Fraser deliberately left out the properties of the core in his list, but one obvious difference that could differentiate cores would be the cell fill, the portion of the unit cell actually comprising the bolometer. I have seen core designs with far less than half of the pixel actually contributing to the image, and others with well over a half. The higher the percentage the harder to do, so an obvious place to cut corners for cost reasons.
If I were to prioritize the ideas presented here I would rate lens quality and focus at the top, and too close to prioritize one over the other, followed by raw core performance, contamination (Fraser's number 6 expanded to include finger oils on the lens and other such antics) and LCD module quality.