INFORMATION

Science, technology, and ethics are all forms of information that depend on information to work. Furthermore there exist sciences, technologies, and ethics of information. To disentangle some of the main relations among these aspects of information, it is helpful to start with a simple example.

Monday morning. John turns the ignition key of his car, but nothing happens: The engine does not even cough. Not surprisingly the low-battery indicator is flashing. After a few more unsuccessful attempts, John calls the garage and explains that, last night, his wife had forgotten to turn off the car's lights—this is a lie, John did but is too ashamed to admit it—and now the battery is dead. John is told that the car's operation manual explains how to use jumper cables to start the engine. Luckily his neighbor has everything John needs. He follows the instructions, starts the car, and drives to the office.

This everyday example illustrates the many ways in which people understand one of their most important resources: information. The information galaxy is vast, and this entry will explore only two main areas: information as content and information as communication. The reader interested in knowing more about the philosophical analysis of the concept should consult the work of Jaakko Hintikka and Patrick Suppes (1970), Philip P. Hanson (1990), and Fred I. Dretske (1999).

Information as Content

It is common to think of information as consisting of data (Floridi 2005). An intuitive way of grasping the notion of data is to imagine an answer without a question. Ultimately data may be described as relational differences: a 0 instead of a 1; a red light flashing; a high or low charge in a battery.

To become information, data need to be well-formed and meaningful. Well-formed means that data are clustered together correctly, according to the rules (syntax) of the chosen language or code. For example, the operation manual from the example above shows the batteries of two cars placed one next to, not one on top of, the other. Meaningful indicates that the data must also comply with the meanings (semantics) of the chosen language or code. So the operation manual contains illustrations that are immediately recognizable.

When meaningful and well-formed data are used to talk about the world and describe it, the result is semantic content (Bar-Hillel and Carnap 1953, Bar-Hillel 1964). Semantic content has a twofold function. Like a pair of pincers, it picks up from or about a situation, a fact, or a state of affairs f, and models or describes f. The battery is dead carves and extracts this piece of information—that the battery of the car is dead—and uses it to model reality into a semantic world in which the battery is dead. Whether the work done by the specific pair of pincers is satisfactory depends on the resource f (realism) and on the purpose for which the pincers are being used (teleologism). Realistically the battery is dead is true. Teleologically it is successful given the goal of communicating to the garage the nature of the problem. The battery is dead would be realistically false and teleologically unsatisfactory if it were used, for instance, to provide an example of something being deceased.

INFORMATION AS TRUE SEMANTIC CONTENT. True semantic content is perhaps the most common sense in which information can be understood (Floridi 2005). It is also one of the most important ways, since information as true semantic content is a necessary condition for knowledge. Some elaboration of this concept is in order. First the data that constitute information allow or invite certain constructs and resist or impede others. Data in this respect work as constraining affordances. Second the data are never accessed and elaborated independently of a level of abstraction (LoA). An LoA is like an interface that establishes the scope and type of data that will be available as a resource for the generation of information (Floridi and Sanders 2004). The battery is what provides electricity to the car is a typical example of information elaborated at a driver's LoA. An engineer's LoA may output something like a 12-volt lead-acid battery is made up of six cells, each cell producing approximately
2.1 volts, and an economist's LoA may suggest that a good quality car battery will cost between $50 and $100 and, if properly maintained, it should last five years or more. Data as constraining affordances—answers waiting for the relevant questions—are transformed into information by being processed semantically at a given LoA (alternatively the right question is associated to the right data at a given LoA).

Once information is available, knowledge can be built in terms of justified or explained information, thus providing the basis of any further scientific investigation. One knows that the battery is dead not by merely guessing correctly, but because one sees the red light of the low-battery indicator flashing and perceives that the engine does not start. The fact that data count as resources for information, and hence for knowledge, rather than sources, provides a constructionist argument against any representationalist theory that interprets knowledge as a sort of picture of the world.

An instance of misinformation arises when some semantic content is false (untrue) (Fox 1983). If the source of the misinformation is aware that the semantic content is false, one may speak of disinformation, for example my wife left the lights on. Disinformation and misinformation are ethically censurable but may be successful teleologically: If one tells the mechanic that one's wife left the lights on last night, the mechanic will still be able to provide the right advice. Likewise information may fail to be teleologically successful; just imagine telling the mechanic that one's car is out of order.

INSTRUCTIONAL INFORMATION. True semantic content is not the only type of information. The operation manual, for example, also provides instructional information, either imperatively—in the form of a recipe: First do this, then do that—or conditionally—in the form of some inferential procedure: If such and such is the case do this, otherwise do that. Instructional information is not about f and does not model f: It constitutes or instantiates f, that is, it is supposed to make f happen. The printed score of a musical composition or the digital files of a program are typical cases of instructional information. The latter clearly has a semantic side. And semantic and instructional information may be joined in performative contexts, such as christening a vessel—for example, "this ship is now called HMS The Informer"—or programming—for example, when declaring the type of a variable. Finally the two types of information may come together in magic spells, where semantic modeling is confused with instructional power and control. Yet, as a test, one should recall that instructional information does not qualify alethically (from aletheia, the Greek word for truth). In the example, it would be silly to ask whether only use batteries with the same-rated voltage is true or false.

ENVIRONMENTAL INFORMATION. When John turned the ignition key, the low-battery indicator flashed. He translated the flashing into (a) semantic information: The battery is dead; and (b) instructional information: The battery needs to be charged or replaced. However the flashing of the indicator is actually an example of environmental information.

Environmental information may be described as natural data: It requires two systems a and b to be coupled in such a way that a being (of type, or in state) F is correlated to b being (of type, or in state) G, thus carrying to the observer the information that b is G (Jon Barwise and Jerry Seligman provide a similar analysis based on Dretske 1999). The correlation is usually nomical (it follows some law). It may be engineered—as in the case of the low-battery indicator (a) whose flashing (F) is triggered by, and hence is informative about, the battery (b) being dead (G). Or it may be natural, as when litmus—a coloring matter from lichens—is used as an acid-alkali indicator (litmus turns red in acid solutions and blue in alkaline solutions). Other typical examples include the correlation between fingerprints and personal identification, or between the age of a plant and its growth rings.

One may be so used to equating the low-battery indicator flashing with the information (that is, meaning) that the battery is dead as to find it hard to distinguish sufficiently between environmental and semantic information. However it is important to remember that environmental information may require or involve no semantics at all. It may consist of correlated data understood as mere differences or constraining affordances. Plants (e.g., a sunflower), animals (e.g., an amoeba) and mechanisms (e.g., a photocell) are certainly capable of making practical use of environmental information even in the absence of any (semantic processing of) meaningful data. Figure 1 summarizes the main distinctions introduced so far.

FIVE TYPES OF INFORMATION. More detail may now be added. First it should be emphasized that the actual format, medium, and language in which information is encoded is often irrelevant. The same semantic, instructional, and environmental information may be analog or digital, printed on paper or viewed on a screen, or in English or some other language. Second thus far it has been implicitly assumed that primary information is the central issue: things like the low-battery indicator flashing, or the words the battery is dead spoken over the phone. But remember how John discovered that the battery was dead. The engine failed to make any of the usual noises. Likewise in Sir Arthur Conan Doyle's Silver Blaze (1892), Sherlock Holmes solves the case by noting something that has escaped everybody else's attention, the unusual silence of the dog. Clearly silence may be very informative. This is a peculiarity of information: Its absence may also be informative. When it is, the difference may be explained by speaking of secondary information.

Apart from secondary information, three other typologies are worth some explanation since they are quite common (the terminology is still far from being standard or fixed, but see Floridi 1999b). Metainformation is information about the nature of information. "The battery is dead is encoded in English" is a simple example. Operational information is information about the dynamics of information. Suppose the car has a yellow light that, when flashing, indicates the entire system that checks that the electronic components of the car is malfunctioning. The fact that the light is off indicates that the low-battery indicator is working properly, thus confirming that the battery is indeed dead. Finally derivative information is information that can be extracted from any form of information whenever the latter is used as a source in search of patterns, clues, or inferential evidence, namely for comparative and quantitative analyses. From a credit card bill concerning the purchase of gasoline, one may derive information about the cardholder's whereabouts at a given time.

Information as Communication

Also important is the concept of information as communication, as in the sense of a transmitted message (Cherry 1978). Some features of information are intuitively quantitative. Information can be encoded, stored, and transmitted. One also expects it to be additive (information a + information b = information a + b) and non-negative. Similar properties of information are investigated by the mathematical theory of communication (MTC, also known as information theory; for an accessible introduction, see Jones 1979).

MTC was developed by Claude E. Shannon (Shannon and Weaver 1998 [1949]) with the primary aim of devising efficient ways of encoding and transferring data. Its two fundamental problems are the ultimate level of data compression (how small can a message be, given the same amount of information to be encoded?) and the ultimate rate of data transmission (how fast can data be transmitted over a channel?). To understand this approach, consider the telephone call to the garage.

The telephone communication with the mechanic is a specific case of a general communication model. The model is described in Figure 2.

John is the informer, the mechanic is the informee, the battery is dead is the message (the informant), there is a coding and decoding procedure through a language (English), a channel of communication (the telephone system), and some possible noise. Informer and informee share the same background knowledge about the collection of usable symbols (the alphabet).

MTC treats information as only a selection of symbols from a set of possible symbols, so a simple way of grasping how MTC quantifies raw information is by considering the number of yes/no questions required to guess what the informer is communicating. When a fair coin is tossed, one question is sufficient to guess whether the outcome is heads (h) or tails (t). Therefore a binary source, like a coin, is said to produce one bit of information. A two-fair-coins system produces four ordered outputs: <h, h, h, t, t, h, t, t> and therefore requires two questions, each output containing two bits of information, and so on. In the example, the low-battery indicator is also a binary device: If it works properly, it either flashes or it does not, exactly like a tossed coin. And since it is more unlikely that it flashes, when it does, the red light is very informative. More generally the lower the probability of p the more informative the occurrence of p is (unfortunately this leads to the paradoxical view that a contradiction—which has probability 0—is the most informative of all contents, unless one maintains that, to qualify as information, p needs to be true [Floridi 2004]).

Before the coin is tossed, the informee does not know which symbol the device will actually produce, so it is in a state of data deficit equal to 1 (Shannon's uncertainty). Once the coin has been tossed, the system produces an amount of raw information that is a function of the possible outputs, in this case two equiprobable symbols, and equal to the data deficit that it removes. The reasoning applies equally well to the letters used in your telephone conversation with the mechanic.

The analysis can be generalized. Call the number of possible symbols N. For N = 1, the amount of information produced by a unary device is 0. For N = 2, by producing an equiprobable symbol, the device delivers one unit of information. And for N = 4, by producing an equiprobable symbol, the device delivers the sum of the amount of information provided by coin A plus the amount of information provided by coin B, that is two units of information. Given an alphabet of N equiprobable symbols, it is possible to rephrase some examples more precisely by using the following equation: log 2 (N) = bits of information per symbol.

Things are made more complicated by the fact that real coins are always biased, and so are low-battery indicators. Likewise in John's conversation with the mechanic a word like batter will make y as the next letter almost certain. To calculate how much information a biased device produces, one must rely on the frequency of the occurrences of symbols in a finite series of occurrences, or on their probabilities, if the occurrences are supposed to go on indefinitely. Once probabilities are taken into account, the previous equation becomes Shannon's formula (where H = uncertainty, what has been called above data deficit):

The quantitative approach just outlined plays a fundamental role in coding theory, hence in cryptography, and in data storage and transmission techniques, which are based on the same principles and concepts. Two of them are so important as to deserve a brief explanation: redundancy and noise.

Redundancy refers to the difference between the physical representation of a message and the mathematical representation of the same message that uses no more bits than necessary. It is basically what can be taken away from a message without loss in communication. John's statement that his wife was responsible for the dead battery was redundant.

Compression procedures work by reducing data redundancy, but redundancy is not always a bad thing, for it can help to counteract equivocation (data sent but never received) and noise (received but unwanted data, like some interference). A message + noise contains more data than the original message by itself, but the aim of a communication process is fidelity, the accurate transfer of the original message from sender to receiver, not data increase. The informee is more likely to reconstruct a message correctly at the end of the transmission if some degree of redundancy counterbalances the inevitable noise and equivocation introduced by the physical process of communication and the environment. This is why, over the phone, John said that the battery is dead and that the lights were left on last night. It was the by whom that was uselessly redundant.

MTC is not a theory of information in the ordinary sense of the word. The term raw information has been used to stress the fact that in MTC information has an entirely technical meaning. Two equiprobable yeses contain the same quantity of raw information, regardless of whether their corresponding questions are Is the battery dead? or Is your wife missing? Likewise if one knows that a device could send with equal probabilities either this whole encyclopedia or just a quote for its price, by receiving one or the other message one would receive very different quantities of data bytes but only one bit of raw information. Since MTC is a theory of information without meaning, and since information – meaning = data, mathematical theory of data communication is a far more appropriate description than information theory.

MTC deals not with semantic information itself but with messages constituted by uninterpreted symbols encoded in well-formed strings of signals, so it is commonly described as a study of information at the syntactic level. This generates some confusion because one may think the syntactic versus semantic dichotomy to be exhaustive. Clearly MTC can be applied in information and communication technologies (ICT) successfully because computers are syntactical devices. It is often through MTC that information becomes a central concept and topic of research in disciplines like chemistry, biology, physics, cognitive science, neuroscience, the philosophy of information (Floridi 2002, Floridi 2004a), and computer ethics (Floridi 1999a).

LUCIANO FLORIDI

SEE ALSO Computer Ethics;Cybernetics;Digital Libraries;Geographic Information Systems;Information Overload;Information Society;Internet;Wiener, Norbert.

BIBLIOGRAPHY

Bar-Hillel, Yehoshua. (1964). Language and Information: Selected Essays on Their Theory and Application. Reading, MA; London: Addison-Wesley. Important collection of relevant essays by one of the philosophers who first tried to apply information theory to semantic information.

Bar-Hillel, Yehoshua, and Rudolf Carnap. (1953). "An Outline of a Theory of Semantic Information." In Language and Information: Selected Essays on Their Theory and Application Reading, MA; London: Addison-Wesley. Influential attempt to quantify semantic information.

Barwise, Jon, and Jerry Seligman. (1997). Information Flow: The Logic of Distributed Systems. Cambridge, UK: Cambridge University Press. Applies situation logic to the study of the dynamics of information.

Cherry, Colin. (1978). On Human Communication: A Review, a Survey, and a Criticism, 3rd edition. Cambridge, MA: MIT Press.

Dretske, Fred I. (1999). Knowledge and the Flow of Information. Stanford, CA: CSLI Publications. Originally published in 1981, Cambridge, MA: MIT Press. A simple and informative introduction to the mathematical theory of communication, even if no longer up-to-date.

Floridi, Luciano. (1999a). "Information Ethics: On the Theoretical Foundations of Computer Ethics." Ethics and Information Technology 1(1): 37–56. Provides a first foundation for an approach in computer ethics known as information ethics.

Floridi, Luciano. (1999b). Philosophy and Computing: An Introduction. London, New York: Routledge. Textbook introduction for philosophy students to computer science and its conceptual challenges.

Floridi, Luciano. (2002). "What Is the Philosophy of Information?" Metaphilosophy 33(1–2): 123–145. Provides a definition of the philosophy of information as a area of research.

Floridi, Luciano. (2004). "Outline of a Theory of Strongly Semantic Information." Minds and Machines 14(2): 197–222. Defends a theory of semantic information based on a veridical interpretation of information, in order to solve the Bar-Hillel-Carnap paradox.

Floridi, Luciano, and J. W. Sanders. (2004). "The Method of Abstraction" In Yearbook of the Artificial: Nature, Culture and Technology: Models in Contemporary Sciences, ed. Massimo Negrotti. Bern: Peter Lang. Develops the method of abstraction for philosophical analysis.

Floridi, Luciano. (2005). "Is Information Meaningful Data?" Philosophy and Phenomenological Research 70(2): 351–370. Defends the thesis that information encapsulates truth, hence that "false" in "false information" is to be understood as meaning "not authentic."

Fox, Christopher J. (1983). Information and Misinformation: An Investigation of the Notions of Information, Misinformation, Informing, and Misinforming. Westport, CT: Greenwood Press. Overview of the concepts mentioned in the title from an information science perspective.

Hanson, Philip P., ed. (1990). Information, Language, and Cognition. Vancouver: University of British Columbia Press. Proceedings of an influential conference on semantic information.

Hintikka, Jaakko, and Patrick Suppes, eds. (1970). Information and Inference. Dordrecht, The Netherlands: Reidel. Influential collection of essays on the philosohy of semantic information.

Jones, Douglas Samuel. (1979). Elementary Information Theory. Oxford: Clarendon Press. Excellent introduction to the mathematical theory of communication for the mathematically uninitiated.

Shannon, Claude E., and Warren Weaver. (1998 [1949]). The Mathematical Theory of Communication. Urbana: University of Illinois Press. Classic presentation of the fundamental results in the mathematical theory of communication.