I'm not convinced the 2012 W3C symposium paper is interpreted properly. Here's a quote from a 2015 paper by the same authors (Rello & Baeza-Yates):
This paper presents the following
[stuff about font size etc., then]
– Black text on white background instead of using
grey scales for the text is significantly preferred by
people with and without dyslexia.
– White text on black background instead of using
grey scales for the background is significantly preferred
by people with and without dyslexia
So it looks to me it's the maximization of contrast that matters. I'm not sure about the black on white vs white on black (right now).
Also from that paper I gather there's not much research on the topic of color schemes for dyslexics (there's much more on font sizes and spacing). The only color-related research from a different group cited in there was Kurniawan and Conroy (2007). And here's what they found (p. 268) studying a group of 5 dyslexics (and a number of controls):
When asked about their most effective colour scheme, each student with dyslexia
provided a slightly different answer. Some said that they did not have any problem
with black and white as long as the paragraphs were short and the page had plenty
of white spaces. Some preferred “any dark coloured text” on baby pink background,
dark blue on cream, and black on “any light coloured background.” Not surprisingly,
all the students without dyslexia preferred black and white. The students without
dyslexia who were given the passages in dyslexia-friendly colours stated that they
felt very dizzy reading the passages on coloured screen, and they had to use their
fingers to follow the sentences because of that. Some suggested that this might be the
main cause that their reading speeds were lower than those of the other groups.
Full ref for the latter paper: Kurniawan, S., Conroy, G.: Comparing comprehension
speed and accuracy of online information in students
with and without dyslexia. Advances in Universal Web
Design and Evaluation: Research, Trends and Opportunities,
Idea Group Publishing, Hershey, PA pp. 257–70
And I think the recommendation for colored presentation is based on a particular cause of dyslexia (p. 262 in Kurniawan and Conroy):
Tinted lenses and coloured overlays. Some people with dyslexia suffer from
a condition called Meares-Irlen syndrome. Text or symbols become blurred
or indistinguishable soon after reading begins. This is particularly the case
when reading information presented in high contrast, such as black text on
a white background. The condition varies in its severity, but in most cases it
can be improved with the use of simple coloured overlays (placed over pages
or computer screen) or tinted lenses.
I'm not sure how many dyslexics suffer from Meares-Irlen syndrome to make the recommendation universal... Also, perusing the Wikipedia page on the latter, I gather it's "very rare" and colored overlays might not actually work to alleviate it; the latter finding is from a 2011 paper (Ritchie et al.) in a medical journal. And 2014 review mentions that paper and other recent negative findings:
Some recent studies failed to find statistically significant effects of colored overlays. Ritchie et al. (2011) had shown that, in the short period, colored overlays do not speed reading up compared to non-colored overlays, whether or not the participants have a diagnosis of visual stress. Ritchie et al. (2012) had shown that – compared to a control condition – not even one year of use of colored overlays results in an increase in reading speed and accuracy. Henderson et al. (2013) had shown that despite the fact that often dyslexic individuals do experience stronger visual stress than controls, neither dyslexics nor controls benefit from the use of colored overlays.
However its conclusions are a bit more nuanced:
Whether colored overlays help reading or not seems at least controversial: although initial evidence was indeed provided, more recent studies both highlight the methodological issue of previous studies and show that colored overlays do not help reading (Ritchie et al., 2011, Ritchie et al., 2012; Henderson et al., 2013), On the ground of contradictory findings as these, the [American Academy of Pediatrics (AAP), 2009] has claimed that there is not empirical evidence toward the efficacy of colored overlays in reading, reading acquisition, or dyslexia, and did not recommend their use.
However, the participants in the studies of Ritchie et al. (2011); Ritchie et al. (2012) were non-dyslexic children, and in the study of Henderson et al. (2013) they were adults, whereas it has been shown that effects of colored overlays are more easily found with dyslexic children (Singleton and Trotter, 2005; Singleton and Henderson, 2007). Whether or not, at least in some conditions, colored overlay works does not seem to be a settled issue. Thus, although from one side, given these contradictory findings, a precautionary, prudent position – as that of the Academy of Pediatrics – on the use of colored overlay seems desirable, especially in clinical or educative contexts, from another side, given that some evidence that the colored overlays work exists, concluding that colored overlays proved not worth in allaying reading problems is premature and, possibly, incorrect.
The lack of efficacy of color overlays is also mentioned in a summary of the book "The Dyslexia Debate", p. 4. That book has a lengthy review in one of main neuroscience journals, namely in BRAIN. While the review does take issue with a number of tenets of the book, it doesn't challenge much the book's take on treatments, so I conclude that the colored overlays not working much is a fairly valid conclusion, although possibly encumbered by diagnostic difficulties with dyslexia subtypes. At the very least, the recommendation to use colored overlays for all dyslexics seems unsupported.
Going back to Rello and Baeza-Yates (2015), they mention that some Firefox extension that allows extensive customization:
Santana et al.  developed the Mozilla Firefox
extension Firexia, a tool that allows readers with
dyslexia to customize websites to improve readability.
They tested Firexia with four users and found that
readers with dyslexia appreciate customization. The
customization settings included in Firexia are based
in previous user studies and recommendations. They
include font type, font size, color, character spacing,
line spacing and column width.
So probably the best technological recommendation (until the science of dyslexia [and its treatments] gets sorted out) is to do something like that.