> 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...