Which raises the question of what was the point of Celsius in the first place?
Fahrenheit had already existed for around 20 years when Celsius was proposed.
With units of distance, area, volume, and mass we commonly deal with a range of several orders of magnitude, and often need to subdivide larger values into equal smaller values.
Because of that immense range it makes a lot of sense to have names for specific multiples or fractions of some base unit.
The metric system made all those names derived units powers of 10 times the base unit, made those names all prefixes applied to the base unit, and used the same prefixes for length, area, volume, and mass.
That was a noticeable improvement over previous systems, where length, area, volume, and mass might use different multiples for their derived named units, the names might not be connected to the base unit, and the names were often only used with one kind of unit.
The only real problem with the metric system was that instead of defining the original meter as 10^-7 times the great circle distance from the equator to the North Pole, it should have been defined as some specific round integer multiple of some existing distance unit that was in common use and reasonably well defined.
But with temperature we almost never in ordinary life or business deal with a large range. We don't need names for specific multiples or fractions.
Fahrenheit's original reference points might not have been the best, but if you wanted to switch to basing on something else you can do that without changing the scale. Fahrenheit's original 0 point was the lowest temperature that could be achieved in the lab via a specific procedure. The other reference point was average human body temperature. On that scale water freezes at around 32℉.
If you wanted to instead have the freezing point of water as your base, you don't need to make a new scale that defines the 0 point as the freezing of water. Instead you can keep the degree size the same, and define the 0 point as 32 degrees below the freezing point of water. Similar for the upper reference point. Define it so that water boils at 212.
That gives you a scale that is defined based on freezing and boiling water (which is what Celsius used, although originally Celsius went the other way--water froze at 100 and boiled at 0--but that was pretty quickly swapped), but that still matches the thermometers that are already in use.
Same thing when later you realize that freezing and boiling point of water aren't the best reference points. We've moved on from freezing/boiling to using the triple point of water, then to the triple point of water with a specific isotropic composition, and now it is based on the Boltzmann constant and has nothing to do with water.
Each time, we set the values so that the resulting scale matches the new scale to within the limits of accuracy that we can measure at the time.
It doesn't really matter what the exact reasons behind the Celsius scale was, it happened hundreds of years ago... The fact is that basically every field of science has settled on it for scientific use, and almost every country has settled on it for everyday use.
Fahrenheit had already existed for around 20 years when Celsius was proposed.
With units of distance, area, volume, and mass we commonly deal with a range of several orders of magnitude, and often need to subdivide larger values into equal smaller values.
Because of that immense range it makes a lot of sense to have names for specific multiples or fractions of some base unit.
The metric system made all those names derived units powers of 10 times the base unit, made those names all prefixes applied to the base unit, and used the same prefixes for length, area, volume, and mass.
That was a noticeable improvement over previous systems, where length, area, volume, and mass might use different multiples for their derived named units, the names might not be connected to the base unit, and the names were often only used with one kind of unit.
The only real problem with the metric system was that instead of defining the original meter as 10^-7 times the great circle distance from the equator to the North Pole, it should have been defined as some specific round integer multiple of some existing distance unit that was in common use and reasonably well defined.
But with temperature we almost never in ordinary life or business deal with a large range. We don't need names for specific multiples or fractions.
Fahrenheit's original reference points might not have been the best, but if you wanted to switch to basing on something else you can do that without changing the scale. Fahrenheit's original 0 point was the lowest temperature that could be achieved in the lab via a specific procedure. The other reference point was average human body temperature. On that scale water freezes at around 32℉.
If you wanted to instead have the freezing point of water as your base, you don't need to make a new scale that defines the 0 point as the freezing of water. Instead you can keep the degree size the same, and define the 0 point as 32 degrees below the freezing point of water. Similar for the upper reference point. Define it so that water boils at 212.
That gives you a scale that is defined based on freezing and boiling water (which is what Celsius used, although originally Celsius went the other way--water froze at 100 and boiled at 0--but that was pretty quickly swapped), but that still matches the thermometers that are already in use.
Same thing when later you realize that freezing and boiling point of water aren't the best reference points. We've moved on from freezing/boiling to using the triple point of water, then to the triple point of water with a specific isotropic composition, and now it is based on the Boltzmann constant and has nothing to do with water.
Each time, we set the values so that the resulting scale matches the new scale to within the limits of accuracy that we can measure at the time.