For all of those who think that a rag and solvent won't clean off all the compound, listen up! Metal is NOT porous enough to absorb ANY of the thermal compound into the matrix. The only metals that can absorb are Powder Metallurgy metals, which are sintered powder particles that leave pores inside the metal matrix. This is the process that is used to make oil impregnated parts. Heatsinks and cores are NOT powder metallurgy parts; they cannot absorb thermal compound, or the oil in it. The only compound left will be residue in the *macroscopic* valleys left when sanding. The valleys will be only as deep as the finest sandpaper you used. They are regularly spaced and of even depth (NOT LIKE THE ARTICLE PICTURE). If you just scrape off the old compound, then you will have the old stuff still in the valleys. But when using a solvent, the solvent actively dissolves the compound, even the compound in the valleys. The rag does not scrape the compound out of the valleys, it absorbes it out, carrying all of the dissolved thermal compound with it. The referenced article is for a semi rigid object being used to *scrape* the compound out of the heatsink, not for a solvent situation!
Furthermore, the valleys are so small and shallow, I would guess that 90% of the old compound can be wiped off due to the flexibility fo the rag fibers and their small size WRT the valley size.
Again, the referenced article does not apply to 1) lapped metal 2) rags and solvent being used. I don't see how you can use it to invalidate the rag+solvent method of removing thermal compound. Its comparing apples and oranges.