Color blindness accessibility of document figures & tables

[sadielbartholomew]

The colours used to convey, or help with the absorption of, information for some figures and tables in the main Conventions document are not sufficiently accessible to those with some facet of colour blindness, as evidenced below. In general, all should ideally be checked for accessibility in this respect and then improved if not up to scratch.

I claim lack of accessibility here because, whilst thinking about the presentation of the document in line with the theme of a recent hackathons at the CF Workshop 2022, I thought to run relevant parts (namely, those with figures and tables utilising colour) through some basic checks made possible from free online tools, notably a very useful 'Colorblind Web Page Filter' tool as linked in my detailed report below.

From such a basic check I observed that most of the figures were good / good enough to be comprehended properly in any case of colour blindness, but there were some that in some cases used colours to highlight key information where certain colours weren't distinguishable, as outlined below. The main culprits seem to be figures I2,to I4 (inclusive) which use a green and a blue to categorise items which can't be distinguished in a few cases.

I would add that I see from the Issue Trackers and discussion that colour blindness has been considered for the creation of diagrams in certain cases, e.g. #193 (comment), which is great! Can we, as a separate but related issue, add something to the contributing guidelines perhaps, or increase awareness for moderators, to ensure that it is a criterion for any new diagrams added to the document that they pass as good enough (to some agreed standard) with respect to (colour blindness, but perhaps more general) accessibility? Maybe we can use the tool above, or similar one(s), as a means to validate new diagrams or adjustments to existing ones?

Report of initial checks on figures and relevant tables

All screenshots shown below are taken from the 'Colorblind Web Page Filter' tool on the linked webpage where the named URL has been submitted.

As mentioned above, most figures were fine in all respects, for example this indicates that Figure 7.1 and 7.2 are OK regarding a certain form of CB:

Example of a figure that is fine w.r.t. C.B. ...

Figures 7.1 and 7.2 (fine, as an illustration)

and scrolling through the options, everything is discernible in any given form listed.

...and ones to highlight for potential C.B. issues

But the following figures looked potentially problematic. I'll start with what I think is the most glaring issues, moving towards those which may or may not be passable (to be discussed!)...

The green, blue and yellow colours used to categorise aspects of the UML in the following, namely Figures I1 to I4 inclusive, can under certain C.B. facets be indistinguishable:

where I also saw that other figures had aspects that might not be comprehensible to the extent we might want, due to cetain colours becoming very similar when certain types of colour blindness are assumed:

[erget]

@sadielbartholomew I think this is a really nice initiative. I myself am colourblind and mostly resign myself to not being able to interpret aspects of diagrams out of laziness - I don't like always making noise about it. It's not a huge barrier, but it often makes me a little sad when I can't grok diagrams as fast/well as other people can. What I mean is this is appreciated ideologically and personally :)

So if I understand your proposal it's 2-fold:

1. Update figures to improve accessibility
2. Update CONTRIBUTING.md to ask people to check figures if they use colours, referring to some common tools that may be useful

Is that right?

[sadielbartholomew]

Hi @erget, thanks for your comment and the speed of it! Firstly, I am sorry to hear you have colour blindness yourself, though it serves to motivate why we should ensure the document is as friendly as possible in this respect. Indeed I saw a figure that ~5% of the (UK, as an example large-sample) population is colour blind in some way, so that's a decent proportion of potential readers.

So if I understand your proposal it's 2-fold:

1. Update figures to improve accessibility
2. Update CONTRIBUTING.md to ask people to check figures if they use colours, referring to some common tools that may be useful

Indeed, nicely summarised :)

As for, (1), I would emphasise that only a small number of figures need updating in my opinion, though it depends what others think in light of the issue and the tools suggested (or indeed others people can find)!

As for the ones I have highlighted above to raise as being potentially insufficient in this respect, let me list those here for ease of processing:

• Figures I1 to I4 (from Appendix I):
  • only need a switch from one or two colours used as box backgrounds to another more distinguishable colour in each case, should be quite simple. @davidhassell I suspect you have ownership on most of these, if not all: would it be simple enough to update them in this respect?
• Figures 8.4 (from Section 8.3.9. Interpolation of Cell Boundaries);
• Tables 9.2 to 9.4 from Section 9 (see 9.3.1. Orthogonal multidimensional array representation forward);
• Figures J5 and the corresponding J3-J4 which introduce those colours (from Appendix J, Interpolation methods).

Further thoughts very welcome from anyone! Just pop them on this thread.