Obviously they didn't apply it here fully, but it would have been possible for them to reuse their text divs on text-only browsers, drop their svgs for some prerendered raster graphics on SVG-less browsers, use their SVGs but have them render flat on 3d-transform-less browsers, have static 3d copies of each page if JS isn't available but 3d was, etc., etc.
It's even possible for established sites to collect statistics on combinations of available browser features, estimate ROI vs. development effort and decide which combinations beyond the least-common-denominator are worth coding for and slowly roll them out one at a time.
Obviously they didn't apply it here fully, but it would have been possible for them to reuse their text divs on text-only browsers, drop their svgs for some prerendered raster graphics on SVG-less browsers, use their SVGs but have them render flat on 3d-transform-less browsers, have static 3d copies of each page if JS isn't available but 3d was, etc., etc.
It's even possible for established sites to collect statistics on combinations of available browser features, estimate ROI vs. development effort and decide which combinations beyond the least-common-denominator are worth coding for and slowly roll them out one at a time.