There is no agreed grain count at which an ordinary heap stops being a heap. The sorites paradox asks why that absence is so difficult to reconcile with reasoning that seems harmless one step at a time.
The argument has three steps:
- A large pile of sand is a heap.
- Whenever a collection is a heap, removing one grain leaves a heap.
- Repeated application of that rule makes a single grain a heap.
The third result conflicts with the clear non-heap at the end. The problem is how an apparently harmless judgement about one removal became the universal rule in step two.
The puzzle, traditionally attributed to Eubulides, concerns the rule as much as the sand. Cian Dorr’s questions about vagueness starts with the ancient questioning strategy and asks what a borderline answer actually means. “I’m unsure” may be a sensible response, but explaining that uncertainty is part of the philosophical job.
The little word doing the damage
The troublesome claim is every step. It is stronger than saying that this particular removal makes no noticeable difference.
Imagine writing an answer beside every pile in a long row. At the large end you write “heap”; at the one-grain end you write “not a heap.” If every neighbouring pair must receive the same answer, those end labels cannot both remain. You cannot get around that by checking only the first few pairs, however convincing they look.
As in the Ship of Theseus, noticing that small changes accumulate identifies a problem without selecting its solution. The paradox asks which commitment to change: the universal rule, the account of truth, the treatment of context, or something else. The Stanford Encyclopedia’s sorites entry surveys the disagreement. It does not announce a settled winner.
The count also simplifies the physical example. For the thought experiment, hold grain size and the way the pile is arranged suitably steady. Scattering the same grains across a floor changes the question. We are testing a vague classification along a controlled series, not discovering a law that heaps depend only on arithmetic.
Four ways to answer
Epistemicism keeps a sharp boundary and says its location is unknowable. Somewhere a one-grain difference changes the truth of “this is a heap,” even though competent speakers cannot identify the transition. This protects ordinary true-or-false judgements. Its difficult question is how linguistic practice determines a boundary nobody can locate.
Supervaluationism considers admissible ways to make the vague expression precise. Something is super-true if it is true under all those sharpenings. A borderline case need not qualify under all of them. The approach can preserve “heap or not heap” without either part being determinately true on its own. It then has to explain what makes a sharpening admissible, including the vagueness that can arise around admissibility itself.
Degree approaches, including fuzzy accounts, allow more than a simple true/false classification. They must explain both the assigned values and how logical inference works with them. Giving a pile “0.6 heapness” is not automatically a measurement, and it is not the same as a 60 percent chance of an unknown yes/no fact.
Contextual approaches examine changes in the standards operative during a conversation or sequence of comparisons. A neighbouring pair can seem alike while the longer process shifts the classification standard. Context helps explain the attraction of the stepwise argument; saying “context matters” alone does not settle every form of the paradox.
The Stanford entry on vagueness discusses competing accounts of borderline truth, including degree theories and supervaluationism. These are families of theories with technical differences, rather than four interchangeable ways of saying “it depends.”
Try a sharpening without mistaking it for a discovery
Suppose a classroom game defines a heap as at least 1,000 grains. The game now has a straightforward answer at 999. You have created a usable rule. You have not shown that everyone previously speaking English was unknowingly following it.
For a simplified supervaluationist exercise, try two candidate sharpenings, with cutoffs at 900 and 1,100. A pile of 1,500 passes both; 100 passes neither; 1,000 passes one. This illustrates agreement at the ends and disagreement in the middle. The invented numbers are teaching aids, not established admissible boundaries for the real word.
Now ask a harder question: why accept those two sharpenings but exclude one that calls a single grain a heap? The exercise exposes the work still required. A list of arbitrary definitions cannot by itself provide a theory of ordinary meaning.
“Taller than” and “tall” do different jobs
You can compare two measured heights without deciding whether either person counts as tall. Christopher Kennedy’s chapter on the sorites in linguistics examines this distinction between comparative constructions and the unmodified positive form of a gradable adjective.
For an everyday example, consider a shelf. You might ask whether it is taller than another shelf, tall enough to hold a particular book, or simply tall. The comparison supplies one reference; the book supplies a practical requirement; the bare adjective leaves more to interpretation. All three questions can be useful without having identical answers or identical sources of uncertainty.
Context sensitivity and vagueness should still be distinguished. Fixing a comparison group can help establish what we are discussing while leaving borderline cases inside that discussion. Moving every troublesome case to a new context would describe the uncertainty away rather than explain it.
Useful language can still have troublesome edges
If a friend asks you to clear the heap of papers off a chair, the sensible next step is usually to move the papers. Requiring a numerical threshold would add work without helping the task.
That example shows that a vague expression can participate in successful communication. It does not prove that vagueness itself causes the success. David Lanius’s What Is the Value of Vagueness? challenges precisely that inference, separating vagueness from generality, ambiguity and other useful features of language.
When a boundary matters, ask what kind of answer you need. A practical convention can tell participants what to do. A philosophical explanation must also account for why the ordinary term generates the puzzle. Both questions deserve answers; they should not be silently exchanged.
The heap is easy to recognise in a clear case. The paradox begins when that confidence is required to survive every tiny step to a clear case at the other end.
