Celcius is really stupid (a rant, I mean dissertation)

Great Debates
61

In the hospital, our monitors register body temp in Celcius. In home health, my thermometer inexplicably sometimes records in Celcius but sometimes in Fahrenheit. Only my ICU experience enables me to instantly know what a person’s temp is in F when the thermometer reads in C without having to use scratch paper.

98.6F= 37C.

Celcius is easier to me when evaluating body temperature but seems foreign when judging air temperature. It’s all a matter of which system you use on a daily basis, for which tasks.

62

“New Shimmer is a floor wax”

“It’s a dessert topping!!”

I for one welcome our new insect overlords… - K. Brockman

63

Erratum wrote:

Apparently, European recipes call for weights more often than volumes, at least when measuring dry ingredients. A scale is a common kitchen appliance over on the other side of the pond.

64

But he said liquids. Quite specific was he.

65

Ah, but then how do you explain the existence of pound cake, huh?!

66

Is pound cake a liquid or a solid?

67

In my business (computer programming), I go by the premise that the replacement of an established system has to be ten times better than the original to be sucessful.

IMHO, The divisiblity and interrelation between the measures in the metric system do indeed make it 10 times better than the English system.

However, as the original post makes clear, the Celcius scale does not offer either divisibility or interrelation benefits. It just redefines the unit. Both scales are decimal (in terms of subdivisions of the unit), and there aren’t any exact major correlations between other units in daily life.

Hail Ants is essentially correct: The Celcius scale does not offer any of the benefits found in the other metric measurement systems.

He’s the sort to stand on a hilltop in a thunderstorm wearing wet copper armor, shouting ‘All Gods are Bastards!’

68

“The divisiblity and interrelation between the measures in the metric system do indeed make it 10 times better than the English system.”

Is this really the case? I believe that it is not. In fact, I believe that the requirement that all units must be factor-of-ten related is a significant weakness of the metric system. In many instances, a factor of ten relationship results in units which are either inconveniently large or inconveniently small for a particular task. In contrast, the imperial system based its units on common activities, and therefore there is a high probability that there will be a “convenient” unit for almost any task.

There are very few everyday tasks which do not have a conveniently sized imperial unit associated with them. In contrast, however, there are many everyday tasks in which the nearby metric unit is an awkward fit. For example, when talking about how tall someone is, the meter is a useless unit, because everyone is “about two meters tall”. We then step down to the next popular metric unit, the centimeter, which is far too small for comfort, because it implies an amount of precision we usually don’t want to when we are talking about height. In contrast, the feet/inches system seems to be very closely suited to the kind of resolution we want to use when discussing height. Now, metric proponents will immediately chime in with “oh, that’s just because you grew up with it”, but I disagree. In order to get meters or centimeters to work on this scale, you either need to use fractional numbers of meters or way too many centimeters. With the imperial system, there is a large probability that you always have a unit which is very close to the “right” size with which to measure something.

I said earlier that the light-year is not a metric unit. We use the light-year because it is so incredibly convenient and communicates so effectively. That is not a coincedence – the metric system cares nothing for convenience of the unit, and everything for conformance to a standard dreamt up by 18th century french intellectuals.

I shall offer the following observation: If powers of ten were so convenient, there would be no nickels or quarters, and there would be no five, twenty, or fifty dollar bills. Power of ten relationships are nice when you are doing math by hand, but they have little relationship to what we actually use a system of weights and measures for – which is communicating.

Now, I will offer an alternative: If ease of calculation is truly important for a system of weights and measures, let’s actually make the system easy to calculate for the things which actually do our calculations now: computers. With a binary or hexadecimal measurement system, we could eliminate my “resolution” complaint, and we could also make calculations easier (no cycles would be wasted converting from base 10 to base 2). “But I don’t know how to do hexadecimal arithmetic!” some will say, “I grew up with the decimal system!”. Metric proponents are always telling us that the system we grew up with shouldn’t influence our choice of what the “best” system is – are they wrong?

69

Huh?

So, you’d remove all digits from the numeric keypad except for zero and one?

70

No, Lib, I’d go for hexadecimal (0,1,2,3,4,5,6,7,8,9,A,B,C,D,E,F) which most people in the computer industry recognize as incredibly useful. Notice that base-16 and base-2 are closely related.

Actually, I’d say we should just stick with what we’ve got, but if we’re going to change to anything, I’d prefer it be to a system that was good rather than just a system which was popular.

71

Erratum, better sit down for this. I think you make an excellent point.

72

Actually, Erratum, the resolution problem has nothing to do with the conversion from base 10 to base 2. Computer arithmetic is already performed in binary, as I know you are aware. The resolution problem occurs because some commonly used fraction cannot be efficiently expressd in base 2. The limitations on register size prevent the necessary bits from being dedicated to an exact representation (where one exists), therefore we have resolution issues. It has nothing to do with the base in which we input the numbers; it is a property of the value as expressed in base 2.

For that matter, the cycles necessary to convert input expressions into binary equivalents is pretty negligible at today’s clock speeds and pipelined architectures. Still, it is a nice philosophical point. I personally like the computational ease of metric for many things, but I never get on the scale and think, “damn – I weigh 100 kilos!”

The best lack all conviction
The worst are full of passionate intensity.
*

73

Lib, I’m speechless. :smiley:

Spiritus Mundi: “some commonly used fraction cannot be efficiently expressd in base 2.”

You mean like 1/3 = .3333333333333… in base 10, but it is .0101010101010101… in base 2? Seems roughly the same to me. Base 2 is much better at 1/2, 1/4, 1/8, 1/16, etc. than base 10. Base 10 can give you 1/5ths and 1/10ths, I suppose, but I don’t find those stunningly useful. 1/3rds don’t work in either system, as I’ve shown. Which fractions are you particularly concerned about?

“the resolution problem has nothing to do with the conversion from base 10 to base 2”

You are confusing two issues here (I didn’t do a good enough job of separating them). The “resolution problem” is a purely human problem, and comes because when you have one unit of size x and another of size 10x, and another or size x/10, then the probability that some arbitrary measurement is a small integer number of those units is very, very small. If, however, you have x/4, x/2, x, 2x, and 4x then the odds of having a convenient unit that allows you to represent a measurement as a small integer number of units is very high. The 10x conversion factors which are required in the metric system frequently create a “no man’s land” between too-large and too-small units. If the units weren’t so far apart (as in a base 2 system), you will almost never get the “no man’s land” effect.

“For that matter, the cycles necessary to convert input expressions into binary equivalents is pretty negligible at today’s clock speeds and pipelined architectures.”

True. However, a mismatch between the input format and the internal representation can lead to errors when people fail to recognize the resolution of the numbers they are using (e.g. how many people really know the difference between single and double precision floating point?). And, at today’s clock speeds and pipelined architectures, the non-power-of-ten conversion factors in the imperial system do not make calculations difficult. The “easy calculations” of the metric system were only an advantage before computers. Personally, I always do any important calculation via computer or calculator, since the chance for human error in the pencil-and-paper approach is too high. Therefore, I would argue that the calculation “simplicity” of the metric system is completely illusory in actual practice. If we care about an “easy calculation” system, we ought to optimize it for calculating machines. If we are going to trust fast calculating machines to do our calculations for us, why try to build an “easy calculation” system in the first place? Metric proponents can’t have it both ways.

74

Erratum:
I had thought that the error you meant was the “rounding” error which is implicit in the computer’s representational scheme and independent of the base used to input the number. Since that is not what you were talking about:

The best lack all conviction
The worst are full of passionate intensity.
*

75

You left out the “IMHO”… I was expressing my opinion. YMMV.

I’m 5’ 9 3/4" tall. Aka 157 cm. About the same level of precision, and centimeters uses one fewer digit.

On the whole, though, your main point is correct. The English units were empircally determined. IMHO, the metric proponents should have used standard English measures and just decimalized them instead of caring “everything for conformance to a standard dreamt up by 18th century french intellectuals.” (chortle!)

The light year and the parsec are still decimilized units (we talk about 10ths and hundreds) and they both have interesting interrelations (the parsec, relating to the angular parallax across the orbit of the earth (I think)). So they do have the benefits I cited for the metric system.

Excuse me? It’s ridiculous to expect every convenient measurement in a decimal-based unit to add up to a power of 10.

Computers and programmers don’t care a whit about what units you use, and how they divide, as long as you’re consistent. And it takes a trivial number of cycles necessary to convert decimal to binary, ocurring only on input and output (internal arithmetic is, of course performed in binary), People, however, perform arithmetic very inefficiently, and we need all the help we can get.

I think you had your tounge in your cheek a little, but you make important points nonetheless.

He’s the sort to stand on a hilltop in a thunderstorm wearing wet copper armor, shouting ‘All Gods are Bastards!’