chriswarbo-net: a39b10207d84dc082ce6126247b207a6ed630d1a

     1: ---
     2: title: The Metric Red Herring
     3: ---
     4: 
     5: <small>Part of [my units pages](/projects/units)</small>
     6: 
     7: This began as a companion to my post about
     8: [improving our units](../../blog/2020-05-22-improving_our_units.html), which
     9: discusses some problems with the metric system, SI and related ideas. This post
    10: was a stand-alone argument that advocates of metric are often doing it for the
    11: wrong reasons.
    12: 
    13: Both are now part of a larger [collection of suggestions to improve and simplify
    14: measurement, notation, numeracy, etc.](./index.html)
    15: 
    16: ## The Powers-of-Ten Red-Herring ##
    17: 
    18: Most discussions about metric versus imperial units seem to dwell on one
    19: specific aspect of metric: that units are related by powers of 10. For example:
    20: 
    21:  - 100cm = 1m
    22:  - 10mm = 1cm
    23:  - 1kN = 1000N
    24: 
    25: However, this is a red herring: those examples have *the same units* on each
    26: side! They only differ in how they factor the multiple; solving these equations
    27: reveals the definition of those factors:
    28: 
    29: The first tells us that `c` ("centi") is a hundredth:
    30: 
    31: ```
    32: 100cm = 1m
    33: 100c  = 1
    34:    c  = 1/100
    35: ```
    36: 
    37: The second tells us that `m` ("milli") is a thousandth:
    38: 
    39: ```
    40: 10mm = 1cm
    41: 10m  = 1c
    42: 10m  = 1/100
    43:   m  = 1/1000
    44: ```
    45: 
    46: The third tells us that `k` ("kilo") is a thousand:
    47: 
    48: ```
    49: 1kN = 1000N
    50:  k  = 1000
    51: ```
    52: 
    53: Notice that these are just statements about numbers: the units we wrote in each
    54: case are irrelevant, since they immediately cancelled-out! We could have done
    55: the same with inches, pounds, etc. There's also nothing special about ten: there
    56: are [convenient factors for many bases](prefix_factors.html), e.g. 3Kim = 3072m.
    57: 
    58: ## Metric is Minimal ##
    59: 
    60: Focusing on "conversions" like 1km = 1000m obscures a much more important
    61: feature of metric: there is only *one* unit of distance. Likewise there is
    62: only *one* unit of force, *one* unit of pressure, and so on for each distinct
    63: [dimension](https://en.wikipedia.org/wiki/Dimensional_analysis).
    64: 
    65: Quantities 'expressed in kilometres' are still expressed in metres, since
    66: kilometres are just multiples of metres. A distance like "5m" is expressed in
    67: metres (five of them, since there's a "5" which means five). A distance like
    68: "7km" is *also* expressed in metres (seven thousand of them, since there's a "7"
    69: which means seven and a "k" which means thousand). "centi", "milli", "femto",
    70: "kilo", "giga", etc. are just generic ways to abbreviate large and small
    71: numbers.
    72: 
    73: If we use a multiple which isn't base 10, like "two dozen metres", that's still
    74: a metric distance; we haven't invented a new system of units with base 12;
    75: "dozen" is just a generic linguistic device meaning "twelves". Compare this to a
    76: foot containing a dozen inches: if we ask a baker for a dozen buns we'll get 12
    77: buns; if we ask for a foot of buns we'll a line of buns about a third of a metre
    78: long. This is because "foot" is *not* a generic multiplier: it is specific to
    79: distance, and it is independent of other distance units (e.g. we don't need to
    80: say "a foot of inches"). The same applies to "inch", "mile", etc. hence they are
    81: all separate units, not just multipliers. In contrast, "metres" and "dozen
    82: metres" are not separate units, and neither are "metres" and "centimetres".
    83: 
    84: (Note that there is a slight wrinkle here, since "kilo" on its own is taken to
    85: mean "kilograms"; this is not too important for this claim, but I explain why
    86: the kilogram is problematic in the "Problems with SI" section of
    87: [the companion post](improving_our_units.html)!)
    88: 
    89: I think this is such a profound advantage that many (most?) people, even those
    90: born and raised with metric, never grasp it explicitly. After all, why argue
    91: that "power-of-ten conversions are easier" when we could go further and say
    92: "there's nothing to convert between"; it can't get any easier than that!
    93: 
    94: ## Metric is Coherent ##
    95: 
    96: In [a 'coherent' system of units](coherence.html), we can combine our units
    97: using multiplication and division, and alway get another unit of our system. The
    98: most obvious example is speed, which is the same as a distance divided by a
    99: time. There are imperial units of speed like the knot, but it's more common to
   100: use the "distance over time" form like "miles per hour", or "metres per second"
   101: in metric. Similarly for pressure, which is often measured in "pounds per square
   102: inch" or "Newtons per square metre".
   103: 
   104: We can do the same thing with any dimensions we like. For example, applying a
   105: force (e.g. Newtons or pounds) over a distance (metres or feet) takes a certain
   106: amount of energy (Joules or calories); hence we can express distances in units
   107: of "energy over force". We can think of this as how far we could push with a
   108: unit of force, before we use up the given energy.
   109: 
   110: These 'derived units' will be some multiple of the 'normal' units for that
   111: dimension, requiring a conversion factor if we want to convert a number between
   112: the two forms. Here are some conversion factors for common imperial units:
   113: 
   114: <div class="summarise">
   115:  <span class="summary">
   116:   Conversion factors between various imperial distance units.
   117:  </span>
   118: 
   119: Quantity            Inches  Feet     Miles    Calories per pound Calories per ounce
   120: ------------------- ------  -------  -------  ------------------ ------------------
   121: 1 inch              1       0.083    2x10^-5^ 0.03               0.43
   122: 1 foot              12      1        2x10^-4^ 0.32               5.18
   123: 1 mile              6x10^4^ 5280     1        1711               3x10^4^
   124: 1 calorie per pound 37.03   3.09     6x10^-4^ 1                  16
   125: 1 calorie per ounce 2.31    0.19     4x10^-5^ 0.06               1
   126: 
   127: </div>
   128: 
   129: There are conversion tables like this for many other combinations of units, e.g.
   130: to find how many slug feet per square hour are in a stone. If the order of the
   131: units is the same in the rows and columns then the main diagonal will always be
   132: 1. For the remaining numbers, we only need to remember one of the 'triangles'
   133: (upper-right or lower-left), since we can get the other by 'reflecting' the
   134: positions across the diagonal and taking the reciprocal of the numbers. (Or you
   135: can [look them up](https://lmgtfy.com/?q=+1+calorie+per+pound+in+miles+!g&s=d)
   136: like I did, rather than cluttering your brain with obsolete junk!)
   137: 
   138: The metric equivalent doesn't have the inch/foot/mile or ounce/pound
   139: redundancies, so the table is much smaller, and hence easier to memorise:
   140: 
   141: <div class="summarise">
   142:  <span class="summary">
   143:   Conversion factors between metric distance units.
   144:  </span>
   145: 
   146: Quantity           Metres Joules per Newton
   147: ------------------ ------ -----------------
   148: 1 metre            1      1
   149: 1 joule per newton 1      1
   150: 
   151: </div>
   152: 
   153: The conversion factors here are all 1. This is not a coincidence, it is *by
   154: design*! [1 Joule is *defined as* 1 Newton
   155: metre.](https://www.bipm.org/utils/common/pdf/si_brochure_8_en.pdf) Likewise:
   156: 
   157: ```
   158: 1 Watt = 1 Joule per second
   159:        = 1 Hertz Joule
   160:        = 1 Hertz Newton metre
   161:        = 1 Hertz Pascal cubic metre
   162:        = 1 Hertz square Coulomb per Farad
   163:        = 1 Newton Coulomb per Tesla Farad metre
   164:        = 1 Newton Joule per Tesla Coulomb metre
   165:        = 1 Newton Volt per Tesla metre
   166:        = 1 Amp Volt
   167:        = 1 Watt
   168: ```
   169: 
   170: This is a *huge* advantage to using metric, which I rarely/never see brought up
   171: in discussions. I'm not sure whether this is because it's subconsciously taken
   172: for granted (like the one-unit-per-dimension feature) or whether it's just used
   173: less frequently in "real life" (e.g. day-to-day estimating, rather than explicit
   174: engineering calculations). Either way, I think this should be celebrated more!
   175: 
   176: In particular, these conversions are based off known scientific laws. For
   177: example Newton's second law of motion, usually written `F = ma`, tells us that
   178: multiplying a mass by an acceleration results in a force. This is actually a
   179: statement of *proportionality*, e.g. doubling the mass will double the force; to
   180: get an equation we need a "constant of proportionality", which is an arbitrary
   181: scaling factor which we usually write as `k`, so the general form of Newton's
   182: second law should really be expressed as `F = kma`. Metric units are *defined
   183: such that* these scaling factors are 1, which gives us simple equations without
   184: having to remember a bunch of proportionality constants (i.e. those shown in the
   185: tables above).
   186: 
   187: ## Exposing My Lies ##
   188: 
   189: I have to admit that the above is slightly inaccurate, for the purposes of
   190: getting across my way of thinking. The first major point to clarify is that when
   191: I say "metric" I'm actually referring mostly to
   192: [the SI system](https://en.wikipedia.org/wiki/International_System_of_Units),
   193: which differs a little from the metric units that are commonly used day-to-day.
   194: In particular, metric often uses extra, redundant units which are not part of
   195: SI, including the litre, tonne and hour.
   196: 
   197: Next, I claimed above that prefixing a unit changes the associated number rather
   198: than the unit, e.g. "2km" is 2000 in the unit of metres, rather than 2 in the
   199: unit of kilometres. In fact, [the SI
   200: definition](https://www.bipm.org/utils/common/pdf/si_brochure_8_en.pdf)
   201: explicitly states that prefixing a unit with a multiplier, like "kilometre",
   202: gives us a new, "derived" unit. This is important for resolving otherwise
   203: ambiguous quantities like "3cm^3^": according to SI, this is 3(cm)^3^ =
   204: 0.000003m^3^, whereas treating prefices in the way I describe would give
   205: 3c(m^3^) = 0.03m^3^. Similar problems arise when dividing, e.g. "per kilometre".
   206: 
   207: Whilst the SI method is well-defined, it still places a mental burden on the
   208: user. What's especially annoying is that the SI rules for units are opposite to
   209: the usual algebraic rules for multiplication and exponentiation, where
   210: ab^3^ = a(b^3^)
   211: 
   212: I cover these in more depth in
   213: [the companion post](improving_our_units.html), but I think the best
   214: thing to do in these situations is stick to the base unit for that dimension
   215: (e.g. "cubic metre" or "per metre"), apply prefix multipliers if they are
   216: unambiguous; or otherwise add
   217: [explicit parentheses](https://en.wikipedia.org/wiki/S-expression).
   218: 
   219: ## Take Aways ##
   220: 
   221: Whilst the metric (or SI) system
   222: [isn't perfect](improving_our_units.html), it's also much better than
   223: those hodge-podges of historical baggage known as imperial units. The most
   224: obvious argument against imperial units is that [there are so many, related in
   225: arbitrary ways](https://www.youtube.com/watch?v=r7x-RGfd0Yk); but
   226: that's a bit of a cheap shot, since few people make regular use of barleycorns,
   227: fathoms or leagues. Likewise the common argument *against* imperial, that powers
   228: of ten make arithmetic easier, is shallow at best, and irrelevant at worst.
   229: Instead, the two *real* advantages of metric (or SI) are having one unit per
   230: dimension, and requiring no conversion factor when combining dimensions.
   231: 
   232: I'm a firm believer that seemingly-innocuous complications, like those found in
   233: imperial units of measurement, are in fact significant risks; they impede
   234: learning, potentially turning people away from areas like maths and science; and
   235: their [compounding, confounding behaviour on the large
   236: scale](https://en.wikipedia.org/wiki/Mars_Climate_Orbiter) constrains what we're
   237: capable of achieving as a species.
   238: 
   239: Every small "gotcha" can be pre-empted on its own, but it takes knowledge and
   240: experience to do so, and some small effort every time. As such "minor" issues
   241: combine together, they can quickly overflow our limited mental capacity,
   242: making it naïve and irresponsible to think their individual avoidability can
   243: hold in general. (As a programmer, such
   244: [seemingly](https://en.wikipedia.org/wiki/Strong_and_weak_typing)
   245: [minor](http://wiki.c2.com/?CeeLanguageAndBufferOverflows)
   246: [problems](https://en.wikipedia.org/wiki/Code_injection) are quite widespread,
   247: and even skilled experts often slip up now and then!)
   248: 
   249: The only way to combat such unnecessary complication is by an aggressive pursuit
   250: of simplicity. Irrelevant details, unwanted degrees of freedom and unnecessary
   251: asymmetries only act to slow us down and trip us up. Imperial units have
   252: infected our collective mind for so long that we're often unable to see the
   253: simplicity that metric provides: we used to waste so much effort converting
   254: between redundant imperial units that, when confronted with a single metric
   255: equivalent, we started treating multiples as if they were different units, just
   256: to make it more familiar.
   257: 
   258: The other advantage, of combining dimensions, is alien to many, despite the
   259: prevalence of examples like "miles per hour" and "pounds per square inch".
   260: Presumably this is due to how horrible it is to convert between imperial units
   261: in this way. It might even be the case that quantities like "miles per hour" and
   262: "pounds per square inch" are acceptable precisely because there's no expectation
   263: that they be convertible to any existing units (other than their constituents,
   264: like "miles" and "hours" for "miles per hour"). This mentality might explain why
   265: someone thought it was a good idea to invent monstrosities like "kilowatthours",
   266: rather than just sticking a "mega" prefix on to the Joule!
   267: 
   268: In any case, we need to embrace the simplicity of metric; grok what it tells us
   269: about the nature of measure and dimension; and use the saved mental effort to
   270: tackle bigger, harder problems.

Generated by git2html.