> while investigating the crystallization rates of metal, Czochralski dipped his pen into molten tin instead of an inkwell. That caused a tin filament to form on the pen’s tip
And later, describes the process for silicon
> Once the silicon melted, he placed a small piece of polycrystalline material—a seed crystal
It seems then, that the seed crystal is not anything special. A typical piece of metal (such as a the pen tip) is made of up multiple single-crystal "grains" with non-crystalline "grain boundaries" between them; this is a polycrystalline material.
The question then is, why does only a single crystal form, rather than multiple crystal filaments oriented at different directions according to whichever grain contacted the starting point of the filament...
The silicon wants to be in the lowest energy state and it does so by forming a face centered cubic crystal (diamond structure). The formation of crystals depends on the cooling rate. If matter cools to fast, it can't form a crystal structure.