I made a previous video about how, since water expands when it freezes, if you try to freeze water inside a rigid container, any of it that becomes ice will expand and pressurize the container, eventually stopping the rest of the water from freezing. So if you have supercooled water inside a rigid container, there aren't just the two competing factions of interior wanting to freeze and surface wanting to melt – there's a third player, pressure. Pressure makes it harder for ice crystals to grow, so pressure, like the surface, wants ice crystals to melt. Obviously, if you have a large amount of water, a tiny amount of it freezing into ice and expanding doesn't pressurize the remaining liquid very much, so a lot of ice can form before there's enough pressure to matter. Pressure simply doesn't come into the picture until way beyond the tipping point size for runaway ice crystal formation. But if you have a tiny amount of water, the pressure can be substantially affected by the formation of even a single tiny ice crystal. For containers that are smaller than a few tenths of a micrometer in size, instead of a tipping point where once an ice crystal gets big enough it keeps growing and growing, in this case the pressure's desire to melt overwhelms the interior's desire to freeze for all crystal sizes, and all ice crystals want to melt regardless of their size. So supercooled water in a tiny, rigid bottle won't freeze no matter how hard you smack it, as long as the bottle is tiny enough. For water 1°C below zero, the bottle needs to be less than 200 nm long.