var intro = 'The data being plotted is Passing Yards vs Passing Attempts.';

var problem0 = '1. Color the marks by if the QB\'s school is in the Big Five, \
                Group of Five, or Independent. -- no restrictions';

var problem1 = '2. Color the marks by if the QB\'s school is in the Big Five, \
                Group of Five, or Independent. -- color-blind safe';

var problem2 = '3. Color the marks by if the QB\'s school is in the Big Five, \
                Group of Five, or Independent. -- photocopy safe';

var problem3 = '4. Plot only QBs from schools in the Big Five conferences. \
                Color the marks by conference. -- no restrictions';

var problem4 = '5. Plot only the top 5 QBs by rank and color the marks by \
                rank. -- choose a multi-hue colormap';

var problem5 = '6. Plot only the top 5 QBs by rank and color the marks by \
                rank. -- choose a single-hue colormap';

var explanation0 = 'Distinctive \
                    hues are appropriate for categorical data, without \
                    implicit ordering. Since this is a small region, \
                    bright, highly saturated colors are used to ensure that \
                    the color coding is distinguishable. The hue channel has \
                    about 6 or 7 discriminable steps. Our categorical data has \
                    3 distinct categories.';

var explanation1 = 'The colors chosen here avoid using only the hue channel \
                    to encode information: the color scheme generated by \
                    ColorBrewer varies in saturation and luminance. \
                    Using alternate channels to encode information produces \
                    color-blind safe visualizations.';

var explanation2 = 'To withstand black and white photocopy we must effectively \
                    manipulate the luminance channel with the colors chosen. \
                    The color scheme chosen achieves this by preserving \
                    differences in lightness between the colors selected.';

var explanation3 = 'Again, we pick distinctive hues for categorical data. \
                    There are only 5 conferences in the Big Five, so we\'re \
                    still good with the 6 to 7 discriminable steps from the \
                    hue channel.';

var explanation4 = 'The color scheme chosen here is appropriate for the \
                    reasons explained in problems 1 and 4. We\'re only \
                    plotting the top 5 ranks, so we\'re within the 6 or 7 \
                    discriminable step limit of the hue channel. The catch is \
                    that hue\'s not particularly suited to encoding order. \
                    A distinct color is associated to each rank, but to \
                    discover which colors are associated to higher ranks \
                    and which to lower ranks, you\'ll have to hover your \
                    mouse over the marks or rely on positioning in the \
                    scatterplot.';

var explanation5 = 'By using only a single color hue, we can rely on the \
                    luminance and saturation channels to encode rank. \
                    These channels are suited for encoding ordered data. \
                    The luminance channel (the one used in this plot) \
                    has 5 or less discriminable steps, \
                    which is enough to encode our top 5 ranks. Darker colors \
                    have been associated with higher ranks (Marcus Mariota), \
                    and lighter colors have been associated with lower ranks \
                    (Jake Waters).';

var extraQuestion = 'Can you do a color-blind safe or photocopy safe version? \
                     why or why not?';

var extraAnswer = 'If the data being encoded were sequential, we might be able \
                   to get away with making only a color-blind safe version of \
                   the plot by encoding via the luminance channel, which can \
                   have up to 5 discriminable steps. But, because the data \
                   being encoded is qualitative we do not obtain enough \
                   discriminable steps from either the luminance \
                   or saturation channels to encode our 5 categories. \
                   Encoding via saturation is not possible because this \
                   channel has at most 3 discriminable steps. \
                   In any case, photocopy versions cannot be made of this \
                   plot. The book suggests avoiding grayscale (luminance) \
                   if more than two to four bins are required. The luminance \
                   channel must be preserved for photocopy safe vis design.';