Conceptual engineering
from 2000 Haslanger's ameliorative project; the name became current after 2015
Stop asking what a concept means and ask what it is for — then rebuild it so it does that job better, and defend the rebuild openly.
- Active
- from 2000
- Where
- Cambridge, Massachusetts
- Status
- In use
- Tradition
- Analytic tools
What it is
Conceptual engineering treats concepts as instruments rather than as objects of discovery. The question is not what our concept of X is, but what concept of X we ought to have given what we want it to do.
It differs from Carnap's explication in ambition and in candour. Explication replaces a vague notion with a precise one for theoretical purposes; conceptual engineering will revise a perfectly precise concept because the current one serves badly, and it treats the choice as a normative question requiring argument.
How it works
Identify the concept and the purpose it is supposed to serve. Ask whether the concept as used serves that purpose. Propose a revision. Argue for it on the grounds of the purpose, not on the grounds of fidelity to current usage. Consider whether the revised concept can actually be adopted, and what happens during the period when both are in circulation.
The last step is the one usually skipped, and it is where most proposals fail.
The moves, in order
- Identify the concept and the purpose it serves
- Show that the concept as used serves that purpose badly
- Propose a revision
- Defend it by reference to the purpose rather than to current usage
- Address whether it can be adopted, and what happens meanwhile
The procedure
Proposition The procurement policy needs a different concept of local business.
- The policy gives a weighting to local businesses in order to keep procurement money circulating in the shire.
- As defined, a local business is one with a registered address in the shire. On that definition the weighting has gone three times in two years to firms whose registered address is an accountant's office and whose staff, plant and banking are all elsewhere.
- So the concept as defined does not serve the purpose it exists for.
- The revision proposed is not what local really means — that argument is unwinnable — but a concept built for the job: a business is local if a majority of the hours worked on the contract will be worked by people residing in the shire.
- This is defended on the purpose: it tracks where the money goes, which is what the weighting was for. It is not defended as capturing what anybody meant by local.
Therefore Adopt the hours-worked concept for the weighting, and state in the policy that it is a defined term for that purpose and not a general definition of local.
The final sentence is what makes this conceptual engineering rather than a definitional argument. The revision is announced as a revision, which is what stops the next dispute being about whether it captured the true meaning.
What it is good for
Policy definitions, legal categories and any concept doing institutional work that it was not designed for. Also the philosophical questions where analysis has visibly stalled.
Where it fails
Two hard problems, neither solved. The first is control: if a concept is determined by how a whole community uses a word, no philosopher and no committee can change it by proposing to, and the proposal simply produces two concepts in circulation. Cappelen presses this hard.
The second is the topic problem. Change a concept enough and you are no longer talking about the original subject, and the revision cannot answer the question that motivated it. Somewhere between a minor refinement and a change of subject is a line nobody has drawn.
And since the revision is justified by a purpose, whoever sets the purpose controls the outcome, which makes the method as good or as bad as the politics behind it.
Common misconceptions
That it is stipulation with a grand name. The difference is that a stipulation needs no defence and an engineered concept requires an argument that it serves the purpose better than the incumbent.
Where it came from
Sally Haslanger's ameliorative project around 2000, asking what concepts of gender and race we ought to have given the purposes of social justice, rather than what they currently are.
What became of it
One of the most active areas in analytic philosophy since about 2015, with the implementation problem as its central unsolved difficulty.
Lasting influence
It has reframed a range of debates as questions about what concepts to adopt rather than what concepts we have, and it has given a name to what legal drafters have always done.
Why it still matters
The most current method on this list.
Put another way
Replacing a spanner that rounds the nut, rather than arguing about what the spanner really is.
Texts
Herman Cappelen, Fixing Language
The most sustained treatment, and the sharpest statement of why the method may not be able to do what it proposes.
Where to start
Haslanger's Gender and Race: (What) Are They? (What) Do We Want Them To Be? — the title states the method.
The surprising bit
The method presented as a break from analysis turns out to describe standard legal drafting practice, which has been engineering concepts for purposes for centuries without a philosophical account of it.
In their own words
What is the point of having these concepts? What cognitive or practical task do they enable us to accomplish?
Who held it
Herman Cappelen
Connections
- Relies on
- Explication
- Set against
- Conceptual analysis, Ordinary language
- Parallel to
- Explication
The verdict
The right question to ask when analysis stalls, and it has no account of how a proposal ever takes hold.