I think the article misses the point: It's not about how complex the data structures are, it's about the result, in all its details.
Comparing different RAW converters (Lightroom, DXO), their image rendering is slightly different. If you compare the colors with the JPEG image, even more so. If the goal is to faithfully reproduce the colors as they were shown in the camera, you depend on the manufacturer's knowledge. To me, it makes not sense to have some "open" DNG format in the middle, when it's flanked by proprietary processing.
It's not about the format, it's about knowing the details, including parameters, of the image processing pipeline to get a certain look.