The alethic nature of semantic information has been, and continues to be a point of contention. At the very least semantic information is understood as semantic content; that is, meaningful, well-formed data. Defenders of the alethic neutrality of semantic information argue that semantic content already qualifies as information, regardless of whether it is true, false or has no alethic value at all. Opponents hold that not just any semantic content qualifies as information. For semantic content to qualify as information, it must also be true; false information or misinformation is not actually a kind of information.
I am inclined to endorse a truth requirement condition in defining semantic information, what can be termed the ‘veridicality thesis’ (VT). So information is defined as well-formed, meaningful and truthful data. The following two sentences are meaningful, well-formed data in the English language:
- Rome is the capital of Italy
- Venice is the capital of Italy
Of these two items of data, only 1 is information, for it is the only true one.
Whilst it might seem that the debate concerning the alethic nature of information is just a trivial terminological one, there is arguably more to it. To begin with, having precise terms supports clarity and effectiveness of analysis. Furthermore, there are practical reasons for working with a truth-based definition of semantic information. For one, it provides a foundation for attempts to define knowledge in terms of information, as both encapsulate truth. Also, it nicely supports a method I am developing to quantitatively analyse information in terms of the connection between information and truthlikeness.
In the last few years, there has been debate within a series of papers between members of these two opposing camps. This recent string of papers around the debate begins with a 2005 paper by Floridi titled ‘Is Semantic Information Meaningful Data?’. This paper contributes to the debate by criticising the standard definition of declarative, objective and semantic (DOS) information as well-formed, meaningful data and revising it to include a necessary truth condition. Information is strongly semantic information, where truth does not supervene on information, but is a necessary condition.
Two independent objections against Floridi’s account of strongly semantic information are mounted in papers from James Fetzer and Gordana Dodig-Crnkovic. he Fetzer paper, titled ‘Information: Does it Have To Be True?’, offers various logical, epistemic, and ordinary-language points in arguing that an account like Floridi’s is too restricted and obscures crucial differences between information, misinformation, and disinformation. In ‘System Modeling and Information Semantics’, Dodig-Crnkovic argues that meaningful data does not necessarily need to be true to constitute information. I found these two papers insubstantial, and their objections easily addressed.
In ‘The Metaphilosophy of Information’ Sebastian Sequoiah-Grayson mounts a defence of Floridi’s theory of strongly semantic information against the objections from Fetzer and Dodig-Crnkovic.
Finally, Floridi himself further contributes towards strengthening his position in the 2007 paper titled ‘In Defence of the Veridical Nature of Semantic Information’.
On top of the arguments presented in the literature for the veridicality thesis (VT), there are a few others I can think of.
One of Floridi’s arguments evokes recognition of a general problem with the idea of counting semantic data as information regardless of whether it is true or false, for when falsity is not separated from the definition of information, conflict arises. At the very least it can be odd to say that of something that it is both information and misinformation.
More significant is the following point. It seems fair to say that most opponents of the veridicality thesis accept that tautological and contradictory propositions do not qualify as information, though they are still determined to argue that contingently false propositions should count as information. It also seems fair to say that information follows a principle of conjunction: for any two propositions A and B, if A is information and B is information, then the compound proposition A & B is information. However, an adherence to these two things leads to the problematic result that when A is information and ~A is information, which is possible with an anti-VT framework, A & ~A is information. For those anti-VT adherents who accept that contradictory propositions do not qualify as information, this puts them in an inextricable bind. For those anti-VT adherents who accept that contradictory propositions can qualify as information, the onus is on them to convince us why.
When talking about semantic information here, I have implicitly been talking about factual semantic information. That is, information about some fact or state of affairs, that can be given propositional expression, in the form of ‘the information that x’. The other type of semantic information is instructional information.
We can also bring consideration of instructional information into the debate. Take the following set of instructions involving direction on how to get from point A to point B. To get from point A to point B:
- travel 50 metres straight
- turn left
- travel 100 metres straight
- turn right
- travel travel 20 metres straight
This instructional information can be reduced to something like the following factual information:
Travelling 50 metres straight, then turning left, then travelling 100 metres straight, then turning right and then travelling 20 metres straight, will get you from point A to point B.
Now, consider the following incorrect list of directions. To get from point A to point B:
- travel 50 metres straight
- turn right
- travel 100 metres straight
- turn leftt
- travel travel 20 metres straight
It would be innapropriate to call this instructional information on how to get from point A to point B, for it does not get one from point A to point B. Its reduction to propositional form gives the false proposition:
Travelling 50 metres straight, then turning right, then travelling 100 metres straight, then turning left and then travelling 20 metres straight, will get you from point A to point B.
If the set of instructions this proposition derives from is not instructional information, then why should this false proposition be considered information?