seem bad", but reflecting the points I have already discussed, and I find the colours helpful, not to show "right"/"wrong" or "pass"/"fail" but "better"/"worse"; you disagree, and I respect your opinion.

Thanks for the link—I wasn't sure where to find it. Since the backup (in the absence of Linux Libertine) uses the same font-family as {{xt}}, this doesn't really help on that count. I look forward to seeing if any other (viable) alternatives emerge. Meanwhile, I note at mw:Typography refresh#Why did we change the body text color? that black (#000000) has been replaced with #252525, so I suppose this should be updated in the table, too? —sroc 💬 13:20, 23 April 2014 (UTC)

I first wrote: "I arbitrarily chose 140 since the comparison between all varieties of teal being considered with the unvisited link colour was 141, so it seemed like as good a place as any the draw a line in the sand. The shading is really just a visual aid to help see where the problem areas lie."

Exactly. You based the selection specifically on the teal shades' scores and deemed anything lower "problem areas" (with absolutely no evidence of actual problems beginning at that level).

Then: "The numbers are more telling..."

And I agree. The numbers are objective measurements that speak for themselves.

Then: "In absence of an objective standard: it's my view that the optimal position is to find a mid-point between potentially conflicting colours in order to provide as much distance as possible" so the shading marks that mid-point."

No intrinsic "mid-point" exists. #C87878 has differences from #006400 and #002bb8 of 340 and 341, respectively (which is even more equidistant). The 140/141 parity enters the equation specifically in the context of your preferred color. And that's only when examining a particular link type's color and that of a template that you happen to use (an arbitrary pairing with no special significance).

To be clear, I'm not advocating that we adopt any shade of red (which would cause problems in other areas, obviously). Though this did remind me to add red links to the table.

It's not about which colours "seem good" or "seem bad",

That wasn't a description of what the color-coding is actually supposed to represent; it was a possible explanation of how bias influenced your decision without any dishonest intent involved.

but reflecting the points I have already discussed, and I find the colours helpful, not to show "right"/"wrong" or "pass"/"fail" but "better"/"worse";

And you drew your "line in the sand" (at which you begin distinguishing "better" from "worse") exactly below your preferred color's level (as indicated at the time). So my suggested color receives a pink mark where it's 8.53% worse than your preferred color, but your preferred color doesn't receive a pink mark where it's 70.92% worse than my suggested color.

you disagree, and I respect your opinion.

And I hope it's clear that no disrespect is intended on my end.

Meanwhile, I note at mw:Typography refresh#Why did we change the body text color? that black (#000000) has been replaced with #252525, so I suppose this should be updated in the table, too?

I've expanded the table accordingly. (Note that the typography refresh currently applies strictly to the Vector skin, and there's been talk of overriding it locally.) —David Levy 21:26, 23 April 2014 (UTC)

I'm sure you understand this by now, but for the avoidance of doubt, I didn't choose 140 because it suited my colour; it represented the mid-point out of the colours that were being compared. If we had a cornucopia of oranges and tans and puces then this might have been different, but I was working from what we had. In hindsight, it would have been better to apply a gradient across the whole table to reflect that "better" and "worse" are on a continuum and to avoid the perception of a specific tipping point.

I see that you have updated the tables including some of the spacing of the wiki markup, but I assume none of the figures that were already in the table have changes; if they have, it would be better to mark up the changes with <del> and <ins> (or just create a new table) so that the intervening commentary still makes sense.

I appreciate the respect that you have demonstrated in our discussions. Thanks again for your civil approach – often easy to be misconstrued in text-based conversations. —sroc 💬 23:48, 23 April 2014 (UTC)

I'm sure you understand this by now, but for the avoidance of doubt, I didn't choose 140 because it suited my colour; it represented the mid-point out of the colours that were being compared.

I realize that the 141/140 parity jumped out at you as a desirable trait. But you've acknowledged that you "arbitrarily chose 140 since the comparison between all varieties of teal being considered with the unvisited link colour was 141, so it seemed like as good a place as any the draw a line in the sand." You intentionally drew the "line" directly below the "varieties of teal being considered", thereby assigning a "problem" marker to my suggested color's #006400 difference (which falls short of said "line" by only 11).

But as I noted, I'm not accusing you of deliberately manipulating the formula to mislead or gain an unfair advantage. I assume that you honestly regarded the "140" cut-off as sensible. I think that your perception of the relevant factors was unconsciously influenced by personal bias (which can happen to anyone).

In hindsight, it would have been better to apply a gradient across the whole table to reflect that "better" and "worse" are on a continuum and to avoid the perception of a specific tipping point.

I agree, but such a scale would have applied to the specific candidate colors included in the table at the time (unless recalibrated whenever one with a higher or lower number was added). It also would have been skewed by #008000's "28" difference from #006400 (just as the aforementioned "360%" and "400%" improvement figures would be misleading in isolation).

I see that you have updated the tables including some of the spacing of the wiki markup, but I assume none of the figures that were already in the table have changes;

That's correct.

I appreciate the respect that you have demonstrated in our discussions. Thanks again for your civil approach – often easy to be misconstrued in text-based conversations.

Thank you for acting in kind. —David Levy 00:43, 24 April 2014 (UTC)

Talkquote styling

sroc, {{talkquote}} is way too 'in-your-face' alrady; I suggest removing the border and go with a light teal background and left-border only. OR follow the formatting of this template. — Edokter (talk) — 10:32, 23 April 2014 (UTC)

@Edokter: If this discussion has taught us anything, it's that backgrounds colours in templates are bad ideas that lead to trouble finding text colours that contrast well. We'd be better off toning down the borders and reducing (or removing) the background colour. —sroc 💬 11:35, 23 April 2014 (UTC)

If this discussion has taught us anything, it's that backgrounds colours in templates are bad ideas that lead to trouble finding text colours that contrast well. We'd be better off toning down the borders and reducing (or removing) the background colour.

My point is: keep it simple and minimize frills. — Edokter (talk) — 13:17, 23 April 2014 (UTC)

I like it. I particularly like that the left-side border is similar to quoting formats in some other media (e.g., some email system, Tumblr comments, etc.), so is somewhat intuitive. Importantly, the background is faint and therefore unlikely to trigger issues with foreground colours. —sroc 💬 13:25, 23 April 2014 (UTC)

I dig it too, for the same reasons. — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼ 11:07, 4 May 2014 (UTC)

Shall we implement it then? Also, is there any consensus about {Tq}? (I wish my main page design had this level of discussion... and this is just a template!) — Edokter (talk) — 11:32, 4 May 2014 (UTC)

Yeah, I'm surprised. Most majorly important MOS issues that would affect tens of thousands of articles don't get this much traffic. Weird priorities. I'd like to see this one re-started as a fresh discussion. It's sprawled and fragmented and turned into editwarring. — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼ 02:35, 5 May 2014 (UTC)

To what "editwarring" are you referring? —David Levy 04:07, 5 May 2014 (UTC)

Now that we have the {{tq}} colour sorted and the {{talkquote}} colours have been adjusted accordingly, do we have consensus to remove the borders save for the left-side border as discussed above? —sroc 💬 23:46, 4 May 2014 (UTC)

Edokter implemented the styling. No objections from me. My only concern is that it the demarcation is greatly reduced under MonoBook. The solution, I believe, is to use a white background on talk pages across all skins. I don't know why we didn't make such a change years ago. —David Levy 00:06, 5 May 2014 (UTC)

That was quick! I assume the talk pages are intentionally a different colour from article space to provide a subtle distinction? —sroc 💬 00:16, 5 May 2014 (UTC)

Not in Vector. Everything is white in Vector. — Edokter (talk) — 00:20, 5 May 2014 (UTC)

(edit conflict) That was the idea, but it seems to have been abandoned with the introduction of Vector (with no apparent widespread consequences). As far as I can tell, the custom color was retained in MonoBook simply because no one thought to remove it. —David Levy 00:22, 5 May 2014 (UTC)

I would darken it slightly to have the same ~~luminosity~~ color as the example above. — Edokter (talk) — 00:20, 5 May 2014 (UTC)

Your suggested background color looks fine to me. —David Levy 00:57, 5 May 2014 (UTC)

I've made the change. —David Levy 04:07, 5 May 2014 (UTC)

Mmmm. It is odd that the colour of the {{tq}} text looks different than the {{talkquote}} border colour (at least to me on my system). Thus:

He said "This is what is looks like" when quoting within a block of text.

I wonder whether it would be beneficial to change the colour of the text in {{talkquote}} to match as well? This may address David Levy's concern that the demarcation is otherwise not clear enough, and would also better unify these templates. Thoughts? —sroc 💬 00:33, 5 May 2014 (UTC)

He said "This is what is looks like" when quoting within a block of text.

Note that you would not normally have the need to use {{tq}} within {{talkquote}}; I've done so here for the sake of comparison. —sroc 💬 00:37, 5 May 2014 (UTC)

The border is thicker, that causes the apparent difference. Use a fat font and they look the same. We can make the border thinner. — Edokter (talk) — 00:40, 5 May 2014 (UTC)

Doesn't the text's anti-aliasing (vs. that of a solid block of color) play a role? Perhaps we could adjust the border's shade to match {{tq}}'s output perceptually. —David Levy 00:57, 5 May 2014 (UTC)

Is there a widely available "fat font" we could adopt for both templates (in place of italics)? Just putting it out there. —sroc 💬 01:01, 5 May 2014 (UTC)

The resultant contrast ratio is 4.44:1 (a bit below our target minimum). —David Levy 00:57, 5 May 2014 (UTC)

That could be resolved by adjusting the background colour:

He said "This is what is looks like" when quoting within a block of text.

#008560 on #f2fefc has a luminosity contrast ratio of 4.50:1. The visual cues of the border and the text colour should be more than enough to mark the quoted text despite the reduction in background shading. —sroc 💬 22:52, 9 May 2014 (UTC)

Firstly, I'll note that the background color has changed since I cited the 4.44:1 contrast ratio. (With the current background color, the contrast ratio would be 4.25:1.)

Secondly, yes, a lighter background color would enable a better contrast ratio. But it also would greatly reduce the template's demarcation. Edokter's suggested color (#eaf8f4) has made a tremendous improvement under MonoBook. A switch to #f2fefc would make the situation even worse than it was before (with the too-light background color that I implemented).

Thirdly, I see no significant benefit to the use of matching text. I see a modest benefit to (and no harm in) having the the border color match, and that's sufficient. Colored text looks fine with shorter quotations, but it can become a bit overpowering with longer ones (which, as you noted previously, tend to fall under {{talkquote}}'s usage). Your suggestion blurs the line between the two templates. —David Levy 00:47, 10 May 2014 (UTC)

Have we reached a consensus on this?

Do we leave the {{talkquote}} text colour as black or change it to match the colour of {{tq}}?
Regarding the border colour, do we:
- Leave it as is and accept that the border and text appear to be slightly different colours?
- Lighten the border colour to match the apparent colour of the text, and if so, how do we work out that border colour?
- Make the border thinner so that the colours appear the same?
- Consider a different (thicker) font that will better reflect the border colour in the text (and adopt that for {{tq}} as well)?

—sroc 💬 22:43, 9 May 2014 (UTC)

1. As discussed above, I see no benefit to changing the text's color.

2. I don't think that the border should be thinner. I'm fine with either leaving it alone or attempting to match {{tq}}'s output perceptually (which would simply entail trial and error, I suppose). A different font is worth perusing, but not for reasons related to color. —David Levy 00:47, 10 May 2014 (UTC)

Tq color

Resolved

@Edokter: I'm all for implementing this at {{talkquote}}. I'm not sure whether we've reached a consensus on {{tq}}: I think we've resolved on a colour (David Levy's #008560), but I only agree that this solves the initial problem if the sans-serif italics option is made the default, which has met some opposition (see the below discussion which has fallen silent). In any case, both templates should reflect the same colour (i.e., the border in {{talkquote}} should match the text in {{tq}}). —sroc 💬 13:43, 4 May 2014 (UTC)

Then someone will just have to force this stalemate. Seeing my proposal shows the least ammount of red in the table above, I think I hit the right color (#006464) from the beginning. Teal is a distinguishable color that is not used anywhere else (yet), and as such, can do without any additional default styling. That will be my position. — Edokter (talk) — 14:52, 4 May 2014 (UTC)

As discussed previously, if {{tq}}'s output is difficult to distinguish from black/near-black text, the template's purpose is defeated. I realize that this doesn't bother you (because you oppose the template's very existence and don't mind if it ceases to function as intended), but others feel differently. —David Levy 15:10, 4 May 2014 (UTC)

I opposed no such thing; I suggested using Italics only. Is it really worth having megabytes of discussion over such a trivial issue? This is starting to look like a bikeshed discussion. I do note my color scores best in all these differential tests, while maintaining legibility on all backgrounds. Now, can someone, anyone, retract their heel from the sand, so we can move forward? — Edokter (talk) — 15:33, 4 May 2014 (UTC)

My apologies for apparently taking your " ''just go like this'' " suggestion too literally. My statement's crux is unaffected; you see no need for the output's color to stand out from black/near-black.

You regard the matter as "trivial" because you don't personally see a problem. Others do see a problem (specifically, one impacting talk page messages' readability).

On what do you base the claim that your color "scores best in all these differential tests"?

You're entitled to your opinion, but you're pleading for others to "retract their heel[s] from the sand" while simultaneously reviving a matter of contention that appeared to have been resolved. —David Levy 16:09, 4 May 2014 (UTC)

Apparently, it has not been resolved, otherwise changes would have been made. So what is it? I'm not trivializing the entire issue, but I do trivialize the entire discussion on the exact color. If there is no consensus on that particular item, then I will have to force a dicision. So... is there a consensus? — Edokter (talk) — 16:15, 4 May 2014 (UTC)

As Sroc noted above, there appeared to have been agreement regarding the color, but you've returned to the discussion and resumed debating that point (while simultaneously deeming it "trivial" and criticising others for focusing on it). —David Levy 16:32, 4 May 2014 (UTC)

Then I have missed something. Please state what the consensus is, because I'm a bit lost here. — Edokter (talk) — 16:58, 4 May 2014 (UTC)

As Sroc explained, there evidently isn't a consensus regarding the overall issue. We settled on a color (#008560), but not a default text style. Sroc wants to reverse the current setup, making italic sans-serif text the default and non-italic serif text a parameter-based option. RexxS opposes this change on the basis that the ability to include nested italic text should take precedence. I'm neutral (and favor pursuing some alternative that addresses everyone's concerns).

I think that it would be okay to proceed with the color change and continue discussing the default text style. Sroc doesn't regard this as a sufficient solution on its own, but there seems to be agreement that it's at least a step in the right direction. (Sroc can correct me if I'm mistaken.) —David Levy 17:27, 4 May 2014 (UTC)

Personally, I think #008560 is the worst possible choice; too close to green and fails in archives, but since you two have agreed on it, I'm not going to oppose it. If it is only the style that is an issue, I choose for the simplest method; no styling (color only). I would like to reserve serif for example text. — Edokter (talk) — 17:57, 4 May 2014 (UTC)

Yes, I agree that #008560 would be a step in the right direction, although not enough overall. Similarly, Edokter's #006464 combined with sans-serif italics would help to differentiate from {{xt}}, but a stronger colour difference would be preferable in cases where the serif option is invoked. I'm tired of debating over colours though and will go wherever the consensus falls. —sroc 💬 20:02, 4 May 2014 (UTC)

That doesn't look like much of a consensus to me... it looks like David has just "out-debated" you. I cannot give my support in this case, so it seems #008560 is out. David Levy, I'm sorry... I should have been more involved in this discussion, but I hate long discussions. That's why I bugged out. — Edokter (talk) — 20:51, 4 May 2014 (UTC)

I was just about to perform the change. (I even worked out a corresponding background color for {{talkquote}}.) Three hours ago, you stated that you were "not going to oppose" the color's use, and now you've vetoed it outright. —David Levy 20:57, 4 May 2014 (UTC)

I have stated my opposition to #008560 very early in the debate, so did sroc. And I hate having to repeat myself, which is why I avoid very long discussions. I was under the impression you two had reached an agreement, but after sroc's comment just now, that is obviously not the case. What would you have me do? Shouldn't't consensus be about a level of agreement instead of testing who has the longest breath? Do you really believe this constitutes consensus? — Edokter (talk) — 21:17, 4 May 2014 (UTC)

I believe that #008560 is a color that Sroc and I agreed to use (and I don't know why Sroc's new comment has led you to doubt this).

Is it Sroc's preferred option? No. And in case you didn't realize this (though I stated it the above discussion), it isn't mine either. My preference is to retain the current shade of green. I suggested #008560 (which took a great deal of trial and error to formulate) as a compromise. It would have been much easier for me to simply oppose changing the color, but I made a sincere effort to address Sroc's concerns (and rather than meeting halfway, I picked a shade closer to Sroc's preference than to my own).

As noted above, you certainly are entitled to your opinion. And had you simply expressed it (even after withdrawing from the discussion for a while), I wouldn't have objected to that.

Instead, you deemed the color choice a "trivial issue", opined that "this is starting to look like a bikeshed discussion", and pleaded for "someone, anyone [to] retract their heel from the sand". But you also argued that we should use your preferred color.

When you realized that Sroc and I had settled on a different color, you overruled the decision on the basis that Sroc had expressed reservations and agreed to compromise for the sake of resolving the matter — which is exactly what you urged us to do when you believed that your proposed color remained in contention. Suddenly, this isn't a "trivial" issue whose discussion amounts to bikeshedding. —David Levy 22:05, 4 May 2014 (UTC)

If one party states "I'm tired of debating...", it is a red flag; that is not consensus, that is throwing in the towel. I have a lot of trouble accepting such outcomes because it elimintates any desire to engage in future discussions with the 'winner'. That's why this doesn't sit right by me. However, it is somewhat of an improvement, so go ahead with the color.

Can you both state in short you preference for the styling? — Edokter (talk) — 22:28, 4 May 2014 (UTC)

I certainly hope that Sroc doesn't view the discussion in that light. My goal was never to wear down others so I could "win". The differences in opinion were obvious from the start, so I set out to compromise. But I'm only one person, and it was entirely possible for ten people to show up and agree with Sroc unanimously (in which case no compromise would have been called for). But it eventually became clear that participation in the discussion was going to be extremely limited (which disappointed me too) and couldn't possibly result in consensus for the status quo or a switch to teal. For the same reason, I'm "tired of debating" too. A compromise is the only way forward, and I don't think that there's anything inherently wrong with that outcome. A clear consensus would be preferable, but we need to take what we can get. Thanks for giving the go-ahead.

Regarding the default styling, I believe that both Sroc and RexxS have made valid points. I have no strong preference for or against the proposed change. Ideally, we can come up with a third option that satisfies all involved. —David Levy 23:09, 4 May 2014 (UTC)

(edit conflict) Ugh! After extensive and vigorous discussion, David Levy and I agreed on #008560. It wasn't either party's first preference, but we weighed the evidence and reached a compromise. That's what establishing consensus is all about: not sticking to one view and defending it to the death, but discussing the issues to reach an agreeable solution.

Now, having been away from the discussion for a while, I've realised that the precise shade is perhaps not as significant as a font format that is distinct from the serif font-family (Georgia, et al) used by the {{xt}} template. My preference remains #008560, but if you persuaded David to accept #006464 then I would go along with this. In either case, I think some other font formatting is needed, and my money's on sans-serif italics as this is the optional format already in use that people will be most familiar with, but if the tide of opinion favour some other formatting, so be it; the important thing to my mind is to move away from the current font-family, at least as the default.

To be clear: I am all in favour of changing the font colour of {{tq}} and the border colour of {{talkquote}} to #008560. —sroc 💬 23:23, 4 May 2014 (UTC)

Background colours on templates

On a related note, I have started a discussion at Wikipedia talk:WikiProject Accessibility#Background colours on templates. —sroc 💬 06:50, 21 April 2014 (UTC)

Font style

I've just seen this discussion having followed the link above. I'm quite happy with any of the colour choices in the tables because I don't believe we should be picking our foreground colours to pander to every choice of background colour that may be dreamed up. The range of foreground colours used should determine the possible backgrounds, not the other way around.

The one point I disagree with strongly is the suggestion of standardising {{tq}} with an italic font. I use this template frequently and want to be able to preserve the semantic formatting of the quoted text as far as possible. Neither the colour nor the font-family convey semantic information in the original text, so may be safely used to delineate a quotation. But italics convey emphasis by convention both on Wikipedia and on the web in general, so should not be used to further delineate a quote. You have two unused attributes to play with in an attempt to have the quote stand out from the surrounding text; taking up font-style as well is overkill and renders emphasised text inside a quotation either in the same style as the rest of the quote as or contrarily rendered as non-italic which conventionally conveys the meaning of unemphasised. --RexxS (talk) 12:49, 21 April 2014 (UTC)

Yes, I mentioned this issue above. I generally prefer to italicize text in this context, but I've used the template's default styling when the preservation of italic text was needed.

The default behavior certainly shouldn't be changed without consensus. —David Levy 14:19, 21 April 2014 (UTC)

This is quite a conundrum, since distinguishing between {{tq}} and {{xt}} is aided by the combination of both colour and font style. Compare (using #007878):

Italics: He said "I didn't propose I didn't do it! but I'll go with it."
Serif: He said "I didn't propose I didn't do it! but I'll go with it."

The italics version is much more easily distinguished from the example text; the serif version is barely distinguishable, in my view (and perhaps only noticeable here because we are looking for it, but not obvious enough when scanning the text unknowingly). Of course, the serif option should still be available where desired, but I think that the italics version should be the default to avoid confusion with example text, as discussed above. Another alternative might be to adopt another different font, but this might just be overkill with so many different fonts in use. —sroc 💬 15:13, 21 April 2014 (UTC)

The more attributes you use up in demarking a quote, the fewer you have left to convey meaning within the quoted text. By your logic we would better off using a 1.2em bold italic serif font in a conspicuous colour with a text shadow (ugh!), but the trick is to only use sufficient different attributes to do the job - and preferably not use ones that are already in use (italics for emphasis in this case). I have never found myself having to distinguish between {{xt}} and {{tq}}, but I've needed to include italics in quoted text quite often. Look at what happens to an inline quote with default italic (no.2):

You said "Another alternative might be to adopt another different font ..." and I agreed.
You said "Another alternative might be to adopt another different font ..." and I agreed.

The semantic markup that you applied has gone missing. The default should be the easiest system - "out-of-the-tin" if you like. Editors shouldn't have to apply options to get emphasised text to remain emphasised when they quote it. --RexxS (talk) 16:43, 21 April 2014 (UTC)

Please do not misrepresent my position. I am not arguing for more attributes (much less the garish ones you suggested!) but to replace an existing one (font) with another (italics or a different font). I didn't realise that the template would obscure italics altogether, but this seems to be a flaw in how the italics is implemented by the template. The usual practice is that italics within italics are interpreted as plain text, and this would be effective if the template used '' marks instead:

You said "Another alternative might be to adopt another different font ...

Unfortunately, because the template uses font style rather than markup to implement italics, this doesn't work – but would be easily fixed. As I said, however, there is another option to choose a different font instead:

You said "Another alternative might be to adopt another different font ...

Again, I'm not advocating this (not this particular font), merely pointing out that there are options. —sroc 💬 00:04, 22 April 2014 (UTC)

I don't see what I'm misrepresenting. I don't know if I can spell out this any more bluntly without seeming rude, but you seem to want to use three attributes to distinguish text that is quoted: color, font-family and font-style. We presently only use two: color and font-family. You haven't justified why you need three, so I asked why not four or five if your only reason is "more is better"? You also seem to want to pick an extra attribute that is already in common use in text to convey meaning (emphasis). That either removes or diminishes the ability to quote text that contains emphasis, and contrary to your assertion of "usual practice", I contend that italics within italics is not plain text, nor should plain text within italics convey emphasis. Check the result of "Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod" if you don't accept my word. You're mistaking our experience of the way that nested raw wiki-markup works with the way that nested html or css works; they are not the same. So, have I really misrepresented your position? --RexxS (talk) 00:47, 22 April 2014 (UTC)

What do you not understand about the word replace? I am not advocating three attributes at all; to say "you seem to want to use three attributes" misrepresents my position, and to extrapolate "four or five" even moreso; nor have I ever suggested "more is better".

Let me be clear. We currently have:

Example — two attributes: font-family and colour

I am proposing to replace the font selection (which is used by another template with a similar colour) with italics by default:

Example — two attributes: font-style and colour

As an alternative, I have suggested replacing the font (using Courier as a crude example):

Example — two attributes: font-family and colour

How do you get to three? —sroc 💬 01:14, 22 April 2014 (UTC)

I can't say that I've ever encountered the inverted formatting that you've described, apart from some accidental use that arose when {{gi}} (now a wrapper for {{tq}} contained the type of wiki markup that you advocate (the replacement of which was considered a bug fix). —David Levy 01:23, 22 April 2014 (UTC)

See Italics#Italics within italics:

If something within a run of italics needs to be italicized itself, the type is normally switched back to non-italicized (roman) type: "I think The Scarlet Letter had a chapter about that, thought Mary". In this example, the title ("The Scarlet Letter") is within an italicized thought process and therefore this title is non-italicized.

It may not be the way HTML handles it, but that's how it works in the real world. 01:28, 22 April 2014 (UTC)

As I said, I've never encountered that "in the real world". Now I've seen it described in an unsourced Wikipedia paragraph. —David Levy 01:38, 22 April 2014 (UTC)

I have seen it in the wild and it was the way I was taught. The article refers to this rule being in The MLA Style Manual. —sroc 💬 02:08, 22 April 2014 (UTC)

According to that unsourced paragraph, The MLA Style Manual includes a recommendation to apply the format to the citation of a book whose title contains another book's title. (No mention of other contexts is made.) —David Levy 03:07, 22 April 2014 (UTC)

What I don't understand is how you can use the word replace when proposing:

The colour of the {{tq}} template be changed to be distinct from that of the {{xt}} and {{!xt}} templates and link colours;
The sans-serif italic font become the default of the {{tq}} template, with serif font as an option.

because I count three attributes there: colour, sans-serif font (with serif font as an option), italics. How many do you count?

There is no convention in web publishing that italics within italics should render as roman. It would break semantic markup conventions. Here's the html your example delivers: "I think The Scarlet Letter had a chapter about that, thought Mary". There are two phrases that are marked up as italic (emphasised) and the title is not emphasised, which is semantically incorrect. On Wikipedia, we use italics for emphasis and for certain names and titles, loan words, and scientific names (see MOS:ITALIC); but not for quotes (see MOS:QUOTE). It is absurd to throw away the ability to keep semantic markup (like ...) when quoting, in order to use the same markup for a purpose for which it is specifically discouraged, "It is normally incorrect to put quotations in italics. They should only be used if the material would otherwise call for italics, such as for emphasis or to indicate use of non-English words." --RexxS (talk) 15:51, 22 April 2014 (UTC)

Again, you seem to misunderstand me. What I propose is:

Default behaviour: Example (italics, colour)
Alternative option: Example (serif font, colour)

This is simply switching the current default (serif font) with the option (italics). Whichever option is selected, exactly two attributes are applied. (To be clear, I never proposed applying serif and italics in this template, if that's what you thought.)

You say that using plain formatting for italics within italics is "semantically incorrect"—on what basis? As I indicated, this is how it is traditionally handled, even if HTML markup wasn't built that way, but wiki markup (using '') copes with it just fine. Semantically the meaning conveyed by italics is preserved in much the same way as nested quotation marks, as illustrated by the example in the article I linked to. —sroc 💬 16:10, 22 April 2014 (UTC)

I believe that RexxS was referring to HTML's semantic structure. Compliance with the established standards enables accurate interpretation by software, including screen readers used by blind and visually impaired people. (See also: Semantic HTML) —David Levy 20:24, 22 April 2014 (UTC)

It's not just me, by the way. For example, see:

DailyWritingTip.com—Italics Within Italics:
If you’re already writing in italics and you need to use italics, you COULD use boldface or underlining, but the generally accepted practice seems to be to UNitalicize. Let’s say I was writing to you about a great book about the making of the film Casablanca, my favorite movie. I would type it like this: I’m reading this great book called The Making of Casablanca: Everyone’s Favorite Movie. Notice that everything’s in italics except the name of the film
TERMIUM Plus—Italics:
Because italic (sloping) type contrasts with roman (vertical) type, a writer can require words or passages to be typeset in italics in order to call special attention to them, to give them special meaning, or to distinguish them from the rest of his or her text. [...]
Note that when the main body of a text is printed in italics, roman type is used for emphasis and for the other purposes described in this chapter.
Ask the English Teacher—Italics (and Parentheses):
By the way, if you must emphasize a word or phrase within a long passage of italics, just use roman text.
National Novel Writing Month—Issues with italics:
By convention, if something is in italics and something within it would otherwise be italicized, then it's put in normal type.

—sroc 💬 16:31, 22 April 2014 (UTC)

Change the color to blue to make the two types of talk quotes match. This will aid in making it clear what is a neutral quote and what isn't. Note that I haven't read the entire discussion, this is based entirely on the original request in this section. Feel free to discount my opinion if there are good reasons to separate the two or a good reason to move one or both to a different color than blue. MjolnirPants Tell me all about it. 18:40, 9 May 2014 (UTC)

Blue is a bad colour choice because it clashes with the link colour. There is a massive discussion above comparing different colours and precise shades, resulting in a choice of a greenish-teal. —sroc 💬 22:30, 9 May 2014 (UTC)

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Remove all styling

I never really understood why user quotes need to be so 'fancy' anyway. This is highly confusing with {{xt}}, and quotes are not examples. How about going back to basics here and "just go like this"? — Edokter (talk) — 10:49, 17 April 2014 (UTC)

This template assists in the comprehension of talk page messages, which goes beyond the ability to distinguish text from its background. —David Levy 23:04, 19 April 2014 (UTC)

I agree with David. Editors are free to quote others without using the template at all and just use bare "quotation marks" or their own "styling", but the template's formatting can be useful when desired (for example, see the above exchange where David and I quote lines from each other's posts in our replies). Edokter is right, though, that it needs to be distinct from {{xt}} to avoid confusion. —sroc 💬 02:54, 20 April 2014 (UTC)

If we do that, editors will resort to using a variety of inline styles, which will be harder to override locally. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:27, 20 April 2014 (UTC)

What Andy said. This is useful. I don't use it anywhere near as much as some people do, but still rely on it. — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼ 04:29, 23 April 2014 (UTC)

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

[1] s a compromise approximating the contrast levels of the current green

[2] revised compromise to reach AA level on #edeaff background

[Note 1]

[Note 2]