25
cess—as the VIM does (JCGM, 2012: 2.1)—does not seem to be correct:
measurement must be an empirical process.
Furthermore, the condition that measurement is empirical imposes a constraint
on the entities that can be considered as its inputs, and therefore as candidates for
measurement: it must be possible to interact with them in an empirical way,
9
again
a key condition to differentiate between measurement and computation. In some
cases this distinction is subtle. For example, is software an entity with which we
interact empirically? No, if a software program is considered as consisting only of
code, and therefore a sequence of zeros and ones; yes, if the program is considered
as a process executed by a given hardware system in a specific context. Accordingly,
while for example the number of lines of code of a software program is something
which is computed rather than measured, the effectiveness of the user interface of
the same program depends on several empirical conditions, and therefore it is an
empirical property that might be the object of measurement, not computation.
10
9 This is generally a two-way interaction, and hence the inputs of a measurement may be affected
by their being measured. In reference to the traditional distinction between observation and experiment, not all experiments are measurements, and some measurements are only specific kinds of
observations, whenever the measured property is unaffected by its being measured. A paradigmatic
example is the case of the measurement of the spectral characteristics of the electromagnetic radiation emitted by stars: a star does not change its state because of this process. However, an intervention might be required on the object under measurement before the measurement and in preparation
for it, an operation sometimes called “signal conditioning”, with the aim of making the property
measurable as expected. For example, electrical resistance measurement typically assumes that a
potential difference has been applied, thus in fact changing the state of the system. This justifies
the idea that usually measurement is a kind of experiment. Other perspectives about measurement
are possible. According to a slightly different construal (which still connects measurement and
quantification, a relation about which we provide some critical comments in the following), “there
are three modes of generating data: by observation, measurement, and experiment. Observation,
whether direct or with the help of instruments and theories, is deliberate and controlled perception,
and it is the basic mode of data generation. […] Measurement […] may be characterized as quantitative observation, or the observation of quantitative properties. Experiment [is] the observation
(and possibly measurement) of changes under our partial control” (Bunge, 1983: p. 91).
10 We do not define here the distinction between measurement and computation, though we aim
instead to provide a pragmatic characterization. In the words of Percy Bridgman, “There are certain human activities which apparently have perfect sharpness. The realm of mathematics and of
logic is such a realm, par excellence. Here we have yes-no sharpness. But this yes-no sharpness is
found only in the realm of things we say, as distinguished from the realm of things we do. Nothing
that happens in the laboratory corresponds to the statement that a given point is either on a given
line or it is not” (1959: p. 226, emphasis added). According to Bridgman’s metaphor, measurement
is something “we do”, and computation is something “we say”. This helps point up the paradigmatic contrast between the exactitude of computation and the uncertainty of the empirical activities of measurement.
Fig. 2.2 The abstract structure of measurement (first version)
2.2 The abstract structure of measurement
cess—as the VIM does (JCGM, 2012: 2.1)—does not seem to be correct:
measurement must be an empirical process.
Furthermore, the condition that measurement is empirical imposes a constraint
on the entities that can be considered as its inputs, and therefore as candidates for
measurement: it must be possible to interact with them in an empirical way,
9
again
a key condition to differentiate between measurement and computation. In some
cases this distinction is subtle. For example, is software an entity with which we
interact empirically? No, if a software program is considered as consisting only of
code, and therefore a sequence of zeros and ones; yes, if the program is considered
as a process executed by a given hardware system in a specific context. Accordingly,
while for example the number of lines of code of a software program is something
which is computed rather than measured, the effectiveness of the user interface of
the same program depends on several empirical conditions, and therefore it is an
empirical property that might be the object of measurement, not computation.
10
9 This is generally a two-way interaction, and hence the inputs of a measurement may be affected
by their being measured. In reference to the traditional distinction between observation and experiment, not all experiments are measurements, and some measurements are only specific kinds of
observations, whenever the measured property is unaffected by its being measured. A paradigmatic
example is the case of the measurement of the spectral characteristics of the electromagnetic radiation emitted by stars: a star does not change its state because of this process. However, an intervention might be required on the object under measurement before the measurement and in preparation
for it, an operation sometimes called “signal conditioning”, with the aim of making the property
measurable as expected. For example, electrical resistance measurement typically assumes that a
potential difference has been applied, thus in fact changing the state of the system. This justifies
the idea that usually measurement is a kind of experiment. Other perspectives about measurement
are possible. According to a slightly different construal (which still connects measurement and
quantification, a relation about which we provide some critical comments in the following), “there
are three modes of generating data: by observation, measurement, and experiment. Observation,
whether direct or with the help of instruments and theories, is deliberate and controlled perception,
and it is the basic mode of data generation. […] Measurement […] may be characterized as quantitative observation, or the observation of quantitative properties. Experiment [is] the observation
(and possibly measurement) of changes under our partial control” (Bunge, 1983: p. 91).
10 We do not define here the distinction between measurement and computation, though we aim
instead to provide a pragmatic characterization. In the words of Percy Bridgman, “There are certain human activities which apparently have perfect sharpness. The realm of mathematics and of
logic is such a realm, par excellence. Here we have yes-no sharpness. But this yes-no sharpness is
found only in the realm of things we say, as distinguished from the realm of things we do. Nothing
that happens in the laboratory corresponds to the statement that a given point is either on a given
line or it is not” (1959: p. 226, emphasis added). According to Bridgman’s metaphor, measurement
is something “we do”, and computation is something “we say”. This helps point up the paradigmatic contrast between the exactitude of computation and the uncertainty of the empirical activities of measurement.
Fig. 2.2 The abstract structure of measurement (first version)
2.2 The abstract structure of measurement