You could make a mix of values. If metl melting temp is 1000 and tung melting temp is 3000, so alloy of metl and tung melting temp would be ~ 2000. And you could use that not only for metals, not only for melting temp, and not only for 2 elements in aloy. For example boyl+boyl+metl= conductive gas, metl+metl+boyl= solid wich makes pressure when heated.
take knife
take bread
slice butter with knife and pull some onto the knife
rub knife's buttery side on bread
smooth out butter distribution
take knife
take bread
find jam jar location
retrieve jam jar
put jam jar onto counter
open jam jar lid
lift some jam out via knife
rub knife's jammy side on bread
smooth out jam distribution
I think this idea may be able to work, but not in varying proportions. For example, I can see you doing something like what happens when one salt and one water particle touch, turning both into salt water particles. You could let that occur with molten metals and other similar things to form 1:1 alloys, and could probably be extended to 1:2, 1:3, etc. up to 1:8 easily. It may be more difficult at higher than 1:8 because it would have to detect 2 pixels away, which would cause problems, but I think alloys at a 1:1-8 ratio are a fun idea that may be worth trying! The alloys themselves would be different elements instead of just tmp variations of existing ones. As for concern over breaking saves involving "element separators", well in real life you don't mix two alloyable metals in liquid form and expect them to stay separate. If they want to separate the elements, they can either stick with powdered metals for the ones that have a powder form, or we could have a simple "centrifuge" or some other similar element whose only purpose is to break down alloys into their components. For example, a centrifuge particle, upon absorbing 2 particles of 1:1 METL/IRON alloy, would excrete a particle of molten iron and a particle of molten metal. Now obviously this may cause some back-ups with the higher alloys, such as the 1:8 type alloys, which would require absorbing 9 particles of alloy before it's able to centrifugate. That's why I doubt my own theory as to the more dilute alloys; I honestly think that 1:3 is the highest practical limit that wouldn't overload centrifuge elements while still allowing a reasonable selection of alloys to produce.
Another issue with this idea some people might have is that you can't have multiple alloy mixtures, such as having both a 1:1 and 1:2 alloy for METL and IRON. My method wouldn't be able to tell which one to form after all. My solution for this is simply don't have multiple concentrations for the same alloy, just have one single concentration that forms upon mixing. This is handled similarly in Minecraft's Tinkers' Construct mod, where any alloy that can form in a furnace full of miscellaneous materials will form, limited by the fact that no two alloys have the same ingredient combinations.
If you want to eliminate the hassle of having to centrifuge alloys, perhaps there would also be a mixer element of some sort that did the opposite: take 2 or more particles of material into itself and excrete the alloy. This would be best for not breaking saves.
Overall I think alloying would be an interesting addition, and not as impossible as people say, though it won't be nearly as detailed and complex as the suggestor originally planned.