Truth and Language 4 - Consensus Theory of Truth
Wikipedia article:
Consensus Theory of Truth
"The consensus theory, invented by Charles Sanders Peirce sees truth as something agreed upon by some specified group, such as all competent investigators. " (from Wikipedia, "Truth")
Like pragmatism, upon first reading this definition, my first reaction was one of immediate rejection, but upon second and third thoughts I decided that there are good ideas here that deserve thought.
First, this theory of truth seems at least well motivated. Many people see scientists as slowly building up an reliable edifice of models and beliefs that is irrefutable. How could we justify equating truth and their work, science? The easiest way is to define one as the other, gratifying our intuitions and giving us, conveniently, a clear process by which to approach truth.
Unfortunately, it has a few obvious faults, both mentioned in the linked article: a) It allows for a proposition to be true even when the meaning of the proposition is not the case--i.e. "Snow is white" could be true even if snow is not white, and b) it is not necessarily well-defined for some, most, or all propositions.
But before looking at these unsettling consequences of the theory, I'd like to take a closer look at the theory itself.
"...truth as something agreed upon by some specified group, such as all competent investigators."
As the article on the theory points out, there is a surface similarity with relativism here--that something can be true purely by virtue of the fact that everybody in the relevant group agrees with it. I have deep objections to relativism that I'll spill some time, but not now. Instead, I want to focus on the notion of competent investigation.
We might imagine the consensus theory of truth providing for us a function, like in math or programming, that takes as input a method of investigation and returns a theory of the world that is the consensus of that investigation.
When I say "method of investigation," I am speaking deliberately broadly. (Part of me immediately revolts and claims that 'method of investigation' is defective as a concept, but I'm lifting myself from that for now) The scientific method might be considered one. Study of texts to discover the secrets of the universe, as ancient Kabbalists, modern workers on the Bible Code, and to a lesser extent evangelical Christians attempt to do, would be another. "Listening to you heart" might be another.
The consensus of a method of investigation M would be, approximately, "Those models or propositions that would be arrived at by anybody and everybody practicing M for (presumably) an arbitrarily long amount of time." If there were no beliefs that everybody practicing M would arrive at, then the consensus is the null set of beliefs.
Let us call the set of beliefs of an individual that have been formed by practicing method M for an arbitrarily long amount of time an individual's M-produced beliefs.
The consensus is then the intersection of the M-produced belief sets of all individuals. Taking this intersection is presumably used to remove individual differences in the investigators out of the equation. This seems like a crude method and has some other problems (who do you include in your group? What if one madman forms radically different beliefs?)
I'm getting lazy now--I don't know how much longer I can write on this topic.
But I want to point out that if it can be shown that M is reliably produces truer and truer beliefs over time, then its consensus might well be the same as the set of all true beliefs. However, this depends on the parameter M being calibrated to the Truth. To define the consensus of an arbitrary method M as Truth would make the choice of M circular. There would be no basis on which to choose scientific investigation over reading tea leaves.
What the concepts of consensus, M-produced beliefs, etc. provide, however, are a hope that we might approach the truth by a systematic process. If we choose M correctly, perhaps the consensus of M is a good thing to strive for.
NOTE: The issue of time is one I have dealt with lightly here, but I think it will become very important later on.
Complexity theory shows us that there is another measure of an algorithm other than how right the answer we get is; we must also value how quickly it gives us answers. Above, we allowed an investigator infinite time to carry out method M. In practice, a theory of truth that provides us only with a promise of truth in the indefinite future is not very satisfying. Can we find a method that is guaranteed to at least approach truth? Can we (we must!) overcome our intuitions about a binary true/false distinction that render approximation useless.
In addition, I suspect that we will have to judge our truths by their complexity as well. I might have all of truth immediately at hand in a database. But the longer it takes me to query that database of truth, the less valuable that truth becomes. Maybe there is a bit of human pragmatism bleeding in here--"What good is truth if it can't help me?" So we need rougher truths that we can query faster (with inaccuracies, perhaps) when we need to, but we have to acknowledge them as being possibly inaccurate...
I'm rambling now, and letting a lot of computer science spill into this epistemology. So sue me.
p.s. I don't know how to spell competent, indefinite, or relevant.
Consensus Theory of Truth
"The consensus theory, invented by Charles Sanders Peirce sees truth as something agreed upon by some specified group, such as all competent investigators. " (from Wikipedia, "Truth")
Like pragmatism, upon first reading this definition, my first reaction was one of immediate rejection, but upon second and third thoughts I decided that there are good ideas here that deserve thought.
First, this theory of truth seems at least well motivated. Many people see scientists as slowly building up an reliable edifice of models and beliefs that is irrefutable. How could we justify equating truth and their work, science? The easiest way is to define one as the other, gratifying our intuitions and giving us, conveniently, a clear process by which to approach truth.
Unfortunately, it has a few obvious faults, both mentioned in the linked article: a) It allows for a proposition to be true even when the meaning of the proposition is not the case--i.e. "Snow is white" could be true even if snow is not white, and b) it is not necessarily well-defined for some, most, or all propositions.
But before looking at these unsettling consequences of the theory, I'd like to take a closer look at the theory itself.
"...truth as something agreed upon by some specified group, such as all competent investigators."
As the article on the theory points out, there is a surface similarity with relativism here--that something can be true purely by virtue of the fact that everybody in the relevant group agrees with it. I have deep objections to relativism that I'll spill some time, but not now. Instead, I want to focus on the notion of competent investigation.
We might imagine the consensus theory of truth providing for us a function, like in math or programming, that takes as input a method of investigation and returns a theory of the world that is the consensus of that investigation.
When I say "method of investigation," I am speaking deliberately broadly. (Part of me immediately revolts and claims that 'method of investigation' is defective as a concept, but I'm lifting myself from that for now) The scientific method might be considered one. Study of texts to discover the secrets of the universe, as ancient Kabbalists, modern workers on the Bible Code, and to a lesser extent evangelical Christians attempt to do, would be another. "Listening to you heart" might be another.
The consensus of a method of investigation M would be, approximately, "Those models or propositions that would be arrived at by anybody and everybody practicing M for (presumably) an arbitrarily long amount of time." If there were no beliefs that everybody practicing M would arrive at, then the consensus is the null set of beliefs.
Let us call the set of beliefs of an individual that have been formed by practicing method M for an arbitrarily long amount of time an individual's M-produced beliefs.
The consensus is then the intersection of the M-produced belief sets of all individuals. Taking this intersection is presumably used to remove individual differences in the investigators out of the equation. This seems like a crude method and has some other problems (who do you include in your group? What if one madman forms radically different beliefs?)
I'm getting lazy now--I don't know how much longer I can write on this topic.
But I want to point out that if it can be shown that M is reliably produces truer and truer beliefs over time, then its consensus might well be the same as the set of all true beliefs. However, this depends on the parameter M being calibrated to the Truth. To define the consensus of an arbitrary method M as Truth would make the choice of M circular. There would be no basis on which to choose scientific investigation over reading tea leaves.
What the concepts of consensus, M-produced beliefs, etc. provide, however, are a hope that we might approach the truth by a systematic process. If we choose M correctly, perhaps the consensus of M is a good thing to strive for.
NOTE: The issue of time is one I have dealt with lightly here, but I think it will become very important later on.
Complexity theory shows us that there is another measure of an algorithm other than how right the answer we get is; we must also value how quickly it gives us answers. Above, we allowed an investigator infinite time to carry out method M. In practice, a theory of truth that provides us only with a promise of truth in the indefinite future is not very satisfying. Can we find a method that is guaranteed to at least approach truth? Can we (we must!) overcome our intuitions about a binary true/false distinction that render approximation useless.
In addition, I suspect that we will have to judge our truths by their complexity as well. I might have all of truth immediately at hand in a database. But the longer it takes me to query that database of truth, the less valuable that truth becomes. Maybe there is a bit of human pragmatism bleeding in here--"What good is truth if it can't help me?" So we need rougher truths that we can query faster (with inaccuracies, perhaps) when we need to, but we have to acknowledge them as being possibly inaccurate...
I'm rambling now, and letting a lot of computer science spill into this epistemology. So sue me.
p.s. I don't know how to spell competent, indefinite, or relevant.