241
on the meaning(s) that someone might associate with them. Data acquisition is
then where the process is grounded. But this still leaves open the question of
what data is. Let us rely again on Weaver’s presentation: “To be sure, [data]
relates not so much to what you do say, as to what you could say. That is, [data
affects] one’s freedom of choice when one selects a message. If one is confronted with a very elementary situation where he has to choose one of two
alternative messages, then it is arbitrarily said that the [syntactic] information,
associated with this situation, is unity. […] The two messages between which
one must choose, in such a selection, can be anything one likes” (p. 9).
1
This
leads to the basic understanding of data as support of selection of difference—
it is this, but it might have been that. That is why any binary system is structurally the simplest provider of data: it is 0 (or false, or white, or …) but it might
have been 1 (or true, or black, or …). More generally, whenever a set of possibilities, X = {x i }, is available, then the selection of one or more of its elements
is a provider of data—i.e., it is what is commonly referred to as “the raw data”.
(B) What Weaver calls the semantic problem broadens the scope of the technical/
syntactic problem, by acknowledging that we usually acquire and manage data
for referring to something, not to perform a purely syntactic activity. Despite
the wealth of results that can be obtained at the syntactic layer—the way to
which was paved by Shannon’s two fundamental theorems about source
entropy and channel capacity—human interest is usually focused on data as
carriers of meanings. In other words, data may have a meaning, if the elements
of the set refer to, or stand for,
2
something else outside the set itself. This
merges them into a second layer, in which the emphasis is on “the relations of
signs to the objects to which the signs are applicable” (Morris, 1946: p. 217).
Data equipped with meaning is called semantic information. The core arguments of this book relate to this: information about the measurand.
(C) Finally, what Weaver calls the effectiveness problem builds upon the semantic
layer and adds the context in which data with meaning is used by some agents
for some purposes, where fitness for purpose is sometimes called the “value”
of data.
3
Indeed, the same syntactic entity, e.g., the string “n-o”, once equipped
1 This analogy between communication and measurement should be read while keeping attention
also to their substantial differences, as highlighted in Sect. 4.2.1.
2 The relation sign stands for entity is very general. Famously, Charles Sanders Peirce identified
three ways in which it can be realized. “If we come to interpret a sign as standing for its object in
virtue of some shared quality, then the sign is an icon. Peirce’s early examples of icons are portraits
[…]. If […] our interpretation comes in virtue of some brute, existential fact, causal connections
say, then the sign is an index. Early examples include the weathercock, and the relationship
between the murderer and his victim […]. And finally, if we generate an interpretant in virtue of
some observed general or conventional connection between sign and object, then the sign is a
symbol. Early examples include the words ‘homme’ and ‘man’ sharing a reference” (Atkin, 2013;
emphasis added). In this semiotic perspective, indication values (i.e., local values) can be interpreted as indexes of measurands, and measured values (i.e., public values) as icons of
measurands.
3 Of course, this is the concept of related to goodness (Schroeder, 2016), which is different
8.1 Introduction
on the meaning(s) that someone might associate with them. Data acquisition is
then where the process is grounded. But this still leaves open the question of
what data is. Let us rely again on Weaver’s presentation: “To be sure, [data]
relates not so much to what you do say, as to what you could say. That is, [data
affects] one’s freedom of choice when one selects a message. If one is confronted with a very elementary situation where he has to choose one of two
alternative messages, then it is arbitrarily said that the [syntactic] information,
associated with this situation, is unity. […] The two messages between which
one must choose, in such a selection, can be anything one likes” (p. 9).
1
This
leads to the basic understanding of data as support of selection of difference—
it is this, but it might have been that. That is why any binary system is structurally the simplest provider of data: it is 0 (or false, or white, or …) but it might
have been 1 (or true, or black, or …). More generally, whenever a set of possibilities, X = {x i }, is available, then the selection of one or more of its elements
is a provider of data—i.e., it is what is commonly referred to as “the raw data”.
(B) What Weaver calls the semantic problem broadens the scope of the technical/
syntactic problem, by acknowledging that we usually acquire and manage data
for referring to something, not to perform a purely syntactic activity. Despite
the wealth of results that can be obtained at the syntactic layer—the way to
which was paved by Shannon’s two fundamental theorems about source
entropy and channel capacity—human interest is usually focused on data as
carriers of meanings. In other words, data may have a meaning, if the elements
of the set refer to, or stand for,
2
something else outside the set itself. This
merges them into a second layer, in which the emphasis is on “the relations of
signs to the objects to which the signs are applicable” (Morris, 1946: p. 217).
Data equipped with meaning is called semantic information. The core arguments of this book relate to this: information about the measurand.
(C) Finally, what Weaver calls the effectiveness problem builds upon the semantic
layer and adds the context in which data with meaning is used by some agents
for some purposes, where fitness for purpose is sometimes called the “value”
of data.
3
Indeed, the same syntactic entity, e.g., the string “n-o”, once equipped
1 This analogy between communication and measurement should be read while keeping attention
also to their substantial differences, as highlighted in Sect. 4.2.1.
2 The relation sign stands for entity is very general. Famously, Charles Sanders Peirce identified
three ways in which it can be realized. “If we come to interpret a sign as standing for its object in
virtue of some shared quality, then the sign is an icon. Peirce’s early examples of icons are portraits
[…]. If […] our interpretation comes in virtue of some brute, existential fact, causal connections
say, then the sign is an index. Early examples include the weathercock, and the relationship
between the murderer and his victim […]. And finally, if we generate an interpretant in virtue of
some observed general or conventional connection between sign and object, then the sign is a
symbol. Early examples include the words ‘homme’ and ‘man’ sharing a reference” (Atkin, 2013;
emphasis added). In this semiotic perspective, indication values (i.e., local values) can be interpreted as indexes of measurands, and measured values (i.e., public values) as icons of
measurands.
3 Of course, this is the concept of
8.1 Introduction